idnits 2.17.1 draft-ietf-p2psip-base-15.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- -- The document has examples using IPv4 documentation addresses according to RFC6890, but does not use any IPv6 documentation addresses. Maybe there should be IPv6 examples, too? Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 1847 has weird spacing: '...Options optio...' == Line 2101 has weird spacing: '...ionType type;...' == Line 2318 has weird spacing: '...tyValue ide...' == Line 2627 has weird spacing: '...ionType typ...' == Line 2629 has weird spacing: '...ionData val...' == (3 more instances...) -- The document seems to contain a disclaimer for pre-RFC5378 work, and may have content which was first submitted before 10 November 2008. The disclaimer is necessary when there are original authors that you have been unable to contact, or if some do not wish to grant the BCP78 rights to the IETF Trust. If you are able to get all authors (current and original) to grant those rights, you can and should remove the disclaimer; otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (May 27, 2011) is 4711 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'NodeIdLength' is mentioned on line 1742, but not defined -- Looks like a reference, but probably isn't: '0' on line 4633 == Missing Reference: 'RFC-AAAA' is mentioned on line 6346, but not defined ** Obsolete normative reference: RFC 2818 (Obsoleted by RFC 9110) ** Obsolete normative reference: RFC 2988 (Obsoleted by RFC 6298) ** Obsolete normative reference: RFC 3023 (Obsoleted by RFC 7303) ** Downref: Normative reference to an Informational RFC: RFC 3174 ** Obsolete normative reference: RFC 3447 (Obsoleted by RFC 8017) ** Obsolete normative reference: RFC 4347 (Obsoleted by RFC 6347) ** Obsolete normative reference: RFC 4395 (Obsoleted by RFC 7595) ** Obsolete normative reference: RFC 5245 (Obsoleted by RFC 8445, RFC 8839) ** Obsolete normative reference: RFC 5246 (Obsoleted by RFC 8446) ** Obsolete normative reference: RFC 5389 (Obsoleted by RFC 8489) ** Obsolete normative reference: RFC 5766 (Obsoleted by RFC 8656) ** Downref: Normative reference to an Informational RFC: RFC 6091 ** Downref: Normative reference to an Informational RFC: RFC 6234 == Outdated reference: A later version (-10) exists of draft-ietf-hip-reload-instance-03 == Outdated reference: A later version (-16) exists of draft-ietf-mmusic-ice-tcp-13 == Outdated reference: A later version (-09) exists of draft-ietf-p2psip-concepts-03 == Outdated reference: A later version (-21) exists of draft-ietf-p2psip-sip-05 -- Obsolete informational reference (is this intentional?): RFC 5201 (Obsoleted by RFC 7401) -- Obsolete informational reference (is this intentional?): RFC 5785 (Obsoleted by RFC 8615) Summary: 13 errors (**), 0 flaws (~~), 13 warnings (==), 7 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 P2PSIP C. Jennings 3 Internet-Draft Cisco 4 Intended status: Standards Track B. Lowekamp, Ed. 5 Expires: November 28, 2011 Skype 6 E. Rescorla 7 RTFM, Inc. 8 S. Baset 9 H. Schulzrinne 10 Columbia University 11 May 27, 2011 13 REsource LOcation And Discovery (RELOAD) Base Protocol 14 draft-ietf-p2psip-base-15 16 Abstract 18 This specification defines REsource LOcation And Discovery (RELOAD), 19 a peer-to-peer (P2P) signaling protocol for use on the Internet. A 20 P2P signaling protocol provides its clients with an abstract storage 21 and messaging service between a set of cooperating peers that form 22 the overlay network. RELOAD is designed to support a P2P Session 23 Initiation Protocol (P2PSIP) network, but can be utilized by other 24 applications with similar requirements by defining new usages that 25 specify the kinds of data that must be stored for a particular 26 application. RELOAD defines a security model based on a certificate 27 enrollment service that provides unique identities. NAT traversal is 28 a fundamental service of the protocol. RELOAD also allows access 29 from "client" nodes that do not need to route traffic or store data 30 for others. 32 Status of this Memo 34 This Internet-Draft is submitted in full conformance with the 35 provisions of BCP 78 and BCP 79. 37 Internet-Drafts are working documents of the Internet Engineering 38 Task Force (IETF). Note that other groups may also distribute 39 working documents as Internet-Drafts. The list of current Internet- 40 Drafts is at http://datatracker.ietf.org/drafts/current/. 42 Internet-Drafts are draft documents valid for a maximum of six months 43 and may be updated, replaced, or obsoleted by other documents at any 44 time. It is inappropriate to use Internet-Drafts as reference 45 material or to cite them other than as "work in progress." 47 This Internet-Draft will expire on November 28, 2011. 49 Copyright Notice 51 Copyright (c) 2011 IETF Trust and the persons identified as the 52 document authors. All rights reserved. 54 This document is subject to BCP 78 and the IETF Trust's Legal 55 Provisions Relating to IETF Documents 56 (http://trustee.ietf.org/license-info) in effect on the date of 57 publication of this document. Please review these documents 58 carefully, as they describe your rights and restrictions with respect 59 to this document. Code Components extracted from this document must 60 include Simplified BSD License text as described in Section 4.e of 61 the Trust Legal Provisions and are provided without warranty as 62 described in the Simplified BSD License. 64 This document may contain material from IETF Documents or IETF 65 Contributions published or made publicly available before November 66 10, 2008. The person(s) controlling the copyright in some of this 67 material may not have granted the IETF Trust the right to allow 68 modifications of such material outside the IETF Standards Process. 69 Without obtaining an adequate license from the person(s) controlling 70 the copyright in such materials, this document may not be modified 71 outside the IETF Standards Process, and derivative works of it may 72 not be created outside the IETF Standards Process, except to format 73 it for publication as an RFC or to translate it into languages other 74 than English. 76 Table of Contents 78 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 8 79 1.1. Basic Setting . . . . . . . . . . . . . . . . . . . . . 9 80 1.2. Architecture . . . . . . . . . . . . . . . . . . . . . . 10 81 1.2.1. Usage Layer . . . . . . . . . . . . . . . . . . . . 13 82 1.2.2. Message Transport . . . . . . . . . . . . . . . . . 14 83 1.2.3. Storage . . . . . . . . . . . . . . . . . . . . . . 14 84 1.2.4. Topology Plugin . . . . . . . . . . . . . . . . . . 15 85 1.2.5. Forwarding and Link Management Layer . . . . . . . . 15 86 1.3. Security . . . . . . . . . . . . . . . . . . . . . . . . 16 87 1.4. Structure of This Document . . . . . . . . . . . . . . . 17 88 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 17 89 3. Overlay Management Overview . . . . . . . . . . . . . . . . . 20 90 3.1. Security and Identification . . . . . . . . . . . . . . 20 91 3.1.1. Shared-Key Security . . . . . . . . . . . . . . . . 21 92 3.2. Clients . . . . . . . . . . . . . . . . . . . . . . . . 21 93 3.2.1. Client Routing . . . . . . . . . . . . . . . . . . . 22 94 3.2.2. Minimum Functionality Requirements for Clients . . . 23 95 3.3. Routing . . . . . . . . . . . . . . . . . . . . . . . . 23 96 3.4. Connectivity Management . . . . . . . . . . . . . . . . 25 97 3.5. Overlay Algorithm Support . . . . . . . . . . . . . . . 26 98 3.5.1. Support for Pluggable Overlay Algorithms . . . . . . 26 99 3.5.2. Joining, Leaving, and Maintenance Overview . . . . . 27 100 3.6. First-Time Setup . . . . . . . . . . . . . . . . . . . . 28 101 3.6.1. Initial Configuration . . . . . . . . . . . . . . . 28 102 3.6.2. Enrollment . . . . . . . . . . . . . . . . . . . . . 29 103 4. Application Support Overview . . . . . . . . . . . . . . . . 29 104 4.1. Data Storage . . . . . . . . . . . . . . . . . . . . . . 29 105 4.1.1. Storage Permissions . . . . . . . . . . . . . . . . 31 106 4.1.2. Replication . . . . . . . . . . . . . . . . . . . . 31 107 4.2. Usages . . . . . . . . . . . . . . . . . . . . . . . . . 32 108 4.3. Service Discovery . . . . . . . . . . . . . . . . . . . 33 109 4.4. Application Connectivity . . . . . . . . . . . . . . . . 33 110 5. Overlay Management Protocol . . . . . . . . . . . . . . . . . 33 111 5.1. Message Receipt and Forwarding . . . . . . . . . . . . . 33 112 5.1.1. Responsible ID . . . . . . . . . . . . . . . . . . . 34 113 5.1.2. Other ID . . . . . . . . . . . . . . . . . . . . . . 34 114 5.1.3. Private ID . . . . . . . . . . . . . . . . . . . . . 36 115 5.2. Symmetric Recursive Routing . . . . . . . . . . . . . . 36 116 5.2.1. Request Origination . . . . . . . . . . . . . . . . 37 117 5.2.2. Response Origination . . . . . . . . . . . . . . . . 37 118 5.3. Message Structure . . . . . . . . . . . . . . . . . . . 38 119 5.3.1. Presentation Language . . . . . . . . . . . . . . . 39 120 5.3.1.1. Common Definitions . . . . . . . . . . . . . . . 39 121 5.3.2. Forwarding Header . . . . . . . . . . . . . . . . . 42 122 5.3.2.1. Processing Configuration Sequence Numbers . . . . 44 123 5.3.2.2. Destination and Via Lists . . . . . . . . . . . . 45 124 5.3.2.3. Forwarding Options . . . . . . . . . . . . . . . 47 125 5.3.3. Message Contents Format . . . . . . . . . . . . . . 48 126 5.3.3.1. Response Codes and Response Errors . . . . . . . 49 127 5.3.4. Security Block . . . . . . . . . . . . . . . . . . . 51 128 5.4. Overlay Topology . . . . . . . . . . . . . . . . . . . . 55 129 5.4.1. Topology Plugin Requirements . . . . . . . . . . . . 55 130 5.4.2. Methods and types for use by topology plugins . . . 55 131 5.4.2.1. Join . . . . . . . . . . . . . . . . . . . . . . 55 132 5.4.2.2. Leave . . . . . . . . . . . . . . . . . . . . . . 56 133 5.4.2.3. Update . . . . . . . . . . . . . . . . . . . . . 57 134 5.4.2.4. RouteQuery . . . . . . . . . . . . . . . . . . . 57 135 5.4.2.5. Probe . . . . . . . . . . . . . . . . . . . . . . 58 136 5.5. Forwarding and Link Management Layer . . . . . . . . . . 60 137 5.5.1. Attach . . . . . . . . . . . . . . . . . . . . . . . 61 138 5.5.1.1. Request Definition . . . . . . . . . . . . . . . 61 139 5.5.1.2. Response Definition . . . . . . . . . . . . . . . 64 140 5.5.1.3. Using ICE With RELOAD . . . . . . . . . . . . . . 65 141 5.5.1.4. Collecting STUN Servers . . . . . . . . . . . . . 65 142 5.5.1.5. Gathering Candidates . . . . . . . . . . . . . . 66 143 5.5.1.6. Prioritizing Candidates . . . . . . . . . . . . . 66 144 5.5.1.7. Encoding the Attach Message . . . . . . . . . . . 67 145 5.5.1.8. Verifying ICE Support . . . . . . . . . . . . . . 67 146 5.5.1.9. Role Determination . . . . . . . . . . . . . . . 67 147 5.5.1.10. Full ICE . . . . . . . . . . . . . . . . . . . . 68 148 5.5.1.11. No-ICE . . . . . . . . . . . . . . . . . . . . . 68 149 5.5.1.12. Subsequent Offers and Answers . . . . . . . . . . 68 150 5.5.1.13. Sending Media . . . . . . . . . . . . . . . . . . 69 151 5.5.1.14. Receiving Media . . . . . . . . . . . . . . . . . 69 152 5.5.2. AppAttach . . . . . . . . . . . . . . . . . . . . . 69 153 5.5.2.1. Request Definition . . . . . . . . . . . . . . . 69 154 5.5.2.2. Response Definition . . . . . . . . . . . . . . . 70 155 5.5.3. Ping . . . . . . . . . . . . . . . . . . . . . . . . 70 156 5.5.3.1. Request Definition . . . . . . . . . . . . . . . 71 157 5.5.3.2. Response Definition . . . . . . . . . . . . . . . 71 158 5.5.4. ConfigUpdate . . . . . . . . . . . . . . . . . . . . 71 159 5.5.4.1. Request Definition . . . . . . . . . . . . . . . 72 160 5.5.4.2. Response Definition . . . . . . . . . . . . . . . 72 161 5.6. Overlay Link Layer . . . . . . . . . . . . . . . . . . . 73 162 5.6.1. Future Overlay Link Protocols . . . . . . . . . . . 74 163 5.6.1.1. HIP . . . . . . . . . . . . . . . . . . . . . . . 75 164 5.6.1.2. ICE-TCP . . . . . . . . . . . . . . . . . . . . . 75 165 5.6.1.3. Message-oriented Transports . . . . . . . . . . . 75 166 5.6.1.4. Tunneled Transports . . . . . . . . . . . . . . . 75 167 5.6.2. Framing Header . . . . . . . . . . . . . . . . . . . 76 168 5.6.3. Simple Reliability . . . . . . . . . . . . . . . . . 77 169 5.6.3.1. Retransmission and Flow Control . . . . . . . . . 78 170 5.6.4. DTLS/UDP with SR . . . . . . . . . . . . . . . . . . 79 171 5.6.5. TLS/TCP with FH, No-ICE . . . . . . . . . . . . . . 79 172 5.6.6. DTLS/UDP with SR, No-ICE . . . . . . . . . . . . . . 80 173 5.7. Fragmentation and Reassembly . . . . . . . . . . . . . . 80 174 6. Data Storage Protocol . . . . . . . . . . . . . . . . . . . . 81 175 6.1. Data Signature Computation . . . . . . . . . . . . . . . 82 176 6.2. Data Models . . . . . . . . . . . . . . . . . . . . . . 83 177 6.2.1. Single Value . . . . . . . . . . . . . . . . . . . . 84 178 6.2.2. Array . . . . . . . . . . . . . . . . . . . . . . . 84 179 6.2.3. Dictionary . . . . . . . . . . . . . . . . . . . . . 85 180 6.3. Access Control Policies . . . . . . . . . . . . . . . . 85 181 6.3.1. USER-MATCH . . . . . . . . . . . . . . . . . . . . . 86 182 6.3.2. NODE-MATCH . . . . . . . . . . . . . . . . . . . . . 86 183 6.3.3. USER-NODE-MATCH . . . . . . . . . . . . . . . . . . 86 184 6.3.4. NODE-MULTIPLE . . . . . . . . . . . . . . . . . . . 86 185 6.4. Data Storage Methods . . . . . . . . . . . . . . . . . . 87 186 6.4.1. Store . . . . . . . . . . . . . . . . . . . . . . . 87 187 6.4.1.1. Request Definition . . . . . . . . . . . . . . . 87 188 6.4.1.2. Response Definition . . . . . . . . . . . . . . . 91 189 6.4.1.3. Removing Values . . . . . . . . . . . . . . . . . 93 190 6.4.2. Fetch . . . . . . . . . . . . . . . . . . . . . . . 93 191 6.4.2.1. Request Definition . . . . . . . . . . . . . . . 94 192 6.4.2.2. Response Definition . . . . . . . . . . . . . . . 96 193 6.4.3. Stat . . . . . . . . . . . . . . . . . . . . . . . . 97 194 6.4.3.1. Request Definition . . . . . . . . . . . . . . . 97 195 6.4.3.2. Response Definition . . . . . . . . . . . . . . . 97 196 6.4.4. Find . . . . . . . . . . . . . . . . . . . . . . . . 99 197 6.4.4.1. Request Definition . . . . . . . . . . . . . . . 99 198 6.4.4.2. Response Definition . . . . . . . . . . . . . . . 100 199 6.4.5. Defining New Kinds . . . . . . . . . . . . . . . . . 101 200 7. Certificate Store Usage . . . . . . . . . . . . . . . . . . . 101 201 8. TURN Server Usage . . . . . . . . . . . . . . . . . . . . . . 102 202 9. Chord Algorithm . . . . . . . . . . . . . . . . . . . . . . . 104 203 9.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 105 204 9.2. Hash Function . . . . . . . . . . . . . . . . . . . . . 105 205 9.3. Routing . . . . . . . . . . . . . . . . . . . . . . . . 105 206 9.4. Redundancy . . . . . . . . . . . . . . . . . . . . . . . 106 207 9.5. Joining . . . . . . . . . . . . . . . . . . . . . . . . 106 208 9.6. Routing Attaches . . . . . . . . . . . . . . . . . . . . 107 209 9.7. Updates . . . . . . . . . . . . . . . . . . . . . . . . 107 210 9.7.1. Handling Neighbor Failures . . . . . . . . . . . . . 109 211 9.7.2. Handling Finger Table Entry Failure . . . . . . . . 110 212 9.7.3. Receiving Updates . . . . . . . . . . . . . . . . . 110 213 9.7.4. Stabilization . . . . . . . . . . . . . . . . . . . 111 214 9.7.4.1. Updating neighbor table . . . . . . . . . . . . . 111 215 9.7.4.2. Refreshing finger table . . . . . . . . . . . . . 111 216 9.7.4.3. Adjusting finger table size . . . . . . . . . . . 112 217 9.7.4.4. Detecting partitioning . . . . . . . . . . . . . 113 218 9.8. Route query . . . . . . . . . . . . . . . . . . . . . . 113 219 9.9. Leaving . . . . . . . . . . . . . . . . . . . . . . . . 114 221 10. Enrollment and Bootstrap . . . . . . . . . . . . . . . . . . 115 222 10.1. Overlay Configuration . . . . . . . . . . . . . . . . . 115 223 10.1.1. Relax NG Grammar . . . . . . . . . . . . . . . . . . 121 224 10.2. Discovery Through Configuration Server . . . . . . . . . 123 225 10.3. Credentials . . . . . . . . . . . . . . . . . . . . . . 124 226 10.3.1. Self-Generated Credentials . . . . . . . . . . . . . 125 227 10.4. Searching for a Bootstrap Node . . . . . . . . . . . . . 126 228 10.5. Contacting a Bootstrap Node . . . . . . . . . . . . . . 126 229 11. Message Flow Example . . . . . . . . . . . . . . . . . . . . 127 230 12. Security Considerations . . . . . . . . . . . . . . . . . . . 133 231 12.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 133 232 12.2. Attacks on P2P Overlays . . . . . . . . . . . . . . . . 134 233 12.3. Certificate-based Security . . . . . . . . . . . . . . . 134 234 12.4. Shared-Secret Security . . . . . . . . . . . . . . . . . 135 235 12.5. Storage Security . . . . . . . . . . . . . . . . . . . . 136 236 12.5.1. Authorization . . . . . . . . . . . . . . . . . . . 136 237 12.5.2. Distributed Quota . . . . . . . . . . . . . . . . . 137 238 12.5.3. Correctness . . . . . . . . . . . . . . . . . . . . 137 239 12.5.4. Residual Attacks . . . . . . . . . . . . . . . . . . 137 240 12.6. Routing Security . . . . . . . . . . . . . . . . . . . . 138 241 12.6.1. Background . . . . . . . . . . . . . . . . . . . . . 138 242 12.6.2. Admissions Control . . . . . . . . . . . . . . . . . 139 243 12.6.3. Peer Identification and Authentication . . . . . . . 139 244 12.6.4. Protecting the Signaling . . . . . . . . . . . . . . 140 245 12.6.5. Residual Attacks . . . . . . . . . . . . . . . . . . 140 246 13. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 141 247 13.1. Well-Known URI Registration . . . . . . . . . . . . . . 141 248 13.2. Port Registrations . . . . . . . . . . . . . . . . . . . 141 249 13.3. Overlay Algorithm Types . . . . . . . . . . . . . . . . 142 250 13.4. Access Control Policies . . . . . . . . . . . . . . . . 142 251 13.5. Application-ID . . . . . . . . . . . . . . . . . . . . . 142 252 13.6. Data Kind-ID . . . . . . . . . . . . . . . . . . . . . . 143 253 13.7. Data Model . . . . . . . . . . . . . . . . . . . . . . . 143 254 13.8. Message Codes . . . . . . . . . . . . . . . . . . . . . 143 255 13.9. Error Codes . . . . . . . . . . . . . . . . . . . . . . 144 256 13.10. Overlay Link Types . . . . . . . . . . . . . . . . . . . 145 257 13.11. Overlay Link Protocols . . . . . . . . . . . . . . . . . 145 258 13.12. Forwarding Options . . . . . . . . . . . . . . . . . . . 146 259 13.13. Probe Information Types . . . . . . . . . . . . . . . . 146 260 13.14. Message Extensions . . . . . . . . . . . . . . . . . . . 146 261 13.15. reload URI Scheme . . . . . . . . . . . . . . . . . . . 147 262 13.15.1. URI Registration . . . . . . . . . . . . . . . . . . 147 263 13.16. Media Type Registration . . . . . . . . . . . . . . . . 148 264 14. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 149 265 15. References . . . . . . . . . . . . . . . . . . . . . . . . . 150 266 15.1. Normative References . . . . . . . . . . . . . . . . . . 150 267 15.2. Informative References . . . . . . . . . . . . . . . . . 151 268 Appendix A. Routing Alternatives . . . . . . . . . . . . . . . . 155 269 A.1. Iterative vs Recursive . . . . . . . . . . . . . . . . . 155 270 A.2. Symmetric vs Forward response . . . . . . . . . . . . . 155 271 A.3. Direct Response . . . . . . . . . . . . . . . . . . . . 156 272 A.4. Relay Peers . . . . . . . . . . . . . . . . . . . . . . 157 273 A.5. Symmetric Route Stability . . . . . . . . . . . . . . . 157 274 Appendix B. Why Clients? . . . . . . . . . . . . . . . . . . . . 158 275 B.1. Why Not Only Peers? . . . . . . . . . . . . . . . . . . 158 276 B.2. Clients as Application-Level Agents . . . . . . . . . . 159 277 Appendix C. Change Log . . . . . . . . . . . . . . . . . . . . . 159 278 C.1. Changes since draft-ietf-p2psip-reload-13 . . . . . . . 159 279 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 159 281 1. Introduction 283 This document defines REsource LOcation And Discovery (RELOAD), a 284 peer-to-peer (P2P) signaling protocol for use on the Internet. It 285 provides a generic, self-organizing overlay network service, allowing 286 nodes to efficiently route messages to other nodes and to efficiently 287 store and retrieve data in the overlay. RELOAD provides several 288 features that are critical for a successful P2P protocol for the 289 Internet: 291 Security Framework: A P2P network will often be established among a 292 set of peers that do not trust each other. RELOAD leverages a 293 central enrollment server to provide credentials for each peer 294 which can then be used to authenticate each operation. This 295 greatly reduces the possible attack surface. 297 Usage Model: RELOAD is designed to support a variety of 298 applications, including P2P multimedia communications with the 299 Session Initiation Protocol [I-D.ietf-p2psip-sip]. RELOAD allows 300 the definition of new application usages, each of which can define 301 its own data types, along with the rules for their use. This 302 allows RELOAD to be used with new applications through a simple 303 documentation process that supplies the details for each 304 application. 306 NAT Traversal: RELOAD is designed to function in environments where 307 many if not most of the nodes are behind NATs or firewalls. 308 Operations for NAT traversal are part of the base design, 309 including using ICE to establish new RELOAD or application 310 protocol connections. 312 High Performance Routing: The very nature of overlay algorithms 313 introduces a requirement that peers participating in the P2P 314 network route requests on behalf of other peers in the network. 315 This introduces a load on those other peers, in the form of 316 bandwidth and processing power. RELOAD has been defined with a 317 simple, lightweight forwarding header, thus minimizing the amount 318 of effort required by intermediate peers. 320 Pluggable Overlay Algorithms: RELOAD has been designed with an 321 abstract interface to the overlay layer to simplify implementing a 322 variety of structured (e.g., distributed hash tables) and 323 unstructured overlay algorithms. This specification also defines 324 how RELOAD is used with the Chord DHT algorithm, which is 325 mandatory to implement. Specifying a default "must implement" 326 overlay algorithm promotes interoperability, while extensibility 327 allows selection of overlay algorithms optimized for a particular 328 application. 330 These properties were designed specifically to meet the requirements 331 for a P2P protocol to support SIP. This document defines the base 332 protocol for the distributed storage and location service, as well as 333 critical usages for NAT traversal and security. The SIP Usage itself 334 is described separately in [I-D.ietf-p2psip-sip]. RELOAD is not 335 limited to usage by SIP and could serve as a tool for supporting 336 other P2P applications with similar needs. RELOAD is also based on 337 the concepts introduced in [I-D.ietf-p2psip-concepts]. 339 1.1. Basic Setting 341 In this section, we provide a brief overview of the operational 342 setting for RELOAD. See the concepts 343 document[I-D.ietf-p2psip-concepts] for more details. A RELOAD 344 Overlay Instance consists of a set of nodes arranged in a connected 345 graph. Each node in the overlay is assigned a numeric Node-ID which, 346 together with the specific overlay algorithm in use, determines its 347 position in the graph and the set of nodes it connects to. The 348 figure below shows a trivial example which isn't drawn from any 349 particular overlay algorithm, but was chosen for convenience of 350 representation. 352 +--------+ +--------+ +--------+ 353 | Node 10|--------------| Node 20|--------------| Node 30| 354 +--------+ +--------+ +--------+ 355 | | | 356 | | | 357 +--------+ +--------+ +--------+ 358 | Node 40|--------------| Node 50|--------------| Node 60| 359 +--------+ +--------+ +--------+ 360 | | | 361 | | | 362 +--------+ +--------+ +--------+ 363 | Node 70|--------------| Node 80|--------------| Node 90| 364 +--------+ +--------+ +--------+ 365 | 366 | 367 +--------+ 368 | Node 85| 369 |(Client)| 370 +--------+ 372 Because the graph is not fully connected, when a node wants to send a 373 message to another node, it may need to route it through the network. 374 For instance, Node 10 can talk directly to nodes 20 and 40, but not 375 to Node 70. In order to send a message to Node 70, it would first 376 send it to Node 40 with instructions to pass it along to Node 70. 377 Different overlay algorithms will have different connectivity graphs, 378 but the general idea behind all of them is to allow any node in the 379 graph to efficiently reach every other node within a small number of 380 hops. 382 The RELOAD network is not only a messaging network. It is also a 383 storage network. Records are stored under numeric addresses which 384 occupy the same space as node identifiers. Peers are responsible for 385 storing the data associated with some set of addresses as determined 386 by their Node-ID. For instance, we might say that every peer is 387 responsible for storing any data value which has an address less than 388 or equal to its own Node-ID, but greater than the next lowest 389 Node-ID. Thus, Node-20 would be responsible for storing values 390 11-20. 392 RELOAD also supports clients. These are nodes which have Node-IDs 393 but do not participate in routing or storage. For instance, in the 394 figure above Node 85 is a client. It can route to the rest of the 395 RELOAD network via Node 80, but no other node will route through it 396 and Node 90 is still responsible for all addresses between 81-90. We 397 refer to non-client nodes as peers. 399 Other applications (for instance, SIP) can be defined on top of 400 RELOAD and use these two basic RELOAD services to provide their own 401 services. 403 1.2. Architecture 405 RELOAD is fundamentally an overlay network. The following figure 406 shows the layered RELOAD architecture. 408 Application 410 +-------+ +-------+ 411 | SIP | | XMPP | ... 412 | Usage | | Usage | 413 +-------+ +-------+ 414 ------------------------------------ Messaging Service Boundary 415 +------------------+ +---------+ 416 | Message |<--->| Storage | 417 | Transport | +---------+ 418 +------------------+ ^ 419 ^ ^ | 420 | v v 421 | +-------------------+ 422 | | Topology | 423 | | Plugin | 424 | +-------------------+ 425 | ^ 426 v v 427 +------------------+ 428 | Forwarding & | 429 | Link Management | 430 +------------------+ 431 ------------------------------------ Overlay Link Service Boundary 432 +-------+ +------+ 433 |TLS | |DTLS | ... 434 +-------+ +------+ 436 The major components of RELOAD are: 438 Usage Layer: Each application defines a RELOAD usage; a set of data 439 kinds and behaviors which describe how to use the services 440 provided by RELOAD. These usages all talk to RELOAD through a 441 common Message Transport Service. 443 Message Transport: Handles end-to-end reliability, manages request 444 state for the usages, and forwards Store and Fetch operations to 445 the Storage component. Delivers message responses to the 446 component initiating the request. 448 Storage: The Storage component is responsible for processing 449 messages relating to the storage and retrieval of data. It talks 450 directly to the Topology Plugin to manage data replication and 451 migration, and it talks to the Message Transport component to send 452 and receive messages. 454 Topology Plugin: The Topology Plugin is responsible for implementing 455 the specific overlay algorithm being used. It uses the Message 456 Transport component to send and receive overlay management 457 messages, to the Storage component to manage data replication, and 458 directly to the Forwarding Layer to control hop-by-hop message 459 forwarding. This component closely parallels conventional routing 460 algorithms, but is more tightly coupled to the Forwarding Layer 461 because there is no single "routing table" equivalent used by all 462 overlay algorithms. 464 Forwarding and Link Management Layer: Stores and implements the 465 routing table by providing packet forwarding services between 466 nodes. It also handles establishing new links between nodes, 467 including setting up connections across NATs using ICE. 469 Overlay Link Layer: Responsible for actually transporting traffic 470 directly between nodes. Each such protocol includes the 471 appropriate provisions for per-hop framing or hop-by-hop ACKs 472 required by unreliable transports. TLS [RFC5246] and DTLS 473 [RFC4347] are the currently defined "link layer" protocols used by 474 RELOAD for hop-by-hop communication. New protocols MAY be 475 defined, as described in Section 5.6.1 and Section 10.1. As this 476 document defines only TLS and DTLS, we use those terms throughout 477 the remainder of the document with the understanding that some 478 future specification may add new overlay link layers. 480 To further clarify the roles of the various layers, this figure 481 parallels the architecture with each layer's role from an overlay 482 perspective and implementation layer in the internet: 484 | Internet Model | 485 Real | Equivalent | Reload 486 Internet | in Overlay | Architecture 487 -------------+-----------------+------------------------------------ 488 | | +-------+ +-------+ 489 | Application | | SIP | | XMPP | ... 490 | | | Usage | | Usage | 491 | | +-------+ +-------+ 492 | | ---------------------------------- 493 | |+------------------+ +---------+ 494 | Transport || Message |<--->| Storage | 495 | || Transport | +---------+ 496 | |+------------------+ ^ 497 | | ^ ^ | 498 | | | v v 499 Application | | | +-------------------+ 500 | (Routing) | | | Topology | 501 | | | | Plugin | 502 | | | +-------------------+ 503 | | | ^ 504 | | v v 505 | Network | +------------------+ 506 | | | Forwarding & | 507 | | | Link Management | 508 | | +------------------+ 509 | | ---------------------------------- 510 Transport | Link | +-------+ +------+ 511 | | |TLS | |DTLS | ... 512 | | +-------+ +------+ 513 -------------+-----------------+------------------------------------ 514 Network | 515 | 516 Link | 518 1.2.1. Usage Layer 520 The top layer, called the Usage Layer, has application usages, such 521 as the SIP Registration Usage [I-D.ietf-p2psip-sip], that use the 522 abstract Message Transport Service provided by RELOAD. The goal of 523 this layer is to implement application-specific usages of the generic 524 overlay services provided by RELOAD. The usage defines how a 525 specific application maps its data into something that can be stored 526 in the overlay, where to store the data, how to secure the data, and 527 finally how applications can retrieve and use the data. 529 The architecture diagram shows both a SIP usage and an XMPP usage. A 530 single application may require multiple usages; for example a 531 softphone application may also require a voicemail usage. An usage 532 may define multiple kinds of data that are stored in the overlay and 533 may also rely on kinds originally defined by other usages. 535 Because the security and storage policies for each kind are dictated 536 by the usage defining the kind, the usages may be coupled with the 537 Storage component to provide security policy enforcement and to 538 implement appropriate storage strategies according to the needs of 539 the usage. The exact implementation of such an interface is outside 540 the scope of this specification. 542 1.2.2. Message Transport 544 The Message Transport component provides a generic message routing 545 service for the overlay. The Message Transport layer is responsible 546 for end-to-end message transactions, including retransmissions. Each 547 peer is identified by its location in the overlay as determined by 548 its Node-ID. A component that is a client of the Message Transport 549 can perform two basic functions: 551 o Send a message to a given peer specified by Node-ID or to the peer 552 responsible for a particular Resource-ID. 553 o Receive messages that other peers sent to a Node-ID or Resource-ID 554 for which the receiving peer is responsible. 556 All usages rely on the Message Transport component to send and 557 receive messages from peers. For instance, when a usage wants to 558 store data, it does so by sending Store requests. Note that the 559 Storage component and the Topology Plugin are themselves clients of 560 the Message Transport, because they need to send and receive messages 561 from other peers. 563 The Message Transport Service is similar to those described as 564 providing "Key based routing" (KBR), although as RELOAD supports 565 different overlay algorithms (including non-DHT overlay algorithms) 566 that calculate keys in different ways, the actual interface must 567 accept Resource Names rather than actual keys. 569 1.2.3. Storage 571 One of the major functions of RELOAD is to allow nodes to store data 572 in the overlay and to retrieve data stored by other nodes or by 573 themselves. The Storage component is responsible for processing data 574 storage and retrieval messages. For instance, the Storage component 575 might receive a Store request for a given resource from the Message 576 Transport. It would then query the appropriate usage before storing 577 the data value(s) in its local data store and sending a response to 578 the Message Transport for delivery to the requesting node. 579 Typically, these messages will come from other nodes, but depending 580 on the overlay topology, a node might be responsible for storing data 581 for itself as well, especially if the overlay is small. 583 A peer's Node-ID determines the set of resources that it will be 584 responsible for storing. However, the exact mapping between these is 585 determined by the overlay algorithm in use. The Storage component 586 will only receive a Store request from the Message Transport if this 587 peer is responsible for that Resource-ID. The Storage component is 588 notified by the Topology Plugin when the Resource-IDs for which it is 589 responsible change, and the Storage component is then responsible for 590 migrating resources to other peers, as required. 592 1.2.4. Topology Plugin 594 RELOAD is explicitly designed to work with a variety of overlay 595 algorithms. In order to facilitate this, the overlay algorithm 596 implementation is provided by a Topology Plugin so that each overlay 597 can select an appropriate overlay algorithm that relies on the common 598 RELOAD core protocols and code. 600 The Topology Plugin is responsible for maintaining the overlay 601 algorithm Routing Table, which is consulted by the Forwarding and 602 Link Management Layer before routing a message. When connections are 603 made or broken, the Forwarding and Link Management Layer notifies the 604 Topology Plugin, which adjusts the routing table as appropriate. The 605 Topology Plugin will also instruct the Forwarding and Link Management 606 Layer to form new connections as dictated by the requirements of the 607 overlay algorithm Topology. The Topology Plugin issues periodic 608 update requests through Message Transport to maintain and update its 609 Routing Table. 611 As peers enter and leave, resources may be stored on different peers, 612 so the Topology Plugin also keeps track of which peers are 613 responsible for which resources. As peers join and leave, the 614 Topology Plugin instructs the Storage component to issue resource 615 migration requests as appropriate, in order to ensure that other 616 peers have whatever resources they are now responsible for. The 617 Topology Plugin is also responsible for providing for redundant data 618 storage to protect against loss of information in the event of a peer 619 failure and to protect against compromised or subversive peers. 621 1.2.5. Forwarding and Link Management Layer 623 The Forwarding and Link Management Layer is responsible for getting a 624 message to the next peer, as determined by the Topology Plugin. This 625 Layer establishes and maintains the network connections as required 626 by the Topology Plugin. This layer is also responsible for setting 627 up connections to other peers through NATs and firewalls using ICE, 628 and it can elect to forward traffic using relays for NAT and firewall 629 traversal. 631 This layer provides a generic interface that allows the topology 632 plugin to control the overlay and resource operations and messages. 633 Since each overlay algorithm is defined and functions differently, we 634 generically refer to the table of other peers that the overlay 635 algorithm maintains and uses to route requests (neighbors) as a 636 Routing Table. The Topology Plugin actually owns the Routing Table, 637 and forwarding decisions are made by querying the Topology Plugin for 638 the next hop for a particular Node-ID or Resource-ID. If this node 639 is the destination of the message, the message is delivered to the 640 Message Transport. 642 This layer also utilizes a framing header to encapsulate messages as 643 they are forwarding along each hop. This header aids reliability 644 congestion control, flow control, etc. It has meaning only in the 645 context of that individual link. 647 The Forwarding and Link Management Layer sits on top of the Overlay 648 Link Layer protocols that carry the actual traffic. This 649 specification defines how to use DTLS and TLS protocols to carry 650 RELOAD messages. 652 1.3. Security 654 RELOAD's security model is based on each node having one or more 655 public key certificates. In general, these certificates will be 656 assigned by a central server which also assigns Node-IDs, although 657 self-signed certificates can be used in closed networks. These 658 credentials can be leveraged to provide communications security for 659 RELOAD messages. RELOAD provides communications security at three 660 levels: 662 Connection Level: Connections between peers are secured with TLS, 663 DTLS, or potentially some to be defined future protocol. 664 Message Level: Each RELOAD message must be signed. 665 Object Level: Stored objects must be signed by the creating peer. 667 These three levels of security work together to allow peers to verify 668 the origin and correctness of data they receive from other peers, 669 even in the face of malicious activity by other peers in the overlay. 670 RELOAD also provides access control built on top of these 671 communications security features. Because the peer responsible for 672 storing a piece of data can validate the signature on the data being 673 stored, the responsible peer can determine whether a given operation 674 is permitted or not. 676 RELOAD also provides an optional shared secret based admission 677 control feature using shared secrets and TLS-PSK. In order to form a 678 TLS connection to any node in the overlay, a new node needs to know 679 the shared overlay key, thus restricting access to authorized users 680 only. This feature is used together with certificate-based access 681 control, not as a replacement for it. It is typically used when 682 self-signed certificates are being used but would generally not be 683 used when the certificates were all signed by an enrollment server. 685 1.4. Structure of This Document 687 The remainder of this document is structured as follows. 689 o Section 2 provides definitions of terms used in this document. 690 o Section 3 provides an overview of the mechanisms used to establish 691 and maintain the overlay. 692 o Section 4 provides an overview of the mechanism RELOAD provides to 693 support other applications. 694 o Section 5 defines the protocol messages that RELOAD uses to 695 establish and maintain the overlay. 696 o Section 6 defines the protocol messages that are used to store and 697 retrieve data using RELOAD. 698 o Section 7 defines the Certificate Store Usage that is fundamental 699 to RELOAD security. 700 o Section 8 defines the TURN Server Usage needed to locate TURN 701 servers for NAT traversal. 702 o Section 9 defines a specific Topology Plugin using Chord. 703 o Section 10 defines the mechanisms that new RELOAD nodes use to 704 join the overlay for the first time. 705 o Section 11 provides an extended example. 707 2. Terminology 709 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 710 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 711 document are to be interpreted as described in RFC 2119 [RFC2119]. 713 We use the terminology and definitions from the Concepts and 714 Terminology for Peer to Peer SIP [I-D.ietf-p2psip-concepts] draft 715 extensively in this document. Other terms used in this document are 716 defined inline when used and are also defined below for reference. 718 DHT: A distributed hash table. A DHT is an abstract hash table 719 service realized by storing the contents of the hash table across 720 a set of peers. 722 Overlay Algorithm: An overlay algorithm defines the rules for 723 determining which peers in an overlay store a particular piece of 724 data and for determining a topology of interconnections amongst 725 peers in order to find a piece of data. 727 Overlay Instance: A specific overlay algorithm and the collection of 728 peers that are collaborating to provide read and write access to 729 it. There can be any number of overlay instances running in an IP 730 network at a time, and each operates in isolation of the others. 732 Peer: A host that is participating in the overlay. Peers are 733 responsible for holding some portion of the data that has been 734 stored in the overlay and also route messages on behalf of other 735 hosts as required by the Overlay Algorithm. 737 Client: A host that is able to store data in and retrieve data from 738 the overlay but which is not participating in routing or data 739 storage for the overlay. 741 Kind: A kind defines a particular type of data that can be stored in 742 the overlay. Applications define new Kinds to story the data they 743 use. Each Kind is identified with a unique integer called a 744 Kind-ID. 746 Node: We use the term "Node" to refer to a host that may be either a 747 Peer or a Client. Because RELOAD uses the same protocol for both 748 clients and peers, much of the text applies equally to both. 749 Therefore we use "Node" when the text applies to both Clients and 750 Peers and the more specific term (i.e. client or peer) when the 751 text applies only to Clients or only to Peers. 753 Node-ID: A fixed-length value that uniquely identifies a node. 754 Node-IDs of all 0s and all 1s are reserved and are invalid Node- 755 IDs. A value of zero is not used in the wire protocol but can be 756 used to indicate an invalid node in implementations and APIs. The 757 Node-ID of all 1s is used on the wire protocol as a wildcard. 759 Resource: An object or group of objects associated with a string 760 identifier. See "Resource Name" below. 762 Resource Name: The potentially human readable name by which a 763 resource is identified. In unstructured P2P networks, the 764 resource name is sometimes used directly as a Resource-ID. In 765 structured P2P networks the resource name is typically mapped into 766 a Resource-ID by using the string as the input to hash function. 767 A SIP resource, for example, is often identified by its AOR which 768 is an example of a Resource Name. 770 Resource-ID: A value that identifies some resources and which is 771 used as a key for storing and retrieving the resource. Often this 772 is not human friendly/readable. One way to generate a Resource-ID 773 is by applying a mapping function to some other unique name (e.g., 774 user name or service name) for the resource. The Resource-ID is 775 used by the distributed database algorithm to determine the peer 776 or peers that are responsible for storing the data for the 777 overlay. In structured P2P networks, Resource-IDs are generally 778 fixed length and are formed by hashing the resource name. In 779 unstructured networks, resource names may be used directly as 780 Resource-IDs and may be variable lengths. 782 Connection Table: The set of nodes to which a node is directly 783 connected. This includes nodes with which Attach handshakes have 784 been done but which have not sent any Updates. 786 Routing Table: The set of peers which a node can use to route 787 overlay messages. In general, these peers will all be on the 788 connection table but not vice versa, because some peers will have 789 Attached but not sent updates. Peers may send messages directly 790 to peers that are in the connection table but may only route 791 messages to other peers through peers that are in the routing 792 table. 794 Destination List: A list of IDs through which a message is to be 795 routed. A single Node-ID is a trivial form of destination list. 797 Usage: A usage is an application that wishes to use the overlay for 798 some purpose. Each application wishing to use the overlay defines 799 a set of data kinds that it wishes to use. The SIP usage defines 800 the location data kind. 802 The term "maximum request lifetime" is the maximum time a request 803 will wait for a response; it defaults to 15 seconds. The term 804 "successor replacement hold-down time" is the amount of time to wait 805 before starting replication when a new successor is found; it 806 defaults to 30 seconds. 808 3. Overlay Management Overview 810 The most basic function of RELOAD is as a generic overlay network. 811 Nodes need to be able to join the overlay, form connections to other 812 nodes, and route messages through the overlay to nodes to which they 813 are not directly connected. This section provides an overview of the 814 mechanisms that perform these functions. 816 3.1. Security and Identification 818 Every node in the RELOAD overlay is identified by a Node-ID. The 819 Node-ID is used for three major purposes: 821 o To address the node itself. 822 o To determine its position in the overlay topology when the overlay 823 is structured. 824 o To determine the set of resources for which the node is 825 responsible. 827 Each node has a certificate [RFC5280] containing a Node-ID, which is 828 unique within an overlay instance. 830 The certificate serves multiple purposes: 832 o It entitles the user to store data at specific locations in the 833 Overlay Instance. Each data kind defines the specific rules for 834 determining which certificates can access each Resource-ID/Kind-ID 835 pair. For instance, some kinds might allow anyone to write at a 836 given location, whereas others might restrict writes to the owner 837 of a single certificate. 838 o It entitles the user to operate a node that has a Node-ID found in 839 the certificate. When the node forms a connection to another 840 peer, it uses this certificate so that a node connecting to it 841 knows it is connected to the correct node (technically: a (D)TLS 842 association with client authentication is formed.) In addition, 843 the node can sign messages, thus providing integrity and 844 authentication for messages which are sent from the node. 845 o It entitles the user to use the user name found in the 846 certificate. 848 If a user has more than one device, typically they would get one 849 certificate for each device. This allows each device to act as a 850 separate peer. 852 RELOAD supports multiple certificate issuance models. The first is 853 based on a central enrollment process which allocates a unique name 854 and Node-ID and puts them in a certificate for the user. All peers 855 in a particular Overlay Instance have the enrollment server as a 856 trust anchor and so can verify any other peer's certificate. 858 In some settings, a group of users want to set up an overlay network 859 but are not concerned about attack by other users in the network. 860 For instance, users on a LAN might want to set up a short term ad hoc 861 network without going to the trouble of setting up an enrollment 862 server. RELOAD supports the use of self-generated, self-signed 863 certificates. When self-signed certificates are used, the node also 864 generates its own Node-ID and username. The Node-ID is computed as a 865 digest of the public key, to prevent Node-ID theft; however this 866 model is still subject to a number of known attacks (most notably 867 Sybil attacks [Sybil]) and can only be safely used in closed networks 868 where users are mutually trusting. 870 The general principle here is that the security mechanisms (TLS and 871 message signatures) are always used, even if the certificates are 872 self-signed. This allows for a single set of code paths in the 873 systems with the only difference being whether certificate 874 verification is required to chain to a single root of trust. 876 3.1.1. Shared-Key Security 878 RELOAD also provides an admission control system based on shared 879 keys. In this model, the peers all share a single key which is used 880 to authenticate the peer-to-peer connections via TLS-PSK/TLS-SRP. 882 3.2. Clients 884 RELOAD defines a single protocol that is used both as the peer 885 protocol and as the client protocol for the overlay. This simplifies 886 implementation, particularly for devices that may act in either role, 887 and allows clients to inject messages directly into the overlay. 889 We use the term "peer" to identify a node in the overlay that routes 890 messages for nodes other than those to which it is directly 891 connected. Peers typically also have storage responsibilities. We 892 use the term "client" to refer to nodes that do not have routing or 893 storage responsibilities. When text applies to both peers and 894 clients, we will simply refer such devices as "nodes." 896 RELOAD's client support allows nodes that are not participating in 897 the overlay as peers to utilize the same implementation and to 898 benefit from the same security mechanisms as the peers. Clients 899 possess and use certificates that authorize the user to store data at 900 certain locations in the overlay. The Node-ID in the certificate is 901 used to identify the particular client as a member of the overlay and 902 to authenticate its messages. 904 In RELOAD, unlike some other designs, clients are not a first-class 905 concept. From the perspective of a peer, a client is simply a node 906 which has not yet sent any Updates or Joins. It might never do so 907 (if it's a client) or it might eventually do so (if it's just a node 908 that's taking a long time to join). The routing and storage rules 909 for RELOAD provide for correct behavior by peers regardless of 910 whether other nodes attached to them are clients or peers. Of 911 course, a client implementation must know that it intends to be a 912 client, but this localizes complexity only to that node. 914 For more discussion of the motivation for RELOAD's client support, 915 see Appendix B. 917 3.2.1. Client Routing 919 Clients may insert themselves in the overlay in two ways: 921 o Establish a connection to the peer responsible for the client's 922 Node-ID in the overlay. Then requests may be sent from/to the 923 client using its Node-ID in the same manner as if it were a peer, 924 because the responsible peer in the overlay will handle the final 925 step of routing to the client. This may require a TURN relay in 926 cases where NATs or firewalls prevent a client from forming a 927 direct connections with its responsible peer. Note that clients 928 that choose this option MUST process Update messages from the 929 peer. Those updates can indicate that the peer no longer is 930 responsible for the Client's Node-ID. The client then MUST form a 931 connection to the appropriate peer. Failure to do so will result 932 in the client no longer receiving messages. 933 o Establish a connection with an arbitrary peer in the overlay 934 (perhaps based on network proximity or an inability to establish a 935 direct connection with the responsible peer). In this case, the 936 client will rely on RELOAD's Destination List feature to ensure 937 reachability. The client can initiate requests, and any node in 938 the overlay that knows the Destination List to its current 939 location can reach it, but the client is not directly reachable 940 using only its Node-ID. If the client is to receive incoming 941 requests from other members of the overlay, the Destination List 942 required to reach it must be learnable via other mechanisms, such 943 as being stored in the overlay by a usage. A client connected 944 this way using a certificate with only a single Node-ID MAY 945 proceed to use the connection without performing an Attach. A 946 client wishing to connect using this mechanism with a certificate 947 with multiple Node-IDs MUST perform an Attach after using Ping to 948 probe the Node-ID of the node to which it is connected. 950 3.2.2. Minimum Functionality Requirements for Clients 952 A node may act as a client simply because it does not have the 953 resources or even an implementation of the topology plugin required 954 to act as a peer in the overlay. In order to exchange RELOAD 955 messages with a peer, a client must meet a minimum level of 956 functionality. Such a client must: 958 o Implement RELOAD's connection-management operations that are used 959 to establish the connection with the peer. 960 o Implement RELOAD's data retrieval methods (with client 961 functionality). 962 o Be able to calculate Resource-IDs used by the overlay. 963 o Possess security credentials required by the overlay it is 964 implementing. 966 A client speaks the same protocol as the peers, knows how to 967 calculate Resource-IDs, and signs its requests in the same manner as 968 peers. While a client does not necessarily require a full 969 implementation of the overlay algorithm, calculating the Resource-ID 970 requires an implementation of the appropriate algorithm for the 971 overlay. 973 3.3. Routing 975 This section will discuss the requirements RELOAD's routing 976 capabilities must meet, then describe the routing features in the 977 protocol, and then provide a brief overview of how they are used. 978 Appendix A discusses some alternative designs and the tradeoffs that 979 would be necessary to support them. 981 RELOAD's routing capabilities must meet the following requirements: 983 NAT Traversal: RELOAD must support establishing and using 984 connections between nodes separated by one or more NATs, including 985 locating peers behind NATs for those overlays allowing/requiring 986 it. 987 Clients: RELOAD must support requests from and to clients that do 988 not participate in overlay routing. 989 Client promotion: RELOAD must support clients that become peers at a 990 later point as determined by the overlay algorithm and deployment. 991 Low state: RELOAD's routing algorithms must not require 992 significant state to be stored on intermediate peers. 993 Return routability in unstable topologies: At some points in 994 times, different nodes may have inconsistent information about the 995 connectivity of the routing graph. In all cases, the response to 996 a request needs to delivered to the node that sent the request and 997 not to some other node. 999 RELOAD's routing provides three mechanisms designed to assist in 1000 meeting these needs: 1002 Destination Lists: While in principle it is possible to just 1003 inject a message into the overlay with a bare Node-ID as the 1004 destination, RELOAD provides a source routing capability in the 1005 form of "Destination Lists". A "Destination List provides a list 1006 of the nodes through which a message must flow. 1007 Via Lists: In order to allow responses to follow the same path as 1008 requests, each message also contains a "Via List", which is added 1009 to by each node a message traverses. This via list can then be 1010 inverted and used as a destination list for the response. 1011 RouteQuery: The RouteQuery method allows a node to query a peer 1012 for the next hop it will use to route a message. This method is 1013 useful for diagnostics and for iterative routing. 1015 The basic routing mechanism used by RELOAD is Symmetric Recursive. 1016 We will first describe symmetric recursive routing and then discuss 1017 its advantages in terms of the requirements discussed above. 1019 Symmetric recursive routing requires that a message follow a path 1020 through the overlay to the destination without returning to the 1021 originating node: each peer forwards the message closer to its 1022 destination. The return path of the response is then the same path 1023 followed in reverse. For example, a message following a route from A 1024 to Z through B and X: 1026 A B X Z 1027 ------------------------------- 1029 ----------> 1030 Dest=Z 1031 ----------> 1032 Via=A 1033 Dest=Z 1034 ----------> 1035 Via=A, B 1036 Dest=Z 1038 <---------- 1039 Dest=X, B, A 1040 <---------- 1041 Dest=B, A 1042 <---------- 1043 Dest=A 1045 Note that the preceding Figure does not indicate whether A is a 1046 client or peer: A forwards its request to B and the response is 1047 returned to A in the same manner regardless of A's role in the 1048 overlay. 1050 This figure shows use of full via-lists by intermediate peers B and 1051 X. However, if B and/or X are willing to store state, then they may 1052 elect to truncate the lists, save that information internally (keyed 1053 by the transaction id), and return the response message along the 1054 path from which it was received when the response is received. This 1055 option requires greater state to be stored on intermediate peers but 1056 saves a small amount of bandwidth and reduces the need for modifying 1057 the message en route. Selection of this mode of operation is a 1058 choice for the individual peer; the techniques are interoperable even 1059 on a single message. The figure below shows B using full via lists 1060 but X truncating them to X1 and saving the state internally. 1062 A B X Z 1063 ------------------------------- 1065 ----------> 1066 Dest=Z 1067 ----------> 1068 Via=A 1069 Dest=Z 1070 ----------> 1071 Dest=Z, X1 1073 <---------- 1074 Dest=X,X1 1075 <---------- 1076 Dest=B, A 1077 <---------- 1078 Dest=A 1080 RELOAD also supports a basic Iterative routing mode (where the 1081 intermediate peers merely return a response indicating the next hop, 1082 but do not actually forward the message to that next hop themselves). 1083 Iterative routing is implemented using the RouteQuery method, which 1084 requests this behavior. Note that iterative routing is selected only 1085 by the initiating node. 1087 3.4. Connectivity Management 1089 In order to provide efficient routing, a peer needs to maintain a set 1090 of direct connections to other peers in the Overlay Instance. Due to 1091 the presence of NATs, these connections often cannot be formed 1092 directly. Instead, we use the Attach request to establish a 1093 connection. Attach uses ICE [RFC5245] to establish the connection. 1095 It is assumed that the reader is familiar with ICE. 1097 Say that peer A wishes to form a direct connection to peer B. It 1098 gathers ICE candidates and packages them up in an Attach request 1099 which it sends to B through usual overlay routing procedures. B does 1100 its own candidate gathering and sends back a response with its 1101 candidates. A and B then do ICE connectivity checks on the candidate 1102 pairs. The result is a connection between A and B. At this point, A 1103 and B can add each other to their routing tables and send messages 1104 directly between themselves without going through other overlay 1105 peers. 1107 There are two cases where Attach is not used. The first is when a 1108 peer is joining the overlay and is not connected to any peers. In 1109 order to support this case, some small number of "bootstrap nodes" 1110 typically need to be publicly accessible so that new peers can 1111 directly connect to them. Section 10 contains more detail on this. 1112 The second case is when a client node connects to a node at an 1113 arbitrary IP address, rather than to its responsible peer, as 1114 described in the second bullet point of Section 3.2.1. 1116 In general, a peer needs to maintain connections to all of the peers 1117 near it in the Overlay Instance and to enough other peers to have 1118 efficient routing (the details depend on the specific overlay). If a 1119 peer cannot form a connection to some other peer, this isn't 1120 necessarily a disaster; overlays can route correctly even without 1121 fully connected links. However, a peer should try to maintain the 1122 specified link set and if it detects that it has fewer direct 1123 connections, should form more as required. This also implies that 1124 peers need to periodically verify that the connected peers are still 1125 alive and if not try to reform the connection or form an alternate 1126 one. 1128 3.5. Overlay Algorithm Support 1130 The Topology Plugin allows RELOAD to support a variety of overlay 1131 algorithms. This specification defines a DHT based on Chord [Chord], 1132 which is mandatory to implement, but the base RELOAD protocol is 1133 designed to support a variety of overlay algorithms. 1135 3.5.1. Support for Pluggable Overlay Algorithms 1137 RELOAD defines three methods for overlay maintenance: Join, Update, 1138 and Leave. However, the contents of those messages, when they are 1139 sent, and their precise semantics are specified by the actual overlay 1140 algorithm; RELOAD merely provides a framework of commonly-needed 1141 methods that provides uniformity of notation (and ease of debugging) 1142 for a variety of overlay algorithms. 1144 3.5.2. Joining, Leaving, and Maintenance Overview 1146 When a new peer wishes to join the Overlay Instance, it must have a 1147 Node-ID that it is allowed to use and a set of credentials which 1148 match that Node-ID. When an enrollment server is used that Node-ID 1149 will be in the certificate the node received from the enrollment 1150 server. The details of the joining procedure are defined by the 1151 overlay algorithm, but the general steps for joining an Overlay 1152 Instance are: 1154 o Forming connections to some other peers. 1155 o Acquiring the data values this peer is responsible for storing. 1156 o Informing the other peers which were previously responsible for 1157 that data that this peer has taken over responsibility. 1159 The first thing the peer needs to do is to form a connection to some 1160 "bootstrap node". Because this is the first connection the peer 1161 makes, these nodes must have public IP addresses so that they can be 1162 connected to directly. Once a peer has connected to one or more 1163 bootstrap nodes, it can form connections in the usual way by routing 1164 Attach messages through the overlay to other nodes. Once a peer has 1165 connected to the overlay for the first time, it can cache the set of 1166 nodes it has connected to with public IP addresses for use as future 1167 bootstrap nodes. 1169 Once a peer has connected to a bootstrap node, it then needs to take 1170 up its appropriate place in the overlay. This requires two major 1171 operations: 1173 o Forming connections to other peers in the overlay to populate its 1174 Routing Table. 1175 o Getting a copy of the data it is now responsible for storing and 1176 assuming responsibility for that data. 1178 The second operation is performed by contacting the Admitting Peer 1179 (AP), the node which is currently responsible for that section of the 1180 overlay. 1182 The details of this operation depend mostly on the overlay algorithm 1183 involved, but a typical case would be: 1185 1. JP (Joining Peer) sends a Join request to AP (Admitting Peer) 1186 announcing its intention to join. 1187 2. AP sends a Join response. 1188 3. AP does a sequence of Stores to JP to give it the data it will 1189 need. 1191 4. AP does Updates to JP and to other peers to tell it about its own 1192 routing table. At this point, both JP and AP consider JP 1193 responsible for some section of the Overlay Instance. 1194 5. JP makes its own connections to the appropriate peers in the 1195 Overlay Instance. 1197 After this process is completed, JP is a full member of the Overlay 1198 Instance and can process Store/Fetch requests. 1200 Note that the first node is a special case. When ordinary nodes 1201 cannot form connections to the bootstrap nodes, then they are not 1202 part of the overlay. However, the first node in the overlay can 1203 obviously not connect to other nodes. In order to support this case, 1204 potential first nodes (which must also serve as bootstrap nodes 1205 initially) must somehow be instructed (perhaps by configuration 1206 settings) that they are the entire overlay, rather than not part of 1207 it. 1209 Note that clients do not perform either of these operations. 1211 3.6. First-Time Setup 1213 Previous sections addressed how RELOAD works once a node has 1214 connected. This section provides an overview of how users get 1215 connected to the overlay for the first time. RELOAD is designed so 1216 that users can start with the name of the overlay they wish to join 1217 and perhaps a username and password, and leverage that into having a 1218 working peer with minimal user intervention. This helps avoid the 1219 problems that have been experienced with conventional SIP clients 1220 where users are required to manually configure a large number of 1221 settings. 1223 3.6.1. Initial Configuration 1225 In the first phase of the process, the user starts out with the name 1226 of the overlay and uses this to download an initial set of overlay 1227 configuration parameters. The node does a DNS SRV lookup on the 1228 overlay name to get the address of a configuration server. It can 1229 then connect to this server with HTTPS to download a configuration 1230 document which contains the basic overlay configuration parameters as 1231 well as a set of bootstrap nodes which can be used to join the 1232 overlay. 1234 If a node already has the valid configuration document that it 1235 received by some out of band method, this step can be skipped. 1237 3.6.2. Enrollment 1239 If the overlay is using centralized enrollment, then a user needs to 1240 acquire a certificate before joining the overlay. The certificate 1241 attests both to the user's name within the overlay and to the Node- 1242 IDs which they are permitted to operate. In that case, the 1243 configuration document will contain the address of an enrollment 1244 server which can be used to obtain such a certificate. The 1245 enrollment server may (and probably will) require some sort of 1246 username and password before issuing the certificate. The enrollment 1247 server's ability to restrict attackers' access to certificates in the 1248 overlay is one of the cornerstones of RELOAD's security. 1250 4. Application Support Overview 1252 RELOAD is not intended to be used alone, but rather as a substrate 1253 for other applications. These applications can use RELOAD for a 1254 variety of purposes: 1256 o To store data in the overlay and retrieve data stored by other 1257 nodes. 1258 o As a discovery mechanism for services such as TURN. 1259 o To form direct connections which can be used to transmit 1260 application-level messages without using the overlay. 1262 This section provides an overview of these services. 1264 4.1. Data Storage 1266 RELOAD provides operations to Store and Fetch data. Each location in 1267 the Overlay Instance is referenced by a Resource-ID. However, each 1268 location may contain data elements corresponding to multiple kinds 1269 (e.g., certificate, SIP registration). Similarly, there may be 1270 multiple elements of a given kind, as shown below: 1272 +--------------------------------+ 1273 | Resource-ID | 1274 | | 1275 | +------------+ +------------+ | 1276 | | Kind 1 | | Kind 2 | | 1277 | | | | | | 1278 | | +--------+ | | +--------+ | | 1279 | | | Value | | | | Value | | | 1280 | | +--------+ | | +--------+ | | 1281 | | | | | | 1282 | | +--------+ | | +--------+ | | 1283 | | | Value | | | | Value | | | 1284 | | +--------+ | | +--------+ | | 1285 | | | +------------+ | 1286 | | +--------+ | | 1287 | | | Value | | | 1288 | | +--------+ | | 1289 | +------------+ | 1290 +--------------------------------+ 1292 Each kind is identified by a Kind-ID, which is a code point either 1293 assigned by IANA or allocated out of a private range. As part of the 1294 kind definition, protocol designers may define constraints, such as 1295 limits on size, on the values which may be stored. For many kinds, 1296 the set may be restricted to a single value; some sets may be allowed 1297 to contain multiple identical items while others may only have unique 1298 items. Note that a kind may be employed by multiple usages and new 1299 usages are encouraged to use previously defined kinds where possible. 1300 We define the following data models in this document, though other 1301 usages can define their own structures: 1303 single value: There can be at most one item in the set and any value 1304 overwrites the previous item. 1306 array: Many values can be stored and addressed by a numeric index. 1308 dictionary: The values stored are indexed by a key. Often this key 1309 is one of the values from the certificate of the peer sending the 1310 Store request. 1312 In order to protect stored data from tampering, by other nodes, each 1313 stored value is digitally signed by the node which created it. When 1314 a value is retrieved, the digital signature can be verified to detect 1315 tampering. 1317 4.1.1. Storage Permissions 1319 A major issue in peer-to-peer storage networks is minimizing the 1320 burden of becoming a peer, and in particular minimizing the amount of 1321 data which any peer is required to store for other nodes. RELOAD 1322 addresses this issue by only allowing any given node to store data at 1323 a small number of locations in the overlay, with those locations 1324 being determined by the node's certificate. When a peer uses a Store 1325 request to place data at a location authorized by its certificate, it 1326 signs that data with the private key that corresponds to its 1327 certificate. Then the peer responsible for storing the data is able 1328 to verify that the peer issuing the request is authorized to make 1329 that request. Each data kind defines the exact rules for determining 1330 what certificate is appropriate. 1332 The most natural rule is that a certificate authorizes a user to 1333 store data keyed with their user name X. This rule is used for all 1334 the kinds defined in this specification. Thus, only a user with a 1335 certificate for "alice@example.org" could write to that location in 1336 the overlay. However, other usages can define any rules they choose, 1337 including publicly writable values. 1339 The digital signature over the data serves two purposes. First, it 1340 allows the peer responsible for storing the data to verify that this 1341 Store is authorized. Second, it provides integrity for the data. 1342 The signature is saved along with the data value (or values) so that 1343 any reader can verify the integrity of the data. Of course, the 1344 responsible peer can "lose" the value but it cannot undetectably 1345 modify it. 1347 The size requirements of the data being stored in the overlay are 1348 variable. For instance, a SIP AOR and voicemail differ widely in the 1349 storage size. RELOAD leaves it to the Usage and overlay 1350 configuration to limit size imbalance of various kinds. 1352 4.1.2. Replication 1354 Replication in P2P overlays can be used to provide: 1356 persistence: if the responsible peer crashes and/or if the storing 1357 peer leaves the overlay 1358 security: to guard against DoS attacks by the responsible peer or 1359 routing attacks to that responsible peer 1360 load balancing: to balance the load of queries for popular 1361 resources. 1363 A variety of schemes are used in P2P overlays to achieve some of 1364 these goals. Common techniques include replicating on neighbors of 1365 the responsible peer, randomly locating replicas around the overlay, 1366 or replicating along the path to the responsible peer. 1368 The core RELOAD specification does not specify a particular 1369 replication strategy. Instead, the first level of replication 1370 strategies are determined by the overlay algorithm, which can base 1371 the replication strategy on its particular topology. For example, 1372 Chord places replicas on successor peers, which will take over 1373 responsibility should the responsible peer fail [Chord]. 1375 If additional replication is needed, for example if data persistence 1376 is particularly important for a particular usage, then that usage may 1377 specify additional replication, such as implementing random 1378 replications by inserting a different well known constant into the 1379 Resource Name used to store each replicated copy of the resource. 1380 Such replication strategies can be added independent of the 1381 underlying algorithm, and their usage can be determined based on the 1382 needs of the particular usage. 1384 4.2. Usages 1386 By itself, the distributed storage layer just provides infrastructure 1387 on which applications are built. In order to do anything useful, a 1388 usage must be defined. Each Usage needs to specify several things: 1390 o Registers Kind-ID code points for any kinds that the Usage 1391 defines. 1392 o Defines the data structure for each of the kinds. 1393 o Defines access control rules for each of the kinds. 1394 o Defines how the Resource Name is formed that is hashed to form the 1395 Resource-ID where each kind is stored. 1396 o Describes how values will be merged after a network partition. 1397 Unless otherwise specified, the default merging rule is to act as 1398 if all the values that need to be merged were stored and as if the 1399 order they were stored in corresponds to the stored time values 1400 associated with (and carried in) their values. Because the stored 1401 time values are those associated with the peer which did the 1402 writing, clock skew is generally not an issue. If two nodes are 1403 on different partitions, write to the same location, and have 1404 clock skew, this can create merge conflicts. However because 1405 RELOAD deliberately segregates storage so that data from different 1406 users and peers is stored in different locations, and a single 1407 peer will typically only be in a single network partition, this 1408 case will generally not arise. 1410 The kinds defined by a usage may also be applied to other usages. 1411 However, a need for different parameters, such as different size 1412 limits, would imply the need to create a new kind. 1414 4.3. Service Discovery 1416 RELOAD does not currently define a generic service discovery 1417 algorithm as part of the base protocol, although a simplistic TURN- 1418 specific discovery mechanism is provided. A variety of service 1419 discovery algorithms can be implemented as extensions to the base 1420 protocol, such as the service discovery algorithm ReDIR 1421 [opendht-sigcomm05] or [I-D.maenpaa-p2psip-service-discovery]. 1423 4.4. Application Connectivity 1425 There is no requirement that a RELOAD usage must use RELOAD's 1426 primitives for establishing its own communication if it already 1427 possesses its own means of establishing connections. For example, 1428 one could design a RELOAD-based resource discovery protocol which 1429 used HTTP to retrieve the actual data. 1431 For more common situations, however, it is the overlay itself - 1432 rather than an external authority such as DNS - which is used to 1433 establish a connection. RELOAD provides connectivity to applications 1434 using the AppAttach method. For example, if a P2PSIP node wishes to 1435 establish a SIP dialog with another P2PSIP node, it will use 1436 AppAttach to establish a direct connection with the other node. This 1437 new connection is separate from the peer protocol connection. It is 1438 a dedicated UDP or TCP flow used only for the SIP dialog. 1440 5. Overlay Management Protocol 1442 This section defines the basic protocols used to create, maintain, 1443 and use the RELOAD overlay network. We start by defining the basic 1444 concept of how message destinations are interpreted when routing 1445 messages. We then describe the symmetric recursive routing model, 1446 which is RELOAD's default routing algorithm. We then define the 1447 message structure and then finally define the messages used to join 1448 and maintain the overlay. 1450 5.1. Message Receipt and Forwarding 1452 When a peer receives a message, it first examines the overlay, 1453 version, and other header fields to determine whether the message is 1454 one it can process. If any of these are incorrect (e.g., the message 1455 is for an overlay in which the peer does not participate) it is an 1456 error. The peer SHOULD generate an appropriate error but local 1457 policy can override this and cause the messages to be silently 1458 dropped. 1460 Once the peer has determined that the message is correctly formatted, 1461 it examines the first entry on the destination list. There are three 1462 possible cases here: 1464 o The first entry on the destination list is an ID for which the 1465 peer is responsible. A peer is always responsible for the 1466 wildcard Node-ID. 1467 o The first entry on the destination list is an ID for which another 1468 peer is responsible. 1469 o The first entry on the destination list is a private ID that is 1470 being used for destination list compression. This is described 1471 later (note that private IDs can be distinguished from Node-IDs 1472 and Resource-IDs on the wire; see Section 5.3.2.2). 1474 These cases are handled as discussed below. 1476 5.1.1. Responsible ID 1478 If the first entry on the destination list is an ID for which the 1479 node is responsible, there are several sub-cases to consider. 1481 o If the entry is a Resource-ID, then it MUST be the only entry on 1482 the destination list. If there are other entries, the message 1483 MUST be silently dropped. Otherwise, the message is destined for 1484 this node and it passes it up to the upper layers. 1485 o If the entry is a Node-ID which equals this node's Node-ID, then 1486 the message is destined for this node. If this is the only entry 1487 on the destination list, the message is destined for this node and 1488 is passed up to the upper layers. Otherwise the entry is removed 1489 from the destination list and the message is passed to the Message 1490 Transport. If the message is a response and there is state for 1491 the transaction ID, the state is reinserted into the destination 1492 list before the message is further processed. 1493 o If the entry is the wildcard Node-ID, the message is destined for 1494 this node and it passes it up to the upper layers. 1495 o If the entry is a Node-ID which is not equal to this node, then 1496 the node MUST drop the message silently unless the Node-ID 1497 corresponds to a node which is directly connected to this node 1498 (i.e., a client). In that case, it MUST forward the message to 1499 the destination node as described in the next section. 1501 Note that this implies that in order to address a message to "the 1502 peer that controls region X", a sender sends to Resource-ID X, not 1503 Node-ID X. 1505 5.1.2. Other ID 1507 If neither of the other three cases applies, then the peer MUST 1508 forward the message towards the first entry on the destination list. 1510 This means that it MUST select one of the peers to which it is 1511 connected and which is likely to be responsible for the first entry 1512 on the destination list. If the first entry on the destination list 1513 is in the peer's connection table, then it SHOULD forward the message 1514 to that peer directly. Otherwise, the peer consults the routing 1515 table to forward the message. 1517 Any intermediate peer which forwards a RELOAD request MUST arrange 1518 that if it receives a response to that message the response can be 1519 routed back through the set of nodes through which the request 1520 passed. This may be arranged in one of two ways: 1522 o The peer MAY add an entry to the via list in the forwarding header 1523 that will enable it to determine the correct node. 1524 o The peer MAY keep per-transaction state which will allow it to 1525 determine the correct node. 1527 As an example of the first strategy, if node D receives a message 1528 from node C with via list (A, B), then D would forward to the next 1529 node (E) with via list (A, B, C). Now, if E wants to respond to the 1530 message, it reverses the via list to produce the destination list, 1531 resulting in (D, C, B, A). When D forwards the response to C, the 1532 destination list will contain (C, B, A). 1534 As an example of the second strategy, if node D receives a message 1535 from node C with transaction ID X and via list (A, B), it could store 1536 (X, C) in its state database and forward the message with the via 1537 list unchanged. When D receives the response, it consults its state 1538 database for transaction id X, determines that the request came from 1539 C, and forwards the response to C. 1541 Intermediate peers which modify the via list are not required to 1542 simply add entries. The only requirement is that the peer be able to 1543 reconstruct the correct destination list on the return route. RELOAD 1544 provides explicit support for this functionality in the form of 1545 private IDs, which can replace any number of via list entries. For 1546 instance, in the above example, Node D might send E a via list 1547 containing only the private ID (I). E would then use the destination 1548 list (D, I) to send its return message. When D processes this 1549 destination list, it would detect that I is a private ID, recover the 1550 via list (A, B, C), and reverse that to produce the correct 1551 destination list (C, B, A) before sending it to C. This feature is 1552 called List Compression. It MAY either be a compressed version of 1553 the original via list or an index into a state database containing 1554 the original via list. 1556 No matter what mechanism for storing via list state is used, if an 1557 intermediate peer exits the overlay, then on the return trip the 1558 message cannot be forwarded and will be dropped. The ordinary 1559 timeout and retransmission mechanisms provide stability over this 1560 type of failure. 1562 Note that if an intermediate peer retains per-transaction state 1563 instead of modifying the via list, it needs some mechanism for timing 1564 out that state, otherwise its state database will grow without bound. 1565 Whatever algorithm is used, unless a FORWARD_CRITICAL forwarding 1566 option or overlay configuration option explicitly indicates this 1567 state is not needed, the state MUST be maintained for at least the 1568 value of the overlay reliability timer (3 seconds) and MAY be kept 1569 longer. Future extension, such as [I-D.jiang-p2psip-relay], may 1570 define mechanisms for determining when this state does not need to be 1571 retained. 1573 None of the above mechanisms are required for responses, since there 1574 is no need to ensure that subsequent requests follow the same path. 1576 To be precise on the responsibility of the intermediate node, suppose 1577 that an intermediate node, A, receives a message from node B with via 1578 list X-Y-Z. Node A MUST implement an algorithm that ensures that A 1579 returns a response to this request to node B with the destination 1580 list B-Z-Y-X, provided that the node to which A forwards the request 1581 follows the same contract. Node A normally learns the Node-ID B is 1582 using via an Attach, but a node using a certificate with a single 1583 Node-ID MAY elect to not send an Attach (see Section 3.2.1 bullet 2). 1584 If a node with a certificate with multiple Node-IDs attempts to route 1585 a message other than a Ping or Attach through a node without 1586 performing an Attach, the receiving node MUST reject the request with 1587 an Error_Forbidden error. The node MUST implement support for 1588 returning responses to a Ping or Attach request made by a joining 1589 node Attaching to its responsible peer. 1591 5.1.3. Private ID 1593 If the first entry in the destination list is a private id (e.g., a 1594 compressed via list), the peer MUST replace that entry with the 1595 original via list that it replaced and then re-examine the 1596 destination list to determine which of the above cases now applies. 1598 5.2. Symmetric Recursive Routing 1600 This Section defines RELOAD's symmetric recursive routing algorithm, 1601 which is the default algorithm used by nodes to route messages 1602 through the overlay. All implementations MUST implement this routing 1603 algorithm. An overlay may be configured to use alternative routing 1604 algorithms, and alternative routing algorithms may be selected on a 1605 per-message basis. 1607 5.2.1. Request Origination 1609 In order to originate a message to a given Node-ID or Resource-ID, a 1610 node constructs an appropriate destination list. The simplest such 1611 destination list is a single entry containing the Node-ID or 1612 Resource-ID. The resulting message will use the normal overlay 1613 routing mechanisms to forward the message to that destination. The 1614 node can also construct a more complicated destination list for 1615 source routing. 1617 Once the message is constructed, the node sends the message to some 1618 adjacent peer. If the first entry on the destination list is 1619 directly connected, then the message MUST be routed down that 1620 connection. Otherwise, the topology plugin MUST be consulted to 1621 determine the appropriate next hop. 1623 Parallel searches for the resource are a common solution to improve 1624 reliability in the face of churn or of subversive peers. Parallel 1625 searches for usage-specified replicas are managed by the usage layer. 1626 However, a single request can also be routed through multiple 1627 adjacent peers, even when known to be sub-optimal, to improve 1628 reliability [vulnerabilities-acsac04]. Such parallel searches MAY be 1629 specified by the topology plugin. 1631 Because messages may be lost in transit through the overlay, RELOAD 1632 incorporates an end-to-end reliability mechanism. When an 1633 originating node transmits a request it MUST set a 3 second timer. 1634 If a response has not been received when the timer fires, the request 1635 is retransmitted with the same transaction identifier. The request 1636 MAY be retransmitted up to 4 times (for a total of 5 messages). 1637 After the timer for the fifth transmission fires, the message SHALL 1638 be considered to have failed. Note that this retransmission 1639 procedure is not followed by intermediate nodes. They follow the 1640 hop-by-hop reliability procedure described in Section 5.6.3. 1642 The above algorithm can result in multiple requests being delivered 1643 to a node. Receiving nodes MUST generate semantically equivalent 1644 responses to retransmissions of the same request (this can be 1645 determined by transaction id) if the request is received within the 1646 maximum request lifetime (15 seconds). For some requests (e.g., 1647 Fetch) this can be accomplished merely by processing the request 1648 again. For other requests, (e.g., Store) it may be necessary to 1649 maintain state for the duration of the request lifetime. 1651 5.2.2. Response Origination 1653 When a peer sends a response to a request using this routing 1654 algorithm, it MUST construct the destination list by reversing the 1655 order of the entries on the via list. This has the result that the 1656 response traverses the same peers as the request traversed, except in 1657 reverse order (symmetric routing). 1659 5.3. Message Structure 1661 RELOAD is a message-oriented request/response protocol. The messages 1662 are encoded using binary fields. All integers are represented in 1663 network byte order. The general philosophy behind the design was to 1664 use Type, Length, Value fields to allow for extensibility. However, 1665 for the parts of a structure that were required in all messages, we 1666 just define these in a fixed position, as adding a type and length 1667 for them is unnecessary and would simply increase bandwidth and 1668 introduces new potential for interoperability issues. 1670 Each message has three parts, concatenated as shown below: 1672 +-------------------------+ 1673 | Forwarding Header | 1674 +-------------------------+ 1675 | Message Contents | 1676 +-------------------------+ 1677 | Security Block | 1678 +-------------------------+ 1680 The contents of these parts are as follows: 1682 Forwarding Header: Each message has a generic header which is used 1683 to forward the message between peers and to its final destination. 1684 This header is the only information that an intermediate peer 1685 (i.e., one that is not the target of a message) needs to examine. 1687 Message Contents: The message being delivered between the peers. 1688 From the perspective of the forwarding layer, the contents are 1689 opaque, however, they are interpreted by the higher layers. 1691 Security Block: A security block containing certificates and a 1692 digital signature over the "Message Contents" section. Note that 1693 this signature can be computed without parsing the message 1694 contents. All messages MUST be signed by their originator. 1696 The following sections describe the format of each part of the 1697 message. 1699 5.3.1. Presentation Language 1701 The structures defined in this document are defined using a C-like 1702 syntax based on the presentation language used to define TLS. 1703 [RFC5246] Advantages of this style include: 1705 o It familiar enough looking that most readers can grasp it quickly. 1706 o The ability to define nested structures allows a separation 1707 between high-level and low-level message structures. 1708 o It has a straightforward wire encoding that allows quick 1709 implementation, but the structures can be comprehended without 1710 knowing the encoding. 1711 o The ability to mechanically compile encoders and decoders. 1713 Several idiosyncrasies of this language are worth noting. 1715 o All lengths are denoted in bytes, not objects. 1716 o Variable length values are denoted like arrays with angle 1717 brackets. 1718 o "select" is used to indicate variant structures. 1720 For instance, "uint16 array<0..2^8-2>;" represents up to 254 bytes 1721 but only up to 127 values of two bytes (16 bits) each. 1723 5.3.1.1. Common Definitions 1725 The following definitions are used throughout RELOAD and so are 1726 defined here. They also provide a convenient introduction to how to 1727 read the presentation language. 1729 An enum represents an enumerated type. The values associated with 1730 each possibility are represented in parentheses and the maximum value 1731 is represented as a nameless value, for purposes of describing the 1732 width of the containing integral type. For instance, Boolean 1733 represents a true or false: 1735 enum { false (0), true(1), (255)} Boolean; 1737 A boolean value is either a 1 or a 0. The max value of 255 indicates 1738 this is represented as a single byte on the wire. 1740 The NodeId, shown below, represents a single Node-ID. 1742 typedef opaque NodeId[NodeIdLength]; 1744 A NodeId is a fixed-length structure represented as a series of 1745 bytes, with the most significant byte first. The length is set on a 1746 per-overlay basis within the range of 16-20 bytes (128 to 160 bits). 1747 (See Section 10.1 for how NodeIdLength is set.) Note: the use of 1748 "typedef" here is an extension to the TLS language, but its meaning 1749 should be relatively obvious. Note the [ size ] syntax defines a 1750 fixed length element that does not include the length of the element 1751 in the on the wire encoding. 1753 A ResourceId, shown below, represents a single Resource-ID. 1755 typedef opaque ResourceId<0..2^8-1>; 1757 Like a NodeId, a ResourceId is an opaque string of bytes, but unlike 1758 NodeIds, ResourceIds are variable length, up to 254 bytes (2040 bits) 1759 in length. On the wire, each ResourceId is preceded by a single 1760 length byte (allowing lengths up to 255). Thus, the 3-byte value 1761 "FOO" would be encoded as: 03 46 4f 4f. Note the < range > syntax 1762 defines a variable length element that does include the length of the 1763 element in the on the wire encoding. The number of bytes to encode 1764 the length on the wire is derived by range; i.e., it is the minimum 1765 number of bytes which can encode the largest range value. 1767 A more complicated example is IpAddressPort, which represents a 1768 network address and can be used to carry either an IPv6 or IPv4 1769 address: 1771 enum {reservedAddr(0), ipv4_address (1), ipv6_address (2), 1772 (255)} AddressType; 1774 struct { 1775 uint32 addr; 1776 uint16 port; 1777 } IPv4AddrPort; 1779 struct { 1780 uint128 addr; 1781 uint16 port; 1782 } IPv6AddrPort; 1784 struct { 1785 AddressType type; 1786 uint8 length; 1788 select (type) { 1789 case ipv4_address: 1790 IPv4AddrPort v4addr_port; 1792 case ipv6_address: 1793 IPv6AddrPort v6addr_port; 1795 /* This structure can be extended */ 1796 }; 1797 } IpAddressPort; 1799 The first two fields in the structure are the same no matter what 1800 kind of address is being represented: 1802 type: the type of address (v4 or v6). 1803 length: the length of the rest of the structure. 1805 By having the type and the length appear at the beginning of the 1806 structure regardless of the kind of address being represented, an 1807 implementation which does not understand new address type X can still 1808 parse the IpAddressPort field and then discard it if it is not 1809 needed. 1811 The rest of the IpAddressPort structure is either an IPv4AddrPort or 1812 an IPv6AddrPort. Both of these simply consist of an address 1813 represented as an integer and a 16-bit port. As an example, here is 1814 the wire representation of the IPv4 address "192.0.2.1" with port 1815 "6100". 1817 01 ; type = IPv4 1818 06 ; length = 6 1819 c0 00 02 01 ; address = 192.0.2.1 1820 17 d4 ; port = 6100 1822 Unless a given structure that uses a select explicitly allows for 1823 unknown types in the select, any unknown type SHOULD be treated as an 1824 parsing error and the whole message discarded with no response. 1826 5.3.2. Forwarding Header 1828 The forwarding header is defined as a ForwardingHeader structure, as 1829 shown below. 1831 struct { 1832 uint32 relo_token; 1833 uint32 overlay; 1834 uint16 configuration_sequence; 1835 uint8 version; 1836 uint8 ttl; 1837 uint32 fragment; 1838 uint32 length; 1839 uint64 transaction_id; 1840 uint32 max_response_length; 1841 uint16 via_list_length; 1842 uint16 destination_list_length; 1843 uint16 options_length; 1844 Destination via_list[via_list_length]; 1845 Destination destination_list 1846 [destination_list_length]; 1847 ForwardingOptions options[options_length]; 1848 } ForwardingHeader; 1850 The contents of the structure are: 1852 relo_token: The first four bytes identify this message as a RELOAD 1853 message. This field MUST contain the value 0xd2454c4f (the string 1854 'RELO' with the high bit of the first byte set.). 1856 overlay: The 32 bit checksum/hash of the overlay being used. The 1857 variable length string representing the overlay name is hashed 1858 with SHA-1 [RFC3174] and the low order 32 bits are used. The 1859 purpose of this field is to allow nodes to participate in multiple 1860 overlays and to detect accidental misconfiguration. This is not a 1861 security critical function. 1863 configuration_sequence: The sequence number of the configuration 1864 file. 1866 version: The version of the RELOAD protocol being used. This is a 1867 fixed point integer between 0.1 and 25.4. This document describes 1868 version 0.1, with a value of 0x01. [[ Note to RFC Editor: Please 1869 update this to version 1.0 with value of 0x0a and remove this 1870 note. ]] 1872 ttl: An 8 bit field indicating the number of iterations, or hops, a 1873 message can experience before it is discarded. The TTL value MUST 1874 be decremented by one at every hop along the route the message 1875 traverses. If the TTL is 0, the message MUST NOT be propagated 1876 further and MUST be discarded, and a "Error_TTL_Exceeded" error 1877 should be generated. The initial value of the TTL SHOULD be 100 1878 unless defined otherwise by the overlay configuration. 1880 fragment: This field is used to handle fragmentation. The high 1881 order two bits are used to indicate the fragmentation status: If 1882 the high bit (0x80000000) is set, it indicates that the message is 1883 a fragment. If the next bit (0x40000000) is set, it indicates 1884 that this is the last fragment. The next six bits (0x20000000 to 1885 0x01000000) are reserved and SHOULD be set to zero. The remainder 1886 of the field is used to indicate the fragment offset; see 1887 Section 5.7 1889 length: The count in bytes of the size of the message, including the 1890 header. 1892 transaction_id: A unique 64 bit number that identifies this 1893 transaction and also allows receivers to disambiguate transactions 1894 which are otherwise identical. In order to provide a high 1895 probability that transaction IDs are unique, they MUST be randomly 1896 generated. Responses use the same Transaction ID as the request 1897 they correspond to. Transaction IDs are also used for fragment 1898 reassembly. 1900 max_response_length: The maximum size in bytes of a response. Used 1901 by requesting nodes to avoid receiving (unexpected) very large 1902 responses. If this value is non-zero, responding peers MUST check 1903 that any response would not exceed it and if so generate an 1904 Error_Response_Too_Large value. This value SHOULD be set to zero 1905 for responses. 1907 via_list_length: The length of the via list in bytes. Note that in 1908 this field and the following two length fields we depart from the 1909 usual variable-length convention of having the length immediately 1910 precede the value in order to make it easier for hardware decoding 1911 engines to quickly determine the length of the header. 1913 destination_list_length: The length of the destination list in 1914 bytes. 1916 options_length: The length of the header options in bytes. 1918 via_list: The via_list contains the sequence of destinations through 1919 which the message has passed. The via_list starts out empty and 1920 grows as the message traverses each peer. 1922 destination_list: The destination_list contains a sequence of 1923 destinations which the message should pass through. The 1924 destination list is constructed by the message originator. The 1925 first element in the destination list is where the message goes 1926 next. The list shrinks as the message traverses each listed peer. 1928 options: Contains a series of ForwardingOptions entries. See 1929 Section 5.3.2.3. 1931 5.3.2.1. Processing Configuration Sequence Numbers 1933 In order to be part of the overlay, a node MUST have a copy of the 1934 overlay configuration document. In order to allow for configuration 1935 document changes, each version of the configuration document has a 1936 sequence number which is monotonically increasing mod 65536. Because 1937 the sequence number may in principle wrap, greater than or less than 1938 are interpreted by modulo arithmetic as in TCP. 1940 When a destination node receives a request, it MUST check that the 1941 configuration_sequence field is equal to its own configuration 1942 sequence number. If they do not match, it MUST generate an error, 1943 either Error_Config_Too_Old or Error_Config_Too_New. In addition, if 1944 the configuration file in the request is too old, it MUST generate a 1945 ConfigUpdate message to update the requesting node. This allows new 1946 configuration documents to propagate quickly throughout the system. 1947 The one exception to this rule is that if the configuration_sequence 1948 field is equal to 0xffff, and the message type is ConfigUpdate, then 1949 the message MUST be accepted regardless of the receiving node's 1950 configuration sequence number. Since 65535 is a special value, peers 1951 sending a new configuration when the configuration sequence is 1952 currently 65534 MUST set the configuration sequence number to 0 when 1953 they send out a new configuration. 1955 5.3.2.2. Destination and Via Lists 1957 The destination list and via lists are sequences of Destination 1958 values: 1960 enum {reserved(0), node(1), resource(2), compressed(3), 1961 /* 128-255 not allowed */ (255) } 1962 DestinationType; 1964 select (destination_type) { 1965 case node: 1966 NodeId node_id; 1968 case resource: 1969 ResourceId resource_id; 1971 case compressed: 1972 opaque compressed_id<0..2^8-1>; 1974 /* This structure may be extended with new types */ 1975 } DestinationData; 1977 struct { 1978 DestinationType type; 1979 uint8 length; 1980 DestinationData destination_data; 1981 } Destination; 1983 struct { 1984 uint16 compressed_id; /* top bit MUST be 1 */ 1985 } Destination; 1987 If a destination structure has its first bit set to 1, then it is a 1988 16 bit integer. If the first bit is not set, then it is a structure 1989 starting with DestinationType. If it is a 16 bit integer, it is 1990 treated as if it were a full structure with a DestinationType of 1991 compressed and a compressed_id that was 2 bytes long with the value 1992 of the 16 bit integer. When the destination structure is not a 16 1993 bit integer, it is the TLV structure with the following contents: 1995 type 1996 The type of the DestinationData Payload Data Unit (PDU). This may 1997 be one of "node", "resource", or "compressed". 1999 length 2000 The length of the destination_data. 2002 destination_data 2003 The destination value itself, which is an encoded DestinationData 2004 structure, depending on the value of "type". 2006 Note: This structure encodes a type, length, value. The length 2007 field specifies the length of the DestinationData values, which 2008 allows the addition of new DestinationTypes. This allows an 2009 implementation which does not understand a given DestinationType 2010 to skip over it. 2012 A DestinationData can be one of three types: 2014 node 2015 A Node-ID. 2017 compressed 2018 A compressed list of Node-IDs and/or resources. Because this 2019 value was compressed by one of the peers, it is only meaningful to 2020 that peer and cannot be decoded by other peers. Thus, it is 2021 represented as an opaque string. 2023 resource 2024 The Resource-ID of the resource which is desired. This type MUST 2025 only appear in the final location of a destination list and MUST 2026 NOT appear in a via list. It is meaningless to try to route 2027 through a resource. 2029 One possible encoding of the 16 bit integer version as an opaque 2030 identifier is to encode an index into a connection table. To avoid 2031 misrouting responses in the event a response is delayed and the 2032 connection table entry has changed, the identifier SHOULD be split 2033 between an index and a generation counter for that index. At 2034 startup, the generation counters should be initialized to random 2035 values. An implementation could use 12 bits for the connection table 2036 index and 3 bits for the generation counter. (Note that this does 2037 not suggest a 4096 entry connection table for every node, only the 2038 ability to encode for a larger connection table.) When a connection 2039 table slot is used for a new connection, the generation counter is 2040 incremented (with wrapping). Connection table slots are used on a 2041 rotating basis to maximize the time interval between uses of the same 2042 slot for different connections. When routing a message to an entry 2043 in the destination list encoding a connection table entry, the node 2044 confirms that the generation counter matches the current generation 2045 counter of that index before forwarding the message. If it does not 2046 match, the message is silently dropped. 2048 5.3.2.3. Forwarding Options 2050 The Forwarding header can be extended with forwarding header options, 2051 which are a series of ForwardingOptions structures: 2053 enum { reservedForwarding(0), (255) } 2054 ForwardingOptionsType; 2056 struct { 2057 ForwardingOptionsType type; 2058 uint8 flags; 2059 uint16 length; 2060 select (type) { 2061 /* This type may be extended */ 2062 } option; 2063 } ForwardingOption; 2065 Each ForwardingOption consists of the following values: 2067 type 2068 The type of the option. This structure allows for unknown options 2069 types. 2071 length 2072 The length of the rest of the structure. 2074 flags 2075 Three flags are defined FORWARD_CRITICAL(0x01), 2076 DESTINATION_CRITICAL(0x02), and RESPONSE_COPY(0x04). These flags 2077 MUST NOT be set in a response. If the FORWARD_CRITICAL flag is 2078 set, any node that would forward the message but does not 2079 understand this options MUST reject the request with an 2080 Error_Unsupported_Forwarding_Option error response. If the 2081 DESTINATION_CRITICAL flag is set, any node that generates a 2082 response to the message but does not understand the forwarding 2083 option MUST reject the request with an 2084 Error_Unsupported_Forwarding_Option error response. If the 2085 RESPONSE_COPY flag is set, any node generating a response MUST 2086 copy the option from the request to the response except that the 2087 RESPONSE_COPY, FORWARD_CRITICAL and DESTINATION_CRITICAL flags 2088 must be cleared. 2090 option 2091 The option value. 2093 5.3.3. Message Contents Format 2095 The second major part of a RELOAD message is the contents part, which 2096 is defined by MessageContents: 2098 enum { reservedMessagesExtension(0), (2^16-1) } MessageExtensionType; 2100 struct { 2101 MessageExtensionType type; 2102 Boolean critical; 2103 opaque extension_contents<0..2^32-1>; 2104 } MessageExtension; 2106 struct { 2107 uint16 message_code; 2108 opaque message_body<0..2^32-1>; 2109 MessageExtensions extensions<0..2^32-1>; 2110 } MessageContents; 2112 The contents of this structure are as follows: 2114 message_code 2115 This indicates the message that is being sent. The code space is 2116 broken up as follows. 2118 0 Reserved 2120 1 .. 0x7fff Requests and responses. These code points are always 2121 paired, with requests being odd and the corresponding response 2122 being the request code plus 1. Thus, "probe_request" (the 2123 Probe request) has value 1 and "probe_answer" (the Probe 2124 response) has value 2 2126 0xffff Error 2127 The message codes are defined in Section 13.8 2128 message_body 2129 The message body itself, represented as a variable-length string 2130 of bytes. The bytes themselves are dependent on the code value. 2131 See the sections describing the various RELOAD methods (Join, 2132 Update, Attach, Store, Fetch, etc.) for the definitions of the 2133 payload contents. 2135 extensions 2136 Extensions to the message. Currently no extensions are defined, 2137 but new extensions can be defined by the process described in 2138 Section 13.14. 2140 All extensions have the following form: 2142 type 2143 The extension type. 2145 critical 2146 Whether this extension must be understood in order to process the 2147 message. If critical = True and the recipient does not understand 2148 the message, it MUST generate an Error_Unknown_Extension error. 2149 If critical = False, the recipient MAY choose to process the 2150 message even if it does not understand the extension. 2152 extension_contents 2153 The contents of the extension (extension-dependent). 2155 5.3.3.1. Response Codes and Response Errors 2157 A peer processing a request returns its status in the message_code 2158 field. If the request was a success, then the message code is the 2159 response code that matches the request (i.e., the next code up). The 2160 response payload is then as defined in the request/response 2161 descriptions. 2163 If the request has failed, then the message code is set to 0xffff 2164 (error) and the payload MUST be an error_response PDU, as shown 2165 below. 2167 When the message code is 0xffff, the payload MUST be an 2168 ErrorResponse. 2170 public struct { 2171 uint16 error_code; 2172 opaque error_info<0..2^16-1>; 2173 } ErrorResponse; 2175 The contents of this structure are as follows: 2177 error_code 2178 A numeric error code indicating the error that occurred. 2180 error_info 2181 An optional arbitrary byte string. Unless otherwise specified, 2182 this will be a UTF-8 text string providing further information 2183 about what went wrong. 2185 The following error code values are defined. The numeric values for 2186 these are defined in Section 13.9. 2188 Error_Forbidden: The requesting node does not have permission to 2189 make this request. 2191 Error_Not_Found: The resource or peer cannot be found or does not 2192 exist. 2194 Error_Request_Timeout: A response to the request has not been 2195 received in a suitable amount of time. The requesting node MAY 2196 resend the request at a later time. 2198 Error_Data_Too_Old: A store cannot be completed because the 2199 storage_time precedes the existing value. 2201 Error_Data_Too_Old: A store cannot be completed because the 2202 storage_time precedes the existing value. 2204 Error_Data_Too_Large: A store cannot be completed because the 2205 requested object exceeds the size limits for that kind. 2207 Error_Generation_Counter_Too_Low: A store cannot be completed 2208 because the generation counter precedes the existing value. 2210 Error_Incompatible_with_Overlay: A peer receiving the request is 2211 using a different overlay, overlay algorithm, or hash algorithm. 2213 Error_Unsupported_Forwarding_Option: A peer receiving the request 2214 with a forwarding options flagged as critical but the peer does 2215 not support this option. See section Section 5.3.2.3. 2217 Error_TTL_Exceeded: A peer receiving the request where the TTL got 2218 decremented to zero. See section Section 5.3.2. 2220 Error_Message_Too_Large: A peer receiving the request that was too 2221 large. See section Section 5.6. 2223 Error_Response_Too_Large: A peer would have generated a response 2224 that is too large per the max_response_length field. 2226 Error_Config_Too_Old: A destination peer received a request with a 2227 configuration sequence that's too old. See Section 5.3.2.1. 2229 Error_Config_Too_New: A destination node received a request with a 2230 configuration sequence that's too new. See Section 5.3.2.1. 2232 Error_Unknown_Kind: A destination node received a request with an 2233 unknown kind-id. See Section 6.4.1.2. 2235 Error_In_Progress: An Attach is already in progress to this peer. 2236 See Section 5.5.1.2. 2238 Error_Unknown_Extension: A destination node received a request with 2239 an unknown extension. 2241 5.3.4. Security Block 2243 The third part of a RELOAD message is the security block. The 2244 security block is represented by a SecurityBlock structure: 2246 struct { 2247 CertificateType type; 2248 opaque certificate<0..2^16-1>; 2249 } GenericCertificate; 2251 struct { 2252 GenericCertificate certificates<0..2^16-1>; 2253 Signature signature; 2254 } SecurityBlock; 2256 The contents of this structure are: 2258 certificates 2259 A bucket of certificates. 2261 signature 2262 A signature over the message contents. 2264 The certificates bucket SHOULD contain all the certificates necessary 2265 to verify every signature in both the message and the internal 2266 message objects. This is the only location in the message which 2267 contains certificates, thus allowing for only a single copy of each 2268 certificate to be sent. In systems which have some alternate 2269 certificate distribution mechanism, some certificates MAY be omitted. 2270 However, implementors should note that this creates the possibility 2271 that messages may not be immediately verifiable because certificates 2272 must first be retrieved. 2274 Each certificate is represented by a GenericCertificate structure, 2275 which has the following contents: 2277 type 2278 The type of the certificate, as defined in [RFC6091]. Only the 2279 use of X.509 certificates is defined in this draft. 2281 certificate 2282 The encoded version of the certificate. For X.509 certificates, 2283 it is the DER form. 2285 The signature is computed over the payload and parts of the 2286 forwarding header. The payload, in case of a Store, may contain an 2287 additional signature computed over a StoreReq structure. All 2288 signatures are formatted using the Signature element. This element 2289 is also used in other contexts where signatures are needed. The 2290 input structure to the signature computation varies depending on the 2291 data element being signed. 2293 enum { reservedSignerIdentity(0), 2294 cert_hash(1), cert_hash_node_id(2), 2295 none(3) 2296 (255)} SignerIdentityType; 2298 struct { 2299 select (identity_type) { 2301 case cert_hash; 2302 HashAlgorithm hash_alg; // From TLS 2303 opaque certificate_hash<0..2^8-1>; 2305 case cert_hash_node_id: 2306 HashAlgorithm hash_alg; // From TLS 2307 opaque certificate_node_id_hash<0..2^8-1>; 2309 case none: 2310 /* empty */ 2311 /* This structure may be extended with new types if necessary*/ 2312 }; 2313 } SignerIdentityValue; 2315 struct { 2316 SignerIdentityType identity_type; 2317 uint16 length; 2318 SignerIdentityValue identity[SignerIdentity.length]; 2319 } SignerIdentity; 2321 struct { 2322 SignatureAndHashAlgorithm algorithm; // From TLS 2323 SignerIdentity identity; 2324 opaque signature_value<0..2^16-1>; 2325 } Signature; 2327 The signature construct contains the following values: 2329 algorithm 2330 The signature algorithm in use. The algorithm definitions are 2331 found in the IANA TLS SignatureAlgorithm Registry and 2332 HashAlgorithm registries. All implementations MUST support 2333 RSASSA-PKCS1-v1_5 [RFC3447] signatures with SHA-256 hashes. 2335 identity 2336 The identity used to form the signature. 2338 signature_value 2339 The value of the signature. 2341 There are two permitted identity formats, one for a certificate with 2342 only one node-id and one for a certificate with multiple node-ids. 2343 In the first case, the cert_hash type SHOULD be used. The hash_alg 2344 field is used to indicate the algorithm used to produce the hash. 2345 The certificate_hash contains the hash of the certificate object 2346 (i.e., the DER-encoded certificate). 2348 In the second case, the cert_hash_node_id type MUST be used. The 2349 hash_alg is as in cert_hash but the cert_hash_node_id is computed 2350 over the NodeId used to sign concatenated with the certificate. 2351 I.e., H(NodeID || certificate). The NodeId is represented without 2352 any framing or length fields, as simple raw bytes. This is safe 2353 because NodeIds are fixed-length for a given overlay. 2355 For signatures over messages the input to the signature is computed 2356 over: 2358 overlay || transaction_id || MessageContents || SignerIdentity 2360 where overlay and transaction_id come from the forwarding header and 2361 || indicates concatenation. 2363 The input to signatures over data values is different, and is 2364 described in Section 6.1. 2366 All RELOAD messages MUST be signed. Upon receipt, the receiving node 2367 MUST verify the signature and the authorizing certificate. This 2368 check provides a minimal level of assurance that the sending node is 2369 a valid part of the overlay as well as cryptographic authentication 2370 of the sending node. In addition, responses MUST be checked as 2371 follows: 2373 1. The response to a message sent to a specific Node-ID MUST have 2374 been sent by that Node-ID. 2375 2. The response to a message sent to a Resource-Id MUST have been 2376 sent by a Node-ID which is as close to or closer to the target 2377 Resource-Id than any node in the requesting node's neighbor 2378 table. 2380 The second condition serves as a primitive check for responses from 2381 wildly wrong nodes but is not a complete check. Note that in periods 2382 of churn, it is possible for the requesting node to obtain a closer 2383 neighbor while the request is outstanding. This will cause the 2384 response to be rejected and the request to be retransmitted. 2386 In addition, some methods (especially Store) have additional 2387 authentication requirements, which are described in the sections 2388 covering those methods. 2390 5.4. Overlay Topology 2392 As discussed in previous sections, RELOAD does not itself implement 2393 any overlay topology. Rather, it relies on Topology Plugins, which 2394 allow a variety of overlay algorithms to be used while maintaining 2395 the same RELOAD core. This section describes the requirements for 2396 new topology plugins and the methods that RELOAD provides for overlay 2397 topology maintenance. 2399 5.4.1. Topology Plugin Requirements 2401 When specifying a new overlay algorithm, at least the following need 2402 to be described: 2404 o Joining procedures, including the contents of the Join message. 2405 o Stabilization procedures, including the contents of the Update 2406 message, the frequency of topology probes and keepalives, and the 2407 mechanism used to detect when peers have disconnected. 2408 o Exit procedures, including the contents of the Leave message. 2409 o The length of the Resource-IDs. For DHTs, the hash algorithm to 2410 compute the hash of an identifier. 2411 o The procedures that peers use to route messages. 2412 o The replication strategy used to ensure data redundancy. 2414 All overlay algorithms MUST specify maintenance procedures that send 2415 Updates to clients and peers that have established connections to the 2416 peer responsible for a particular ID when the responsibility for that 2417 ID changes. Because tracking this information is difficult, overlay 2418 algorithms MAY simply specify that an Update is sent to all members 2419 of the Connection Table whenever the range of IDs for which the peer 2420 is responsible changes. 2422 5.4.2. Methods and types for use by topology plugins 2424 This section describes the methods that topology plugins use to join, 2425 leave, and maintain the overlay. 2427 5.4.2.1. Join 2429 A new peer (but one that already has credentials) uses the JoinReq 2430 message to join the overlay. The JoinReq is sent to the responsible 2431 peer depending on the routing mechanism described in the topology 2432 plugin. This notifies the responsible peer that the new peer is 2433 taking over some of the overlay and it needs to synchronize its 2434 state. 2436 struct { 2437 NodeId joining_peer_id; 2438 opaque overlay_specific_data<0..2^16-1>; 2439 } JoinReq; 2441 The minimal JoinReq contains only the Node-ID which the sending peer 2442 wishes to assume. Overlay algorithms MAY specify other data to 2443 appear in this request. Receivers of the JoinReq MUST verify that 2444 the joining_peer_id field matches the Node-ID used to sign the 2445 message and if not MUST reject the message with an Error_Forbidden 2446 error. 2448 If the request succeeds, the responding peer responds with a JoinAns 2449 message, as defined below: 2451 struct { 2452 opaque overlay_specific_data<0..2^16-1>; 2453 } JoinAns; 2455 If the request succeeds, the responding peer MUST follow up by 2456 executing the right sequence of Stores and Updates to transfer the 2457 appropriate section of the overlay space to the joining peer. In 2458 addition, overlay algorithms MAY define data to appear in the 2459 response payload that provides additional info. 2461 In general, nodes which cannot form connections SHOULD report an 2462 error. However, implementations MUST provide some mechanism whereby 2463 nodes can determine that they are potentially the first node and take 2464 responsibility for the overlay. This specification does not mandate 2465 any particular mechanism, but a configuration flag or setting seems 2466 appropriate. 2468 5.4.2.2. Leave 2470 The LeaveReq message is used to indicate that a node is exiting the 2471 overlay. A node SHOULD send this message to each peer with which it 2472 is directly connected prior to exiting the overlay. 2474 struct { 2475 NodeId leaving_peer_id; 2476 opaque overlay_specific_data<0..2^16-1>; 2477 } LeaveReq; 2479 LeaveReq contains only the Node-ID of the leaving peer. Overlay 2480 algorithms MAY specify other data to appear in this request. 2481 Receivers of the LeaveReq MUST verify that the leaving_peer_id field 2482 matches the Node-ID used to sign the message and if not MUST reject 2483 the message with an Error_Forbidden error. 2485 Upon receiving a Leave request, a peer MUST update its own routing 2486 table, and send the appropriate Store/Update sequences to re- 2487 stabilize the overlay. 2489 5.4.2.3. Update 2491 Update is the primary overlay-specific maintenance message. It is 2492 used by the sender to notify the recipient of the sender's view of 2493 the current state of the overlay (its routing state), and it is up to 2494 the recipient to take whatever actions are appropriate to deal with 2495 the state change. In general, peers send Update messages to all 2496 their adjacencies whenever they detect a topology shift. 2498 When a peer receives an Attach request with the send_update flag set 2499 to "true" (Section 5.4.2.4.1, it MUST send an Update message back to 2500 the sender of the Attach request after the completion of the 2501 corresponding ICE check and TLS connection. Note that the sender of 2502 a such Attach request may not have joined the overlay yet. 2504 When a peer detects through an Update that it is no longer 2505 responsible for any data value it is storing, it MUST attempt to 2506 Store a copy to the correct node unless it knows the newly 2507 responsible node already has a copy of the data. This prevents data 2508 loss during large-scale topology shifts such as the merging of 2509 partitioned overlays. 2511 The contents of the UpdateReq message are completely overlay- 2512 specific. The UpdateAns response is expected to be either success or 2513 an error. 2515 5.4.2.4. RouteQuery 2517 The RouteQuery request allows the sender to ask a peer where they 2518 would route a message directed to a given destination. In other 2519 words, a RouteQuery for a destination X requests the Node-ID for the 2520 node that the receiving peer would next route to in order to get to 2521 X. A RouteQuery can also request that the receiving peer initiate an 2522 Update request to transfer the receiving peer's routing table. 2524 One important use of the RouteQuery request is to support iterative 2525 routing. The sender selects one of the peers in its routing table 2526 and sends it a RouteQuery message with the destination_object set to 2527 the Node-ID or Resource-ID it wishes to route to. The receiving peer 2528 responds with information about the peers to which the request would 2529 be routed. The sending peer MAY then use the Attach method to attach 2530 to that peer(s), and repeat the RouteQuery. Eventually, the sender 2531 gets a response from a peer that is closest to the identifier in the 2532 destination_object as determined by the topology plugin. At that 2533 point, the sender can send messages directly to that peer. 2535 5.4.2.4.1. Request Definition 2537 A RouteQueryReq message indicates the peer or resource that the 2538 requesting node is interested in. It also contains a "send_update" 2539 option allowing the requesting node to request a full copy of the 2540 other peer's routing table. 2542 struct { 2543 Boolean send_update; 2544 Destination destination; 2545 opaque overlay_specific_data<0..2^16-1>; 2546 } RouteQueryReq; 2548 The contents of the RouteQueryReq message are as follows: 2550 send_update 2551 A single byte. This may be set to "true" to indicate that the 2552 requester wishes the responder to initiate an Update request 2553 immediately. Otherwise, this value MUST be set to "false". 2555 destination 2556 The destination which the requester is interested in. This may be 2557 any valid destination object, including a Node-ID, compressed ids, 2558 or Resource-ID. 2560 overlay_specific_data 2561 Other data as appropriate for the overlay. 2563 5.4.2.4.2. Response Definition 2565 A response to a successful RouteQueryReq request is a RouteQueryAns 2566 message. This is completely overlay specific. 2568 5.4.2.5. Probe 2570 Probe provides primitive "exploration" services: it allows node to 2571 determine which resources another node is responsible for; and it 2572 allows some discovery services using multicast, anycast, or 2573 broadcast. A probe can be addressed to a specific Node-ID, or the 2574 peer controlling a given location (by using a Resource-ID). In 2575 either case, the target Node-IDs respond with a simple response 2576 containing some status information. 2578 5.4.2.5.1. Request Definition 2580 The ProbeReq message contains a list (potentially empty) of the 2581 pieces of status information that the requester would like the 2582 responder to provide. 2584 enum { reservedProbeInformation(0), responsible_set(1), 2585 num_resources(2), uptime(3), (255)} 2586 ProbeInformationType; 2588 struct { 2589 ProbeInformationType requested_info<0..2^8-1>; 2590 } ProbeReq 2592 The currently defined values for ProbeInformation are: 2594 responsible_set 2595 indicates that the peer should Respond with the fraction of the 2596 overlay for which the responding peer is responsible. 2598 num_resources 2599 indicates that the peer should Respond with the number of 2600 resources currently being stored by the peer. 2602 uptime 2603 indicates that the peer should Respond with how long the peer has 2604 been up in seconds. 2606 5.4.2.5.2. Response Definition 2608 A successful ProbeAns response contains the information elements 2609 requested by the peer. 2611 struct { 2612 select (type) { 2613 case responsible_set: 2614 uint32 responsible_ppb; 2616 case num_resources: 2617 uint32 num_resources; 2619 case uptime: 2620 uint32 uptime; 2621 /* This type may be extended */ 2623 }; 2624 } ProbeInformationData; 2626 struct { 2627 ProbeInformationType type; 2628 uint8 length; 2629 ProbeInformationData value; 2630 } ProbeInformation; 2632 struct { 2633 ProbeInformation probe_info<0..2^16-1>; 2634 } ProbeAns; 2636 A ProbeAns message contains a sequence of ProbeInformation 2637 structures. Each has a "length" indicating the length of the 2638 following value field. This structure allows for unknown option 2639 types. 2641 Each of the current possible Probe information types is a 32-bit 2642 unsigned integer. For type "responsible_ppb", it is the fraction of 2643 the overlay for which the peer is responsible in parts per billion. 2644 For type "num_resources", it is the number of resources the peer is 2645 storing. For the type "uptime" it is the number of seconds the peer 2646 has been up. 2648 The responding peer SHOULD include any values that the requesting 2649 node requested and that it recognizes. They SHOULD be returned in 2650 the requested order. Any other values MUST NOT be returned. 2652 5.5. Forwarding and Link Management Layer 2654 Each node maintains connections to a set of other nodes defined by 2655 the topology plugin. This section defines the methods RELOAD uses to 2656 form and maintain connections between nodes in the overlay. Three 2657 methods are defined: 2659 Attach: used to form RELOAD connections between nodes using ICE 2660 for NAT traversal. When node A wants to connect to node B, it 2661 sends an Attach message to node B through the overlay. The Attach 2662 contains A's ICE parameters. B responds with its ICE parameters 2663 and the two nodes perform ICE to form connection. Attach also 2664 allows two nodes to connect via No-ICE instead of full ICE. 2665 AppAttach: used to form application layer connections between 2666 nodes. 2667 Ping: is a simple request/response which is used to verify 2668 connectivity of the target peer. 2670 5.5.1. Attach 2672 A node sends an Attach request when it wishes to establish a direct 2673 TCP or UDP connection to another node for the purpose of sending 2674 RELOAD messages. A client that can establish a connection directly 2675 need not send an attach as described in the second bullet of 2676 Section 3.2.1 2678 As described in Section 5.1, an Attach may be routed to either a 2679 Node-ID or to a Resource-ID. An Attach routed to a specific Node-ID 2680 will fail if that node is not reached. An Attach routed to a 2681 Resource-ID will establish a connection with the peer currently 2682 responsible for that Resource-ID, which may be useful in establishing 2683 a direct connection to the responsible peer for use with frequent or 2684 large resource updates. 2686 An Attach in and of itself does not result in updating the routing 2687 table of either node. That function is performed by Updates. If 2688 node A has Attached to node B, but not received any Updates from B, 2689 it MAY route messages which are directly addressed to B through that 2690 channel but MUST NOT route messages through B to other peers via that 2691 channel. The process of Attaching is separate from the process of 2692 becoming a peer (using Join and Update), to prevent half-open states 2693 where a node has started to form connections but is not really ready 2694 to act as a peer. Thus, clients (unlike peers) can simply Attach 2695 without sending Join or Update. 2697 5.5.1.1. Request Definition 2699 An Attach request message contains the requesting node ICE connection 2700 parameters formatted into a binary structure. 2702 enum { reservedOverlayLink(0), DTLS-UDP-SR(1), 2703 DTLS-UDP-SR-NO-ICE(3), TLS-TCP-FH-NO-ICE(4), 2704 (255) } OverlayLinkType; 2706 enum { reservedCand(0), host(1), srflx(2), prflx(3), relay(4), 2707 (255) } CandType; 2709 struct { 2710 opaque name<0..2^16-1>; 2711 opaque value<0..2^16-1>; 2712 } IceExtension; 2714 struct { 2715 IpAddressPort addr_port; 2716 OverlayLinkType overlay_link; 2717 opaque foundation<0..255>; 2718 uint32 priority; 2719 CandType type; 2720 select (type){ 2721 case host: 2722 ; /* Nothing */ 2723 case srflx: 2724 case prflx: 2725 case relay: 2726 IpAddressPort rel_addr_port; 2727 }; 2728 IceExtension extensions<0..2^16-1>; 2729 } IceCandidate; 2731 struct { 2732 opaque ufrag<0..2^8-1>; 2733 opaque password<0..2^8-1>; 2734 opaque role<0..2^8-1>; 2735 IceCandidate candidates<0..2^16-1>; 2736 Boolean send_update; 2737 } AttachReqAns; 2739 The values contained in AttachReqAns are: 2741 ufrag 2742 The username fragment (from ICE). 2744 password 2745 The ICE password. 2747 role 2748 An active/passive/actpass attribute from RFC 4145 [RFC4145]. This 2749 value MUST be 'passive' for the offerer (the peer sending the 2750 Attach request) and 'active' for the answerer (the peer sending 2751 the Attach response). 2753 candidates 2754 One or more ICE candidate values, as described below. 2755 send_update 2756 Has the same meaning as the send_update field in RouteQueryReq. 2758 Each ICE candidate is represented as an IceCandidate structure, which 2759 is a direct translation of the information from the ICE string 2760 structures, with the exception of the component ID. Since there is 2761 only one component, it is always 1, and thus left out of the PDU. 2762 The remaining values are specified as follows: 2764 addr_port 2765 corresponds to the connection-address and port productions. 2767 overlay_link 2768 corresponds to the OverlayLinkType production, Overlay Link 2769 protocols used with No-ICE MUST specify "No-ICE" in their 2770 description. Future overlay link values can be added be defining 2771 new OverlayLinkType values in the IANA registry in Section 13.10. 2772 Future extensions to the encapsulation or framing that provide for 2773 backward compatibility with that specified by a previously defined 2774 OverlayLinkType values MUST use that previous value. 2775 OverlayLinkType protocols are defined in Section 5.6 2776 A single AttachReqAns MUST NOT include both candidates whose 2777 OverlayLinkType protocols use ICE (the default) and candidates 2778 that specify "No-ICE". 2780 foundation 2781 corresponds to the foundation production. 2783 priority 2784 corresponds to the priority production. 2786 type 2787 corresponds to the cand-type production. 2789 rel_addr_port 2790 corresponds to the rel-addr and rel-port productions. Only 2791 present for type "relay". 2793 extensions 2794 ICE extensions. The name and value fields correspond to binary 2795 translations of the equivalent fields in the ICE extensions. 2797 These values should be generated using the procedures described in 2798 Section 5.5.1.3. 2800 5.5.1.2. Response Definition 2802 If a peer receives an Attach request, it MUST determine how to 2803 process the request as follows: 2805 o If it has not initiated an Attach request to the originating peer 2806 of this Attach request, it MUST process this request and SHOULD 2807 generate its own response with an AttachReqAns. It should then 2808 begin ICE checks. 2809 o If it has already sent an Attach request to and received the 2810 response from the originating peer of this Attach request, and as 2811 a as a result, an ICE check and TLS connection is in progress, 2812 then it SHOULD generate an Error_In_Progress error instead of an 2813 AttachReqAns. 2814 o If it has already sent an Attach request to but not yet received 2815 the response from the originating peer of this Attach request, it 2816 SHOULD apply the following tie-breaker heuristic to determine how 2817 to handle this Attach request and the incomplete Attach request it 2818 has sent out: 2819 * If the peer's own Node-ID is smaller when compared as big- 2820 endian unsigned integers, it MUST cancel its own incomplete 2821 Attach request. It MUST then process this Attach request, 2822 generate an AttachReqAns response, and proceed with the 2823 corresponding ICE check. 2824 * If the peer's own Node-ID is larger when compared as big-endien 2825 unsigned integers, it MUST generate an Error_In_Progress error 2826 to this Attach request, then proceed to wait for and complete 2827 the Attach and the corresponding ICE check it has originated. 2828 o If the peer is overloaded or detects some other kind of error, it 2829 MAY generate an error instead of an AttachReqAns. 2831 When a peer receives an Attach response, it SHOULD parse the response 2832 and begin its own ICE checks. 2834 5.5.1.3. Using ICE With RELOAD 2836 This section describes the profile of ICE that is used with RELOAD. 2837 RELOAD implementations MUST implement full ICE. 2839 In ICE as defined by [RFC5245], SDP is used to carry the ICE 2840 parameters. In RELOAD, this function is performed by a binary 2841 encoding in the Attach method. This encoding is more restricted than 2842 the SDP encoding because the RELOAD environment is simpler: 2844 o Only a single media stream is supported. 2845 o In this case, the "stream" refers not to RTP or other types of 2846 media, but rather to a connection for RELOAD itself or for SIP 2847 signaling. 2848 o RELOAD only allows for a single offer/answer exchange. Unlike the 2849 usage of ICE within SIP, there is never a need to send a 2850 subsequent offer to update the default candidates to match the 2851 ones selected by ICE. 2853 An agent follows the ICE specification as described in [RFC5245] with 2854 the changes and additional procedures described in the subsections 2855 below. 2857 5.5.1.4. Collecting STUN Servers 2859 ICE relies on the node having one or more STUN servers to use. In 2860 conventional ICE, it is assumed that nodes are configured with one or 2861 more STUN servers through some out of band mechanism. This is still 2862 possible in RELOAD but RELOAD also learns STUN servers as it connects 2863 to other peers. Because all RELOAD peers implement ICE and use STUN 2864 keepalives, every peer is a capable of responding to STUN Binding 2865 requests [RFC5389]. Accordingly, any peer that a node knows about 2866 can be used like a STUN server -- though of course it may be behind a 2867 NAT. 2869 A peer on a well-provisioned wide-area overlay will be configured 2870 with one or more bootstrap nodes. These nodes make an initial list 2871 of STUN servers. However, as the peer forms connections with 2872 additional peers, it builds more peers it can use like STUN servers. 2874 Because complicated NAT topologies are possible, a peer may need more 2875 than one STUN server. Specifically, a peer that is behind a single 2876 NAT will typically observe only two IP addresses in its STUN checks: 2877 its local address and its server reflexive address from a STUN server 2878 outside its NAT. However, if there are more NATs involved, it may 2879 learn additional server reflexive addresses (which vary based on 2880 where in the topology the STUN server is). To maximize the chance of 2881 achieving a direct connection, a peer SHOULD group other peers by the 2882 peer-reflexive addresses it discovers through them. It SHOULD then 2883 select one peer from each group to use as a STUN server for future 2884 connections. 2886 Only peers to which the peer currently has connections may be used. 2887 If the connection to that host is lost, it MUST be removed from the 2888 list of stun servers and a new server from the same group MUST be 2889 selected unless there are no others servers in the group in which 2890 case some other peer MAY be used. 2892 5.5.1.5. Gathering Candidates 2894 When a node wishes to establish a connection for the purposes of 2895 RELOAD signaling or application signaling, it follows the process of 2896 gathering candidates as described in Section 4 of ICE [RFC5245]. 2897 RELOAD utilizes a single component. Consequently, gathering for 2898 these "streams" requires a single component. In the case where a 2899 node has not yet found a TURN server, the agent would not include a 2900 relayed candidate. 2902 The ICE specification assumes that an ICE agent is configured with, 2903 or somehow knows of, TURN and STUN servers. RELOAD provides a way 2904 for an agent to learn these by querying the overlay, as described in 2905 Section 5.5.1.4 and Section 8. 2907 The default candidate selection described in Section 4.1.4 of ICE is 2908 ignored; defaults are not signaled or utilized by RELOAD. 2910 An alternative to using the full ICE supported by the Attach request 2911 is to use No-ICE mechanism by providing candidates with "No-ICE" 2912 Overlay Link protocols. Configuration for the overlay indicates 2913 whether or not these Overlay Link protocols can be used. An overlay 2914 MUST be either all ICE or all No-ICE. 2916 No-ICE will not work in all of the scenarios where ICE would work, 2917 but in some cases, particularly those with no NATs or firewalls, it 2918 will work. Therefore it is RECOMMENDED that full ICE be used even 2919 for a node that has a public, unfiltered IP address, to take 2920 advantage of STUN connectivity checks, etc. 2922 5.5.1.6. Prioritizing Candidates 2924 At the time of writing, UDP is the only transport protocol specified 2925 to work with ICE. However, standardization of additional protocols 2926 for use with ICE is expected, including TCP and datagram-oriented 2927 protocols such as SCTP and DCCP. In particular, UDP encapsulations 2928 for SCTP and DCCP are expected to be standardized in the near future, 2929 greatly expanding the available Overlay Link protocols available for 2930 RELOAD. When additional protocols are available, the following 2931 prioritization is RECOMMENDED: 2933 o Highest priority is assigned to message-oriented protocols that 2934 offer well-understood congestion and flow control without head of 2935 line blocking. For example, SCTP without message ordering, DCCP, 2936 or those protocols encapsulated using UDP. 2937 o Second highest priority is assigned to stream-oriented protocols, 2938 e.g. TCP. 2939 o Lowest priority is assigned to protocols encapsulated over UDP 2940 that do not implement well-established congestion control 2941 algorithms. The DTLS/UDP with SR overlay link protocol is an 2942 example of such a protocol. 2944 5.5.1.7. Encoding the Attach Message 2946 Section 4.3 of ICE describes procedures for encoding the SDP for 2947 conveying RELOAD candidates. Instead of actually encoding an SDP, 2948 the candidate information (IP address and port and transport 2949 protocol, priority, foundation, type and related address) is carried 2950 within the attributes of the Attach request or its response. 2951 Similarly, the username fragment and password are carried in the 2952 Attach message or its response. Section 5.5.1 describes the detailed 2953 attribute encoding for Attach. The Attach request and its response 2954 do not contain any default candidates or the ice-lite attribute, as 2955 these features of ICE are not used by RELOAD. 2957 Since the Attach request contains the candidate information and short 2958 term credentials, it is considered as an offer for a single media 2959 stream that happens to be encoded in a format different than SDP, but 2960 is otherwise considered a valid offer for the purposes of following 2961 the ICE specification. Similarly, the Attach response is considered 2962 a valid answer for the purposes of following the ICE specification. 2964 5.5.1.8. Verifying ICE Support 2966 An agent MUST skip the verification procedures in Section 5.1 and 6.1 2967 of ICE. Since RELOAD requires full ICE from all agents, this check 2968 is not required. 2970 5.5.1.9. Role Determination 2972 The roles of controlling and controlled as described in Section 5.2 2973 of ICE are still utilized with RELOAD. However, the offerer (the 2974 entity sending the Attach request) will always be controlling, and 2975 the answerer (the entity sending the Attach response) will always be 2976 controlled. The connectivity checks MUST still contain the ICE- 2977 CONTROLLED and ICE-CONTROLLING attributes, however, even though the 2978 role reversal capability for which they are defined will never be 2979 needed with RELOAD. This is to allow for a common codebase between 2980 ICE for RELOAD and ICE for SDP. 2982 5.5.1.10. Full ICE 2984 When neither side has provided an No-ICE candidate, connectivity 2985 checks and nominations are used as in regular ICE. 2987 5.5.1.10.1. Connectivity Checks 2989 The processes of forming check lists in Section 5.7 of ICE, 2990 scheduling checks in Section 5.8, and checking connectivity checks in 2991 Section 7 are used with RELOAD without change. 2993 5.5.1.10.2. Concluding ICE 2995 The procedures in Section 8 of ICE are followed to conclude ICE, with 2996 the following exceptions: 2998 o The controlling agent MUST NOT attempt to send an updated offer 2999 once the state of its single media stream reaches Completed. 3000 o Once the state of ICE reaches Completed, the agent can immediately 3001 free all unused candidates. This is because RELOAD does not have 3002 the concept of forking, and thus the three second delay in Section 3003 8.3 of ICE does not apply. 3005 5.5.1.10.3. Media Keepalives 3007 STUN MUST be utilized for the keepalives described in Section 10 of 3008 ICE. 3010 5.5.1.11. No-ICE 3012 No-ICE is selected when either side has provided "no ICE" Overlay 3013 Link candidates. STUN is not used for connectivity checks when doing 3014 No-ICE; instead the DTLS or TLS handshake (or similar security layer 3015 of future overlay link protocols) forms the connectivity check. The 3016 certificate exchanged during the (D)TLS handshake must match the node 3017 that sent the AttachReqAns and if it does not, the connection MUST be 3018 closed. 3020 5.5.1.12. Subsequent Offers and Answers 3022 An agent MUST NOT send a subsequent offer or answer. Thus, the 3023 procedures in Section 9 of ICE MUST be ignored. 3025 5.5.1.13. Sending Media 3027 The procedures of Section 11 of ICE apply to RELOAD as well. 3028 However, in this case, the "media" takes the form of application 3029 layer protocols (RELOAD) over TLS or DTLS. Consequently, once ICE 3030 processing completes, the agent will begin TLS or DTLS procedures to 3031 establish a secure connection. The node which sent the Attach 3032 request MUST be the TLS server. The other node MUST be the TLS 3033 client. The server MUST request TLS client authentication. The 3034 nodes MUST verify that the certificate presented in the handshake 3035 matches the identity of the other peer as found in the Attach 3036 message. Once the TLS or DTLS signaling is complete, the application 3037 protocol is free to use the connection. 3039 The concept of a previous selected pair for a component does not 3040 apply to RELOAD, since ICE restarts are not possible with RELOAD. 3042 5.5.1.14. Receiving Media 3044 An agent MUST be prepared to receive packets for the application 3045 protocol (TLS or DTLS carrying RELOAD, SIP or anything else) at any 3046 time. The jitter and RTP considerations in Section 11 of ICE do not 3047 apply to RELOAD. 3049 5.5.2. AppAttach 3051 A node sends an AppAttach request when it wishes to establish a 3052 direct connection to another node for the purposes of sending 3053 application layer messages. AppAttach is nearly identical to Attach, 3054 except for the purpose of the connection: it is used to transport 3055 non-RELOAD "media". A separate request is used to avoid implementor 3056 confusion between the two methods (this was found to be a real 3057 problem with initial implementations). The AppAttach request and its 3058 response contain an application attribute, which indicates what 3059 protocol is to be run over the connection. 3061 5.5.2.1. Request Definition 3063 An AppAttachReq message contains the requesting node's ICE connection 3064 parameters formatted into a binary structure. 3066 struct { 3067 opaque ufrag<0..2^8-1>; 3068 opaque password<0..2^8-1>; 3069 uint16 application; 3070 opaque role<0..2^8-1>; 3071 IceCandidate candidates<0..2^16-1>; 3072 } AppAttachReq; 3074 The values contained in AppAttachReq and AppAttachAns are: 3076 ufrag 3077 The username fragment (from ICE) 3079 password 3080 The ICE password. 3082 application 3083 A 16-bit application-id as defined in the Section 13.5. This 3084 number represents the IANA registered application that is going to 3085 send data on this connection. For SIP, this is 5060 or 5061. 3087 role 3088 An active/passive/actpass attribute from RFC 4145 [RFC4145]. 3090 candidates 3091 One or more ICE candidate values 3093 The application using connection set up with this request is 3094 responsible for providing sufficiently frequent keep traffic for NAT 3095 and Firewall keep alive and for deciding when to close the 3096 connection. 3098 5.5.2.2. Response Definition 3100 If a peer receives an AppAttach request, it SHOULD process the 3101 request and generate its own response with a AppAttachAns. It should 3102 then begin ICE checks. When a peer receives an AppAttach response, 3103 it SHOULD parse the response and begin its own ICE checks. If the 3104 application ID is not supported, the peer MUST reply with an 3105 Error_Not_Found error. 3107 struct { 3108 opaque ufrag<0..2^8-1>; 3109 opaque password<0..2^8-1>; 3110 uint16 application; 3111 opaque role<0..2^8-1>; 3112 IceCandidate candidates<0..2^16-1>; 3113 } AppAttachAns; 3115 The meaning of the fields is the same as in the AppAttachReq. 3117 5.5.3. Ping 3119 Ping is used to test connectivity along a path. A ping can be 3120 addressed to a specific Node-ID, to the peer controlling a given 3121 location (by using a resource ID), or to the broadcast Node-ID 3122 (2^128-1). 3124 5.5.3.1. Request Definition 3126 struct { 3127 opaque<0..2^16-1> padding; 3128 } PingReq 3130 The Ping request is empty of meaningful contents. However, it may 3131 contain up to 65535 bytes of padding to facilitate the discovery of 3132 overlay maximum packet sizes. 3134 5.5.3.2. Response Definition 3136 A successful PingAns response contains the information elements 3137 requested by the peer. 3139 struct { 3140 uint64 response_id; 3141 uint64 time; 3142 } PingAns; 3144 A PingAns message contains the following elements: 3146 response_id 3147 A randomly generated 64-bit response ID. This is used to 3148 distinguish Ping responses. 3150 time 3151 The time when the Ping response was created represented in the 3152 same way as storage_time defined in Section 6. 3154 5.5.4. ConfigUpdate 3156 The ConfigUpdate method is used to push updated configuration data 3157 across the overlay. Whenever a node detects that another node has 3158 old configuration data, it MUST generate a ConfigUpdate request. The 3159 ConfigUpdate request allows updating of two kinds of data: the 3160 configuration data (Section 5.3.2.1) and kind information 3161 (Section 6.4.1.1). 3163 5.5.4.1. Request Definition 3165 enum { reservedConfigUpdate(0), config(1), kind(2), (255) } 3166 ConfigUpdateType; 3168 typedef uint32 KindId; 3169 typedef opaque KindDescription<0..2^16-1>; 3171 struct { 3172 ConfigUpdateType type; 3173 uint32 length; 3175 select (type) { 3176 case config: 3177 opaque config_data<0..2^24-1>; 3179 case kind: 3180 KindDescription kinds<0..2^24-1>; 3182 /* This structure may be extended with new types*/ 3183 }; 3184 } ConfigUpdateReq; 3186 The ConfigUpdateReq message contains the following elements: 3188 type 3189 The type of the contents of the message. This structure allows 3190 for unknown content types. 3191 length 3192 The length of the remainder of the message. This is included to 3193 preserve backward compatibility and is 32 bits instead of 24 to 3194 facilitate easy conversion between network and host byte order. 3195 config_data (type==config) 3196 The contents of the configuration document. 3197 kinds (type==kind) 3198 One or more XML kind-block productions (see Section 10.1). These 3199 MUST be encoded with UTF-8 and assume a default namespace of 3200 "urn:ietf:params:xml:ns:p2p:config-base". 3202 5.5.4.2. Response Definition 3204 struct { 3205 } ConfigUpdateAns 3207 If the ConfigUpdateReq is of type "config" it MUST only be processed 3208 if all the following are true: 3210 o The sequence number in the document is greater than the current 3211 configuration sequence number. 3212 o The configuration document is correctly digitally signed (see 3213 Section 10 for details on signatures. 3214 Otherwise appropriate errors MUST be generated. 3216 If the ConfigUpdateReq is of type "kind" it MUST only be processed if 3217 it is correctly digitally signed by an acceptable kind signer as 3218 specified in the configuration file. Details on kind-signer field in 3219 the configuration file is described in Section 10.1. In addition, if 3220 the kind update conflicts with an existing known kind (i.e., it is 3221 signed by a different signer), then it should be rejected with 3222 "Error_Forbidden". This should not happen in correctly functioning 3223 overlays. 3225 If the update is acceptable, then the node MUST reconfigure itself to 3226 match the new information. This may include adding permissions for 3227 new kinds, deleting old kinds, or even, in extreme circumstances, 3228 exiting and reentering the overlay, if, for instance, the DHT 3229 algorithm has changed. 3231 The response for ConfigUpdate is empty. 3233 5.6. Overlay Link Layer 3235 RELOAD can use multiple Overlay Link protocols to send its messages. 3236 Because ICE is used to establish connections (see Section 5.5.1.3), 3237 RELOAD nodes are able to detect which Overlay Link protocols are 3238 offered by other nodes and establish connections between them. Any 3239 link protocol needs to be able to establish a secure, authenticated 3240 connection and to provide data origin authentication and message 3241 integrity for individual data elements. RELOAD currently supports 3242 three Overlay Link protocols: 3244 o DTLS [RFC4347] over UDP with Simple Reliability (SR) 3245 o TLS [RFC5246] over TCP with Framing Header, No-ICE 3246 o DTLS [RFC4347] over UDP with SR, No-ICE 3248 Note that although UDP does not properly have "connections", both TLS 3249 and DTLS have a handshake which establishes a similar, stateful 3250 association, and we simply refer to these as "connections" for the 3251 purposes of this document. 3253 If a peer receives a message that is larger than value of max- 3254 message-size defined in the overlay configuration, the peer SHOULD 3255 send an Error_Message_Too_Large error and then close the TLS or DTLS 3256 session from which the message was received. Note that this error 3257 can be sent and the session closed before receiving the complete 3258 message. If the forwarding header is larger than the max-message- 3259 size, the receiver SHOULD close the TLS or DTLS session without 3260 sending an error. 3262 The Framing Header (FH) is used to frame messages and provide timing 3263 when used on a reliable stream-based transport protocol. Simple 3264 Reliability (SR) makes use of the FH to provide congestion control 3265 and semi-reliability when using unreliable message-oriented transport 3266 protocols. We will first define each of these algorithms, then 3267 define overlay link protocols that use them. 3269 Note: We expect future Overlay Link protocols to define replacements 3270 for all components of these protocols, including the framing header. 3271 These protocols have been chosen for simplicity of implementation and 3272 reasonable performance. 3274 Note to implementers: There are inherent tradeoffs in utilizing 3275 short timeouts to determine when a link has failed. To balance the 3276 tradeoffs, an implementation should be able to quickly act to remove 3277 entries from the routing table when there is reason to suspect the 3278 link has failed. For example, in a Chord-derived overlay algorithm, 3279 a closer finger table entry could be substituted for an entry in the 3280 finger table that has experienced a timeout. That entry can be 3281 restored if it proves to resume functioning, or replaced at some 3282 point in the future if necessary. End-to-end retransmissions will 3283 handle any lost messages, but only if the failing entries do not 3284 remain in the finger table for subsequent retransmissions. 3286 5.6.1. Future Overlay Link Protocols 3288 The only currently defined overlay link protocols are TLS and DTLS. 3289 It is possible to define new link-layer protocols and apply them to a 3290 new overlay using the "overlay-link-protocol" configuration directive 3291 (see Section 10.1.). However, any new protocols MUST meet the 3292 following requirements. 3294 Endpoint authentication When a node forms an association with 3295 another endpoint, it MUST be possible to cryptographically verify 3296 that the endpoint has a given Node-Id. 3298 Traffic origin authentication and integrity When a node receives 3299 traffic from another endpoint, it MUST be possible to 3300 cryptographically verify that the traffic came from a given 3301 association and that it has not been modified in transit from the 3302 other endpoint in the association. The overlay link protocol MUST 3303 also provide replay prevention/detection. 3305 Traffic confidentiality When a node sends traffic to another 3306 endpoint, it MUST NOT be possible for a third party not involved 3307 in the association to determine the contents of that traffic. 3309 Any new overlay protocol MUST be defined via RFC 5226 Standards 3310 Action; see Section 13.11. 3312 5.6.1.1. HIP 3314 In a Host Identity Protocol Based Overlay Networking Environment (HIP 3315 BONE) [RFC6079] HIP [RFC5201] provides connection management (e.g., 3316 NAT traversal and mobility) and security for the overlay network. 3317 The P2PSIP Working Group has expressed interest in supporting a HIP- 3318 based link protocol. Such support would require specifying such 3319 details as: 3321 o How to issue certificates which provided identities meaningful to 3322 the HIP base exchange. We anticipate that this would require a 3323 mapping between ORCHIDs and NodeIds. 3324 o How to carry the HIP I1 and I2 messages. 3325 o How to carry RELOAD messages over HIP. 3327 [I-D.ietf-hip-reload-instance] documents work in progress on using 3328 RELOAD with the HIP BONE. 3330 5.6.1.2. ICE-TCP 3332 The ICE-TCP draft [I-D.ietf-mmusic-ice-tcp] should allow TCP to be 3333 supported as an Overlay Link protocol that can be added using ICE. 3335 5.6.1.3. Message-oriented Transports 3337 Modern message-oriented transports offer high performance, good 3338 congestion control, and avoid head of line blocking in case of lost 3339 data. These characteristics make them preferable as underlying 3340 transport protocols for RELOAD links. SCTP without message ordering 3341 and DCCP are two examples of such protocols. However, currently they 3342 are not well-supported by commonly available NATs, and specifications 3343 for ICE session establishment are not available. 3345 5.6.1.4. Tunneled Transports 3347 As of the time of this writing, there is significant interest in the 3348 IETF community in tunneling other transports over UDP, motivated by 3349 the situation that UDP is well-supported by modern NAT hardware, and 3350 similar performance can be achieved to native implementation. 3351 Currently SCTP, DCCP, and a generic tunneling extension are being 3352 proposed for message-oriented protocols. Baset et al. have proposed 3353 tunneling TCP over UDP for similar reasons 3354 [I-D.baset-tsvwg-tcp-over-udp]. Once ICE traversal has been 3355 specified for these tunneled protocols, they should be 3356 straightforward to support as overlay link protocols. 3358 5.6.2. Framing Header 3360 In order to support unreliable links and to allow for quick detection 3361 of link failures when using reliable end-to-end transports, each 3362 message is wrapped in a very simple framing layer (FramedMessage) 3363 which is only used for each hop. This layer contains a sequence 3364 number which can then be used for ACKs. The same header is used for 3365 both reliable and unreliable transports for simplicity of 3366 implementation. 3368 The definition of FramedMessage is: 3370 enum { data(128), ack(129), (255)} FramedMessageType; 3372 struct { 3373 FramedMessageType type; 3375 select (type) { 3376 case data: 3377 uint32 sequence; 3378 opaque message<0..2^24-1>; 3380 case ack: 3381 uint32 ack_sequence; 3382 uint32 received; 3383 }; 3384 } FramedMessage; 3386 The type field of the PDU is set to indicate whether the message is 3387 data or an acknowledgement. 3389 If the message is of type "data", then the remainder of the PDU is as 3390 follows: 3392 sequence 3393 the sequence number. This increments by 1 for each framed message 3394 sent over this transport session. 3396 message 3397 the message that is being transmitted. 3399 Each connection has it own sequence number space. Initially the 3400 value is zero and it increments by exactly one for each message sent 3401 over that connection. 3403 When the receiver receives a message, it SHOULD immediately send an 3404 ACK message. The receiver MUST keep track of the 32 most recent 3405 sequence numbers received on this association in order to generate 3406 the appropriate ack. 3408 If the PDU is of type "ack", the contents are as follows: 3410 ack_sequence 3411 The sequence number of the message being acknowledged. 3413 received 3414 A bitmask indicating if each of the previous 32 sequence numbers 3415 before this packet has been among the 32 packets most recently 3416 received on this connection. When a packet is received with a 3417 sequence number N, the receiver looks at the sequence number of 3418 the previously 32 packets received on this connection. Call the 3419 previously received packet number M. For each of the previous 32 3420 packets, if the sequence number M is less than N but greater than 3421 N-32, the N-M bit of the received bitmask is set to one; otherwise 3422 it is zero. Note that a bit being set to one indicates positively 3423 that a particular packet was received, but a bit being set to zero 3424 means only that it is unknown whether or not the packet has been 3425 received, because it might have been received before the 32 most 3426 recently received packets. 3428 The received field bits in the ACK provide a high degree of 3429 redundancy so that the sender can figure out which packets the 3430 receiver has received and can then estimate packet loss rates. If 3431 the sender also keeps track of the time at which recent sequence 3432 numbers have been sent, the RTT can be estimated. 3434 5.6.3. Simple Reliability 3436 When RELOAD is carried over DTLS or another unreliable link protocol, 3437 it needs to be used with a reliability and congestion control 3438 mechanism, which is provided on a hop-by-hop basis. The basic 3439 principle is that each message, regardless of whether or not it 3440 carries a request or response, will get an ACK and be reliably 3441 retransmitted. The receiver's job is very simple, limited to just 3442 sending ACKs. All the complexity is at the sender side. This allows 3443 the sending implementation to trade off performance versus 3444 implementation complexity without affecting the wire protocol. 3446 5.6.3.1. Retransmission and Flow Control 3448 Because the receiver's role is limited to providing packet 3449 acknowledgements, a wide variety of congestion control algorithms can 3450 be implemented on the sender side while using the same basic wire 3451 protocol. In general, senders MAY implement any rate control scheme 3452 of their choice, provided that it is REQUIRED to be no more 3453 aggressive then TFRC[RFC5348]. 3455 The following section describes a simple, inefficient, scheme that 3456 complies with this requirement. Another alternative would be TFRC-SP 3457 [RFC4828] and use the received bitmask to allow the sender to compute 3458 packet loss event rates. 3460 5.6.3.1.1. Trivial Retransmission 3462 A node SHOULD retransmit a message if it has not received an ACK 3463 after an interval of RTO ("Retransmission TimeOut"). The node MUST 3464 double the time to wait after each retransmission. In each 3465 retransmission, the sequence number is incremented. 3467 The RTO is an estimate of the round-trip time (RTT). Implementations 3468 can use a static value for RTO or a dynamic estimate which will 3469 result in better performance. For implementations that use a static 3470 value, the default value for RTO is 500 ms. Nodes MAY use smaller 3471 values of RTO if it is known that all nodes are within the local 3472 network. The default RTO MAY be chosen larger, and this is 3473 RECOMMENDED if it is known in advance (such as on high latency access 3474 links) that the round-trip time is larger. 3476 Implementations that use a dynamic estimate to compute the RTO MUST 3477 use the algorithm described in RFC 2988[RFC2988], with the exception 3478 that the value of RTO SHOULD NOT be rounded up to the nearest second 3479 but instead rounded up to the nearest millisecond. The RTT of a 3480 successful STUN transaction from the ICE stage is used as the initial 3481 measurement for formula 2.2 of RFC 2988. The sender keeps track of 3482 the time each message was sent for all recently sent messages. Any 3483 time an ACK is received, the sender can compute the RTT for that 3484 message by looking at the time the ACK was received and the time when 3485 the message was sent. This is used as a subsequent RTT measurement 3486 for formula 2.3 of RFC 2988 to update the RTO estimate. (Note that 3487 because retransmissions receive new sequence numbers, all received 3488 ACKs are used.) 3489 The value for RTO is calculated separately for each DTLS session. 3491 Retransmissions continue until a response is received, or until a 3492 total of 5 requests have been sent or there has been a hard ICMP 3493 error [RFC1122] or a TLS alert. The sender knows a response was 3494 received when it receives an ACK with a sequence number that 3495 indicates it is a response to one of the transmissions of this 3496 messages. For example, assuming an RTO of 500 ms, requests would be 3497 sent at times 0 ms, 500 ms, 1500 ms, 3500 ms, and 7500 ms. If all 3498 retransmissions for a message fail, then the sending node SHOULD 3499 close the connection routing the message. 3501 To determine when a link may be failing without waiting for the final 3502 timeout, observe when no ACKs have been received for an entire RTO 3503 interval, and then wait for three retransmissions to occur beyond 3504 that point. If no ACKs have been received by the time the third 3505 retransmission occurs, it is RECOMMENDED that the link be removed 3506 from the routing table. The link MAY be restored to the routing 3507 table if ACKs resume before the connection is closed, as described 3508 above. 3510 Once an ACK has been received for a message, the next message can be 3511 sent, but the peer SHOULD ensure that there is at least 10 ms between 3512 sending any two messages. The only time a value less than 10 ms can 3513 be used is when it is known that all nodes are on a network that can 3514 support retransmissions faster than 10 ms with no congestion issues. 3516 5.6.4. DTLS/UDP with SR 3518 This overlay link protocol consists of DTLS over UDP while 3519 implementing the Simple Reliability protocol. STUN Connectivity 3520 checks and keepalives are used. 3522 5.6.5. TLS/TCP with FH, No-ICE 3524 This overlay link protocol consists of TLS over TCP with the framing 3525 header. Because ICE is not used, STUN connectivity checks are not 3526 used upon establishing the TCP connection, nor are they used for 3527 keepalives. 3529 Because the TCP layer's application-level timeout is too slow to be 3530 useful for overlay routing, the Overlay Link implementation MUST use 3531 the framing header to measure the RTT of the connection and calculate 3532 an RTO as specified in Section 2 of [RFC2988]. The resulting RTO is 3533 not used for retransmissions, but as a timeout to indicate when the 3534 link SHOULD be removed from the routing table. It is RECOMMENDED 3535 that such a connection be retained for 30s to determine if the 3536 failure was transient before concluding the link has failed 3537 permanently. 3539 When sending candidates for TLS/TCP with FH, No-ICE, a passive 3540 candidate MUST be provided. 3542 5.6.6. DTLS/UDP with SR, No-ICE 3544 This overlay link protocol consists of DTLS over UDP while 3545 implementing the Simple Reliability protocol. Because ICE is not 3546 used, no STUN connectivity checks or keepalives are used. 3548 5.7. Fragmentation and Reassembly 3550 In order to allow transmission over datagram protocols such as DTLS, 3551 RELOAD messages may be fragmented. 3553 Any node along the path can fragment the message but only the final 3554 destination reassembles the fragments. When a node takes a packet 3555 and fragments it, each fragment has a full copy of the Forwarding 3556 Header but the data after the Forwarding Header is broken up in 3557 appropriate sized chunks. The size of the payload chunks needs to 3558 take into account space to allow the via and destination lists to 3559 grow. Each fragment MUST contain a full copy of the via and 3560 destination list and MUST contain at least 256 bytes of the message 3561 body. If the via and destination list are so large that this is not 3562 possible, RELOAD fragmentation is not performed and IP-layer 3563 fragmentation is allowed to occur. When a message must be 3564 fragmented, it SHOULD be split into equal-sized fragments that are no 3565 larger than the PMTU of the next overlay link minus 32 bytes. This 3566 is to allow the via list to grow before further fragmentation is 3567 required. 3569 Note that this fragmentation is not optimal for the end-to-end path - 3570 a message may be refragmented multiple times as it traverses the 3571 overlay but is only assembled at the final destination. This option 3572 has been chosen as it is far easier to implement than e2e PMTU 3573 discovery across an ever-changing overlay, and it effectively 3574 addresses the reliability issues of relying on IP-layer 3575 fragmentation. However, PING can be used to allow e2e PMTU to be 3576 implemented if desired. 3578 Upon receipt of a fragmented message by the intended peer, the peer 3579 holds the fragments in a holding buffer until the entire message has 3580 been received. The message is then reassembled into a single message 3581 and processed. In order to mitigate denial of service attacks, 3582 receivers SHOULD time out incomplete fragments after maximum request 3583 lifetime (15 seconds). Note this time was derived from looking at 3584 the end to end retransmission time and saving fragments long enough 3585 for the full end to end retransmissions to take place. Ideally the 3586 receiver would have enough buffer space to deal with as many 3587 fragments as can arrive in the maximum request lifetime. However, if 3588 the receiver runs out of buffer space to reassemble the messages it 3589 MUST drop the message. 3591 When a message is fragmented, the fragment offset value is stored in 3592 the lower 24 bits of the fragment field of the forwarding header. 3593 The offset is the number of bytes between the end of the forwarding 3594 header and the start of the data. The first fragment therefore has 3595 an offset of 0. The first and last bit indicators MUST be 3596 appropriately set. If the message is not fragmented, then both the 3597 first and last fragment bits are set to 1 and the offset is 0 3598 resulting in a fragment value of 0xC0000000. Note that this means 3599 that the first fragment bit is always 1, so isn't actually that 3600 useful. 3602 6. Data Storage Protocol 3604 RELOAD provides a set of generic mechanisms for storing and 3605 retrieving data in the Overlay Instance. These mechanisms can be 3606 used for new applications simply by defining new code points and a 3607 small set of rules. No new protocol mechanisms are required. 3609 The basic unit of stored data is a single StoredData structure: 3611 struct { 3612 uint32 length; 3613 uint64 storage_time; 3614 uint32 lifetime; 3615 StoredDataValue value; 3616 Signature signature; 3617 } StoredData; 3619 The contents of this structure are as follows: 3621 length 3622 The size of the StoredData structure in octets excluding the size 3623 of length itself. 3625 storage_time 3626 The time when the data was stored represented as the number of 3627 milliseconds elapsed since midnight Jan 1, 1970 UTC not counting 3628 leap seconds. This will have the same values for seconds as 3629 standard UNIX time or POSIX time. More information can be found 3630 at [UnixTime]. Any attempt to store a data value with a storage 3631 time before that of a value already stored at this location MUST 3632 generate a Error_Data_Too_Old error. This prevents rollback 3633 attacks. The node SHOULD make a best-effort attempt to use a 3634 correct clock to determine this number, however, the protocol does 3635 not require synchronized clocks: the receiving peer uses the 3636 storage time in the previous store, not its own clock. Clock 3637 values are used so that when clocks are generally synchronized, 3638 data may be stored in a single transaction, rather than querying 3639 for the value of a counter before the actual store. 3640 If a node attempting to store new data in response to a user 3641 request (rather than as an overlay maintenance operation such as 3642 occurs during unpartitioning) is rejected with an 3643 Error_Data_Too_Old error, the node MAY elect to perform its store 3644 using a storage_time that increments the value used with the 3645 previous store. This situation may occur when the clocks of nodes 3646 storing to this location are not properly synchronized. 3648 lifetime 3649 The validity period for the data, in seconds, starting from the 3650 time the peer receives the StoreReq. 3652 value 3653 The data value itself, as described in Section 6.2. 3655 signature 3656 A signature as defined in Section 6.1. 3658 Each Resource-ID specifies a single location in the Overlay Instance. 3659 However, each location may contain multiple StoredData values 3660 distinguished by Kind-ID. The definition of a kind describes both 3661 the data values which may be stored and the data model of the data. 3662 Some data models allow multiple values to be stored under the same 3663 Kind-ID. Section Section 6.2 describes the available data models. 3664 Thus, for instance, a given Resource-ID might contain a single-value 3665 element stored under Kind-ID X and an array containing multiple 3666 values stored under Kind-ID Y. 3668 6.1. Data Signature Computation 3670 Each StoredData element is individually signed. However, the 3671 signature also must be self-contained and cover the Kind-ID and 3672 Resource-ID even though they are not present in the StoredData 3673 structure. The input to the signature algorithm is: 3675 resource_id || kind || storage_time || StoredDataValue || 3676 SignerIdentity 3678 Where || indicates concatenation. 3680 Where these values are: 3682 resource_id 3683 The resource ID where this data is stored. 3685 kind 3686 The Kind-ID for this data. 3688 storage_time 3690 The contents of the storage_time data value. 3691 StoredDataValue 3692 The contents of the stored data value, as described in the 3693 previous sections. 3695 SignerIdentity 3696 The signer identity as defined in Section 5.3.4. 3698 Once the signature has been computed, the signature is represented 3699 using a signature element, as described in Section 5.3.4. 3701 6.2. Data Models 3703 The protocol currently defines the following data models: 3705 o single value 3706 o array 3707 o dictionary 3709 These are represented with the StoredDataValue structure. The actual 3710 dataModel is known from the kind being stored. 3712 struct { 3713 Boolean exists; 3714 opaque value<0..2^32-1>; 3715 } DataValue; 3717 struct { 3718 select (dataModel) { 3719 case single_value: 3720 DataValue single_value_entry; 3722 case array: 3723 ArrayEntry array_entry; 3725 case dictionary: 3726 DictionaryEntry dictionary_entry; 3728 /* This structure may be extended */ 3729 }; 3730 } StoredDataValue; 3732 We now discuss the properties of each data model in turn: 3734 6.2.1. Single Value 3736 A single-value element is a simple sequence of bytes. There may be 3737 only one single-value element for each Resource-ID, Kind-ID pair. 3739 A single value element is represented as a DataValue, which contains 3740 the following two elements: 3742 exists 3743 This value indicates whether the value exists at all. If it is 3744 set to False, it means that no value is present. If it is True, 3745 that means that a value is present. This gives the protocol a 3746 mechanism for indicating nonexistence as opposed to emptiness. 3748 value 3749 The stored data. 3751 6.2.2. Array 3753 An array is a set of opaque values addressed by an integer index. 3754 Arrays are zero based. Note that arrays can be sparse. For 3755 instance, a Store of "X" at index 2 in an empty array produces an 3756 array with the values [ NA, NA, "X"]. Future attempts to fetch 3757 elements at index 0 or 1 will return values with "exists" set to 3758 False. 3760 A array element is represented as an ArrayEntry: 3762 struct { 3763 uint32 index; 3764 DataValue value; 3765 } ArrayEntry; 3767 The contents of this structure are: 3769 index 3770 The index of the data element in the array. 3772 value 3773 The stored data. 3775 6.2.3. Dictionary 3777 A dictionary is a set of opaque values indexed by an opaque key with 3778 one value for each key. A single dictionary entry is represented as 3779 follows: 3781 A dictionary element is represented as a DictionaryEntry: 3783 typedef opaque DictionaryKey<0..2^16-1>; 3785 struct { 3786 DictionaryKey key; 3787 DataValue value; 3788 } DictionaryEntry; 3790 The contents of this structure are: 3792 key 3793 The dictionary key for this value. 3795 value 3796 The stored data. 3798 6.3. Access Control Policies 3800 Every kind which is storable in an overlay MUST be associated with an 3801 access control policy. This policy defines whether a request from a 3802 given node to operate on a given value should succeed or fail. It is 3803 anticipated that only a small number of generic access control 3804 policies are required. To that end, this section describes a small 3805 set of such policies and Section 13.4 establishes a registry for new 3806 policies if required. Each policy has a short string identifier 3807 which is used to reference it in the configuration document. 3809 In the following policies, the term "signer" refers to the signer of 3810 the StoredValue object and, in the case of non-replica stores, to the 3811 signer of the StoreReq message. I.e., in a non-replica store, both 3812 the signer of the StoredValue and the signer of the StoreReq MUST 3813 conform to the policy. In the case of a replica store, the signer of 3814 the StoredValue MUST conform to the policy and the StoreReq itself 3815 MUST be checked as described in Section 6.4.1.1. 3817 6.3.1. USER-MATCH 3819 In the USER-MATCH policy, a given value MUST be written (or 3820 overwritten) if and only if the signer's certificate has a user name 3821 which hashes (using the hash function for the overlay) to the 3822 Resource-ID for the resource. Recall that the certificate may, 3823 depending on the overlay configuration, be self-signed. 3825 6.3.2. NODE-MATCH 3827 In the NODE-MATCH policy, a given value MUST be written (or 3828 overwritten) if and only if the signer's certificate has a specified 3829 Node-ID which hashes (using the hash function for the overlay) to the 3830 Resource-ID for the resource and that Node-ID is the one indicated in 3831 the SignerIdentity value cert_hash. 3833 6.3.3. USER-NODE-MATCH 3835 The USER-NODE-MATCH policy may only be used with dictionary types. 3836 In the USER-NODE-MATCH policy, a given value MUST be written (or 3837 overwritten) if and only if the signer's certificate has a user name 3838 which hashes (using the hash function for the overlay) to the 3839 Resource-ID for the resource. In addition, the dictionary key MUST 3840 be equal to the Node-ID in the certificate and that Node-ID MUST be 3841 the one indicated in the SignerIdentity value cert_hash. 3843 6.3.4. NODE-MULTIPLE 3845 In the NODE-MULTIPLE policy, a given value MUST be written (or 3846 overwritten) if and only if signer's certificate contains a Node-ID 3847 such that H(Node-ID || i) is equal to the Resource-ID for some small 3848 integer value of i and that Node-ID is the one indicated in the 3849 SignerIdentity value cert_hash. When this policy is in use, the 3850 maximum value of i MUST be specified in the kind definition. 3852 Note that as i is not carried on the wire, the verifier MUST iterate 3853 through potential i values up to the maximum value in order to 3854 determine whether a store is acceptable. 3856 6.4. Data Storage Methods 3858 RELOAD provides several methods for storing and retrieving data: 3860 o Store values in the overlay 3861 o Fetch values from the overlay 3862 o Stat: get metadata about values in the overlay 3863 o Find the values stored at an individual peer 3865 These methods are each described in the following sections. 3867 6.4.1. Store 3869 The Store method is used to store data in the overlay. The format of 3870 the Store request depends on the data model which is determined by 3871 the kind. 3873 6.4.1.1. Request Definition 3875 A StoreReq message is a sequence of StoreKindData values, each of 3876 which represents a sequence of stored values for a given kind. The 3877 same Kind-ID MUST NOT be used twice in a given store request. Each 3878 value is then processed in turn. These operations MUST be atomic. 3879 If any operation fails, the state MUST be rolled back to before the 3880 request was received. 3882 The store request is defined by the StoreReq structure: 3884 struct { 3885 KindId kind; 3886 uint64 generation_counter; 3887 StoredData values<0..2^32-1>; 3888 } StoreKindData; 3890 struct { 3891 ResourceId resource; 3892 uint8 replica_number; 3893 StoreKindData kind_data<0..2^32-1>; 3894 } StoreReq; 3896 A single Store request stores data of a number of kinds to a single 3897 resource location. The contents of the structure are: 3899 resource 3900 The resource to store at. 3902 replica_number 3903 The number of this replica. When a storing peer saves replicas to 3904 other peers each peer is assigned a replica number starting from 1 3905 and sent in the Store message. This field is set to 0 when a node 3906 is storing its own data. This allows peers to distinguish replica 3907 writes from original writes. 3909 kind_data 3910 A series of elements, one for each kind of data to be stored. 3912 If the replica number is zero, then the peer MUST check that it is 3913 responsible for the resource and, if not, reject the request. If the 3914 replica number is nonzero, then the peer MUST check that it expects 3915 to be a replica for the resource and that the request sender is 3916 consistent with being the responsible node (i.e., that the receiving 3917 peer does not know of a better node) and, if not, reject the request. 3919 Each StoreKindData element represents the data to be stored for a 3920 single Kind-ID. The contents of the element are: 3922 kind 3923 The Kind-ID. Implementations MUST reject requests corresponding 3924 to unknown kinds. 3926 generation_counter 3927 The expected current state of the generation counter 3928 (approximately the number of times this object has been written; 3929 see below for details). 3931 values 3932 The value or values to be stored. This may contain one or more 3933 stored_data values depending on the data model associated with 3934 each kind. 3936 The peer MUST perform the following checks: 3938 o The Kind-ID is known and supported. 3939 o The signatures over each individual data element (if any) are 3940 valid. If this check fails, the request MUST be rejected with an 3941 Error_Forbidden error. 3942 o Each element is signed by a credential which is authorized to 3943 write this kind at this Resource-ID. If this check fails, the 3944 request MUST be rejected with an Error_Forbidden error. 3946 o For original (non-replica) stores, the StoreReq is signed by a 3947 credential which is authorized to write this kind at this 3948 Resource-Id. If this check fails, the request MUST be rejected 3949 with an Error_Forbidden error. 3950 o For replica stores, the StoreReq is signed by a Node-Id which is a 3951 plausible node to either have originally stored the value or in 3952 the replica set. What this means is overlay specific, but in the 3953 case of Chord replica StoreReqs MUST come from nodes which are 3954 either in the known replica set for a given resource or which are 3955 closer than some node in the replica set. If this check fails, 3956 the request MUST be rejected with an Error_Forbidden error. 3957 o For original (non-replica) stores, the peer MUST check that if the 3958 generation counter is non-zero, it equals the current value of the 3959 generation counter for this kind. This feature allows the 3960 generation counter to be used in a way similar to the HTTP Etag 3961 feature. 3962 o For replica Stores, the peer MUST set the generation counter to 3963 match the generation counter in the message, and MUST NOT check 3964 the generation counter against the current value. Replica Stores 3965 MUST NOT use a generation counter of 0. 3966 o The storage time values are greater than that of any value which 3967 would be replaced by this Store. 3968 o The size and number of the stored values is consistent with the 3969 limits specified in the overlay configuration. 3970 o If the data is signed with identity_type set to "none" and/or 3971 SignatureAndHashAlgorithm values set to {0, 0} ("anonymous" and 3972 "none"), the StoreReq MUST be rejected with an Error_forbidden 3973 error. Only synthesized data returned by the storage can use 3974 these values 3976 If all these checks succeed, the peer MUST attempt to store the data 3977 values. For non-replica stores, if the store succeeds and the data 3978 is changed, then the peer must increase the generation counter by at 3979 least one. If there are multiple stored values in a single 3980 StoreKindData, it is permissible for the peer to increase the 3981 generation counter by only 1 for the entire Kind-ID, or by 1 or more 3982 than one for each value. Accordingly, all stored data values must 3983 have a generation counter of 1 or greater. 0 is used in the Store 3984 request to indicate that the generation counter should be ignored for 3985 processing this request; however the responsible peer should increase 3986 the stored generation counter and should return the correct 3987 generation counter in the response. 3989 When a peer stores data previously stored by another node (e.g., for 3990 replicas or topology shifts) it MUST adjust the lifetime value 3991 downward to reflect the amount of time the value was stored at the 3992 peer. The adjustment SHOULD be implemented by an algorithm 3993 equivalent to the following: at the time the peer initially receives 3994 the StoreReq it notes the local time T. When it then attempts to do a 3995 StoreReq to another node it should decrement the lifetime value by 3996 the difference between the current local time and T. 3998 Unless otherwise specified by the usage, if a peer attempts to store 3999 data previously stored by another node (e.g., for replicas or 4000 topology shifts) and that store fails with either an 4001 Error_Generation_Counter_Too_Low or an Error_Data_Too old error, the 4002 peer MUST fetch the newer data from the peer generating the error and 4003 use that to replace its own copy. This rule allows resynchronization 4004 after partitions heal. 4006 The properties of stores for each data model are as follows: 4008 Single-value: 4009 A store of a new single-value element creates the element if it 4010 does not exist and overwrites any existing value with the new 4011 value. 4013 Array: 4014 A store of an array entry replaces (or inserts) the given value at 4015 the location specified by the index. Because arrays are sparse, a 4016 store past the end of the array extends it with nonexistent values 4017 (exists=False) as required. A store at index 0xffffffff places 4018 the new value at the end of the array regardless of the length of 4019 the array. The resulting StoredData has the correct index value 4020 when it is subsequently fetched. 4022 Dictionary: 4023 A store of a dictionary entry replaces (or inserts) the given 4024 value at the location specified by the dictionary key. 4026 The following figure shows the relationship between these structures 4027 for an example store which stores the following values at resource 4028 "1234" 4030 o The value "abc" in the single value location for kind X 4031 o The value "foo" at index 0 in the array for kind Y 4032 o The value "bar" at index 1 in the array for kind Y 4033 Store 4034 resource=1234 4035 replica_number = 0 4036 / \ 4037 / \ 4038 StoreKindData StoreKindData 4039 kind=X (Single-Value) kind=Y (Array) 4040 generation_counter = 99 generation_counter = 107 4041 | /\ 4042 | / \ 4043 StoredData / \ 4044 storage_time = xxxxxxx / \ 4045 lifetime = 86400 / \ 4046 signature = XXXX / \ 4047 | | | 4048 | StoredData StoredData 4049 | storage_time = storage_time = 4050 | yyyyyyyy zzzzzzz 4051 | lifetime = 86400 lifetime = 33200 4052 | signature = YYYY signature = ZZZZ 4053 | | | 4054 StoredDataValue | | 4055 value="abc" | | 4056 | | 4057 StoredDataValue StoredDataValue 4058 index=0 index=1 4059 value="foo" value="bar" 4061 6.4.1.2. Response Definition 4063 In response to a successful Store request the peer MUST return a 4064 StoreAns message containing a series of StoreKindResponse elements 4065 containing the current value of the generation counter for each 4066 Kind-ID, as well as a list of the peers where the data will be 4067 replicated by the node processing the request. 4069 struct { 4070 KindId kind; 4071 uint64 generation_counter; 4072 NodeId replicas<0..2^16-1>; 4073 } StoreKindResponse; 4075 struct { 4076 StoreKindResponse kind_responses<0..2^16-1>; 4077 } StoreAns; 4079 The contents of each StoreKindResponse are: 4081 kind 4082 The Kind-ID being represented. 4084 generation_counter 4085 The current value of the generation counter for that Kind-ID. 4087 replicas 4088 The list of other peers at which the data was/will be replicated. 4089 In overlays and applications where the responsible peer is 4090 intended to store redundant copies, this allows the storing peer 4091 to independently verify that the replicas have in fact been 4092 stored. It does this verification by using the Stat method. Note 4093 that the storing peer is not required to perform this 4094 verification. 4096 The response itself is just StoreKindResponse values packed end-to- 4097 end. 4099 If any of the generation counters in the request precede the 4100 corresponding stored generation counter, then the peer MUST fail the 4101 entire request and respond with an Error_Generation_Counter_Too_Low 4102 error. The error_info in the ErrorResponse MUST be a StoreAns 4103 response containing the correct generation counter for each kind and 4104 the replica list, which will be empty. For original (non-replica) 4105 stores, a node which receives such an error SHOULD attempt to fetch 4106 the data and, if the storage_time value is newer, replace its own 4107 data with that newer data. This rule improves data consistency in 4108 the case of partitions and merges. 4110 If the data being stored is too large for the allowed limit by the 4111 given usage, then the peer MUST fail the request and generate an 4112 Error_Data_Too_Large error. 4114 If any type of request tries to access a data kind that the node does 4115 not know about, an Error_Unknown_Kind MUST be generated. The 4116 error_info in the Error_Response is: 4118 KindId unknown_kinds<0..2^8-1>; 4120 which lists all the kinds that were unrecognized. A node which 4121 receives this error MUST generate a ConfigUpdate message which 4122 contains the appropriate kind definition (assuming that in fact a 4123 kind was used which was defined in the configuration document). 4125 6.4.1.3. Removing Values 4127 This version of RELOAD (unlike previous versions) does not have an 4128 explicit Remove operation. Rather, values are Removed by storing 4129 "nonexistent" values in their place. Each DataValue contains a 4130 boolean value called "exists" which indicates whether a value is 4131 present at that location. In order to effectively remove a value, 4132 the owner stores a new DataValue with "exists" set to "false": 4134 exists = false 4135 value = {} (0 length) 4137 The owner SHOULD use a lifetime for the nonexistent value at least as 4138 long as the remainder of the lifetime of the value it is replacing; 4139 otherwise it is possible for the original value to be accidentally or 4140 maliciously re-stored after the storing node has expired it. Note 4141 that there is still a window of vulnerability for replay attack after 4142 the original lifetime has expired (as with any store). This attack 4143 can be mitigated by doing a nonexistent store with a very long 4144 lifetime. 4146 Storing nodes MUST treat these nonexistent values the same way they 4147 treat any other stored value, including overwriting the existing 4148 value, replicating them, and aging them out as necessary when 4149 lifetime expires. When a stored nonexistent value's lifetime 4150 expires, it is simply removed from the storing node like any other 4151 stored value expiration. 4153 Note that in the case of arrays and dictionaries, expiration may 4154 create an implicit, unsigned "nonexistent" value to represent a gap 4155 in the data structure, as might happen when any value is aged out. 4156 However, this value isn't persistent nor is it replicated. It is 4157 simply synthesized by the storing node. 4159 6.4.2. Fetch 4161 The Fetch request retrieves one or more data elements stored at a 4162 given Resource-ID. A single Fetch request can retrieve multiple 4163 different kinds. 4165 6.4.2.1. Request Definition 4167 struct { 4168 int32 first; 4169 int32 last; 4170 } ArrayRange; 4172 struct { 4173 KindId kind; 4174 uint64 generation; 4175 uint16 length; 4177 select (dataModel) { 4178 case single_value: ; /* Empty */ 4180 case array: 4181 ArrayRange indices<0..2^16-1>; 4183 case dictionary: 4184 DictionaryKey keys<0..2^16-1>; 4186 /* This structure may be extended */ 4188 } model_specifier; 4189 } StoredDataSpecifier; 4191 struct { 4192 ResourceId resource; 4193 StoredDataSpecifier specifiers<0..2^16-1>; 4194 } FetchReq; 4196 The contents of the Fetch requests are as follows: 4198 resource 4199 The Resource-ID to fetch from. 4201 specifiers 4202 A sequence of StoredDataSpecifier values, each specifying some of 4203 the data values to retrieve. 4205 Each StoredDataSpecifier specifies a single kind of data to retrieve 4206 and (if appropriate) the subset of values that are to be retrieved. 4207 The contents of the StoredDataSpecifier structure are as follows: 4209 kind 4210 The Kind-ID of the data being fetched. Implementations SHOULD 4211 reject requests corresponding to unknown kinds unless specifically 4212 configured otherwise. 4214 dataModel 4215 The data model of the data. This is not transmitted on the wire 4216 but comes from the definition of the kind. 4218 generation 4219 The last generation counter that the requesting node saw. This 4220 may be used to avoid unnecessary fetches or it may be set to zero. 4222 length 4223 The length of the rest of the structure, thus allowing 4224 extensibility. 4226 model_specifier 4227 A reference to the data value being requested within the data 4228 model specified for the kind. For instance, if the data model is 4229 "array", it might specify some subset of the values. 4231 The model_specifier is as follows: 4233 o If the data model is single value, the specifier is empty. 4234 o If the data model is array, the specifier contains a list of 4235 ArrayRange elements, each of which contains two integers. The 4236 first integer is the beginning of the range and the second is the 4237 end of the range. 0 is used to indicate the first element and 4238 0xffffffff is used to indicate the final element. The first 4239 integer must be less than the second. While multiple ranges MAY 4240 be specified, they MUST NOT overlap. 4241 o If the data model is dictionary then the specifier contains a list 4242 of the dictionary keys being requested. If no keys are specified, 4243 than this is a wildcard fetch and all key-value pairs are 4244 returned. 4246 The generation counter is used to indicate the requester's expected 4247 state of the storing peer. If the generation counter in the request 4248 matches the stored counter, then the storing peer returns a response 4249 with no StoredData values. 4251 Note that because the certificate for a user is typically stored at 4252 the same location as any data stored for that user, a requesting node 4253 that does not already have the user's certificate should request the 4254 certificate in the Fetch as an optimization. 4256 6.4.2.2. Response Definition 4258 The response to a successful Fetch request is a FetchAns message 4259 containing the data requested by the requester. 4261 struct { 4262 KindId kind; 4263 uint64 generation; 4264 StoredData values<0..2^32-1>; 4265 } FetchKindResponse; 4267 struct { 4268 FetchKindResponse kind_responses<0..2^32-1>; 4269 } FetchAns; 4271 The FetchAns structure contains a series of FetchKindResponse 4272 structures. There MUST be one FetchKindResponse element for each 4273 Kind-ID in the request. 4275 The contents of the FetchKindResponse structure are as follows: 4277 kind 4278 the kind that this structure is for. 4280 generation 4281 the generation counter for this kind. 4283 values 4284 the relevant values. If the generation counter in the request 4285 matches the generation counter in the stored data, then no 4286 StoredData values are returned. Otherwise, all relevant data 4287 values MUST be returned. A nonexistent value (i.e., one which the 4288 node has no knowledge of) is represented by a synthetic value with 4289 "exists" set to False and has an empty signature. Specifically, 4290 the identity_type is set to "none", the SignatureAndHashAlgorithm 4291 values are set to {0, 0} ("anonymous" and "none" respectively), 4292 and the signature value is of zero length. This removes the need 4293 for the responding node to do signatures for values which do not 4294 exist. These signatures are unnecessary as the entire response is 4295 signed by that node. Note that entries which have been removed by 4296 the procedure of Section 6.4.1.3 and have not yet expired also 4297 have exists = false but have valid signatures from the node which 4298 did the store. 4300 There is one subtle point about signature computation on arrays. If 4301 the storing node uses the append feature (where the 4302 index=0xffffffff), then the index in the StoredData that is returned 4303 will not match that used by the storing node, which would break the 4304 signature. In order to avoid this issue, the index value in the 4305 array is set to zero before the signature is computed. This implies 4306 that malicious storing nodes can reorder array entries without being 4307 detected. 4309 6.4.3. Stat 4311 The Stat request is used to get metadata (length, generation counter, 4312 digest, etc.) for a stored element without retrieving the element 4313 itself. The name is from the UNIX stat(2) system call which performs 4314 a similar function for files in a file system. It also allows the 4315 requesting node to get a list of matching elements without requesting 4316 the entire element. 4318 6.4.3.1. Request Definition 4320 The Stat request is identical to the Fetch request. It simply 4321 specifies the elements to get metadata about. 4323 struct { 4324 ResourceId resource; 4325 StoredDataSpecifier specifiers<0..2^16-1>; 4326 } StatReq; 4328 6.4.3.2. Response Definition 4330 The Stat response contains the same sort of entries that a Fetch 4331 response would contain; however, instead of containing the element 4332 data it contains metadata. 4334 struct { 4335 Boolean exists; 4336 uint32 value_length; 4337 HashAlgorithm hash_algorithm; 4338 opaque hash_value<0..255>; 4339 } MetaData; 4341 struct { 4342 uint32 index; 4343 MetaData value; 4344 } ArrayEntryMeta; 4346 struct { 4347 DictionaryKey key; 4348 MetaData value; 4349 } DictionaryEntryMeta; 4351 struct { 4352 select (model) { 4353 case single_value: 4354 MetaData single_value_entry; 4356 case array: 4357 ArrayEntryMeta array_entry; 4359 case dictionary: 4360 DictionaryEntryMeta dictionary_entry; 4362 /* This structure may be extended */ 4363 }; 4364 } MetaDataValue; 4366 struct { 4367 uint32 value_length; 4368 uint64 storage_time; 4369 uint32 lifetime; 4370 MetaDataValue metadata; 4371 } StoredMetaData; 4373 struct { 4374 KindId kind; 4375 uint64 generation; 4376 StoredMetaData values<0..2^32-1>; 4377 } StatKindResponse; 4379 struct { 4380 StatKindResponse kind_responses<0..2^32-1>; 4381 } StatAns; 4383 The structures used in StatAns parallel those used in FetchAns: a 4384 response consists of multiple StatKindResponse values, one for each 4385 kind that was in the request. The contents of the StatKindResponse 4386 are the same as those in the FetchKindResponse, except that the 4387 values list contains StoredMetaData entries instead of StoredData 4388 entries. 4390 The contents of the StoredMetaData structure are the same as the 4391 corresponding fields in StoredData except that there is no signature 4392 field and the value is a MetaDataValue rather than a StoredDataValue. 4394 A MetaDataValue is a variant structure, like a StoredDataValue, 4395 except for the types of each arm, which replace DataValue with 4396 MetaData. 4398 The only really new structure is MetaData, which has the following 4399 contents: 4401 exists 4402 Same as in DataValue 4404 value_length 4405 The length of the stored value. 4407 hash_algorithm 4408 The hash algorithm used to perform the digest of the value. 4410 hash_value 4411 A digest of the value using hash_algorithm. 4413 6.4.4. Find 4415 The Find request can be used to explore the Overlay Instance. A Find 4416 request for a Resource-ID R and a Kind-ID T retrieves the Resource-ID 4417 (if any) of the resource of kind T known to the target peer which is 4418 closest to R. This method can be used to walk the Overlay Instance by 4419 iteratively fetching R_n+1=nearest(1 + R_n). 4421 6.4.4.1. Request Definition 4423 The FindReq message contains a Resource-ID and a series of Kind-IDs 4424 identifying the resource the peer is interested in. 4426 struct { 4427 ResourceId resource; 4428 KindId kinds<0..2^8-1>; 4429 } FindReq; 4431 The request contains a list of Kind-IDs which the Find is for, as 4432 indicated below: 4434 resource 4435 The desired Resource-ID 4437 kinds 4438 The desired Kind-IDs. Each value MUST only appear once, and if 4439 not the request MUST be rejected with an error. 4441 6.4.4.2. Response Definition 4443 A response to a successful Find request is a FindAns message 4444 containing the closest Resource-ID on the peer for each kind 4445 specified in the request. 4447 struct { 4448 KindId kind; 4449 ResourceId closest; 4450 } FindKindData; 4452 struct { 4453 FindKindData results<0..2^16-1>; 4454 } FindAns; 4456 If the processing peer is not responsible for the specified 4457 Resource-ID, it SHOULD return an Error_Not_Found error code. 4459 For each Kind-ID in the request the response MUST contain a 4460 FindKindData indicating the closest Resource-ID for that Kind-ID, 4461 unless the kind is not allowed to be used with Find in which case a 4462 FindKindData for that Kind-ID MUST NOT be included in the response. 4463 If a Kind-ID is not known, then the corresponding Resource-ID MUST be 4464 0. Note that different Kind-IDs may have different closest Resource- 4465 IDs. 4467 The response is simply a series of FindKindData elements, one per 4468 kind, concatenated end-to-end. The contents of each element are: 4470 kind 4471 The Kind-ID. 4473 closest 4474 The closest resource ID to the specified resource ID. This is 0 4475 if no resource ID is known. 4477 Note that the response does not contain the contents of the data 4478 stored at these Resource-IDs. If the requester wants this, it must 4479 retrieve it using Fetch. 4481 6.4.5. Defining New Kinds 4483 There are two ways to define a new kind. The first is by writing a 4484 document and registering the kind-id with IANA. This is the 4485 preferred method for kinds which may be widely used and reused. The 4486 second method is to simply define the kind and its parameters in the 4487 configuration document using the section of kind-id space set aside 4488 for private use. This method MAY be used to define ad hoc kinds in 4489 new overlays. 4491 However a kind is defined, the definition must include: 4493 o The meaning of the data to be stored (in some textual form). 4494 o The Kind-ID. 4495 o The data model (single value, array, dictionary, etc). 4496 o The access control model. 4498 In addition, when kinds are registered with IANA, each kind is 4499 assigned a short string name which is used to refer to it in 4500 configuration documents. 4502 While each kind needs to define what data model is used for its data, 4503 that does not mean that it must define new data models. Where 4504 practical, kinds should use the existing data models. The intention 4505 is that the basic data model set be sufficient for most applications/ 4506 usages. 4508 7. Certificate Store Usage 4510 The Certificate Store usage allows a peer to store its certificate in 4511 the overlay, thus avoiding the need to send a certificate in each 4512 message - a reference may be sent instead. 4514 A user/peer MUST store its certificate at Resource-IDs derived from 4515 two Resource Names: 4517 o The user name in the certificate. 4519 o The Node-ID in the certificate. 4521 Note that in the second case the certificate is not stored at the 4522 peer's Node-ID but rather at a hash of the peer's Node-ID. The 4523 intention here (as is common throughout RELOAD) is to avoid making a 4524 peer responsible for its own data. 4526 A peer MUST ensure that the user's certificates are stored in the 4527 Overlay Instance. New certificates are stored at the end of the 4528 list. This structure allows users to store an old and a new 4529 certificate that both have the same Node-ID, which allows for 4530 migration of certificates when they are renewed. 4532 This usage defines the following kinds: 4534 Name: CERTIFICATE_BY_NODE 4536 Data Model: The data model for CERTIFICATE_BY_NODE data is array. 4538 Access Control: NODE-MATCH. 4540 Name: CERTIFICATE_BY_USER 4542 Data Model: The data model for CERTIFICATE_BY_USER data is array. 4544 Access Control: USER-MATCH. 4546 8. TURN Server Usage 4548 The TURN server usage allows a RELOAD peer to advertise that it is 4549 prepared to be a TURN server as defined in [RFC5766]. When a node 4550 starts up, it joins the overlay network and forms several connections 4551 in the process. If the ICE stage in any of these connections returns 4552 a reflexive address that is not the same as the peer's perceived 4553 address, then the peer is behind a NAT and not a candidate for a TURN 4554 server. Additionally, if the peer's IP address is in the private 4555 address space range, then it is also not a candidate for a TURN 4556 server. Otherwise, the peer SHOULD assume it is a potential TURN 4557 server and follow the procedures below. 4559 If the node is a candidate for a TURN server it will insert some 4560 pointers in the overlay so that other peers can find it. The overlay 4561 configuration file specifies a turn-density parameter that indicates 4562 how many times each TURN server should record itself in the overlay. 4563 Typically this should be set to the reciprocal of the estimate of 4564 what percentage of peers will act as TURN servers. If the turn- 4565 density is not set to zero, for each value, called d, between 1 and 4566 turn-density, the peer forms a Resource Name by concatenating its 4567 Node-ID and the value d. This Resource Name is hashed to form a 4568 Resource-ID. The address of the peer is stored at that Resource-ID 4569 using type TURN-SERVICE and the TurnServer object: 4571 struct { 4572 uint8 iteration; 4573 IpAddressAndPort server_address; 4574 } TurnServer; 4576 The contents of this structure are as follows: 4578 iteration 4579 the d value 4581 server_address 4582 the address at which the TURN server can be contacted. 4584 Note: Correct functioning of this algorithm depends on having turn- 4585 density be an reasonable estimate of the reciprocal of the 4586 proportion of nodes in the overlay that can act as TURN servers. 4587 If the turn-density value in the configuration file is too low, 4588 then the process of finding TURN servers becomes more expensive as 4589 multiple candidate Resource-IDs must be probed to find a TURN 4590 server. 4592 Peers that provide this service need to support the TURN extensions 4593 to STUN for media relay as defined in [RFC5766]. 4595 This usage defines the following kind to indicate that a peer is 4596 willing to act as a TURN server: 4598 Name TURN-SERVICE 4599 Data Model The TURN-SERVICE kind stores a single value for each 4600 Resource-ID. 4601 Access Control NODE-MULTIPLE, with maximum iteration counter 20. 4603 Peers can find other servers by selecting a random Resource-ID and 4604 then doing a Find request for the appropriate Kind-ID with that 4605 Resource-ID. The Find request gets routed to a random peer based on 4606 the Resource-ID. If that peer knows of any servers, they will be 4607 returned. The returned response may be empty if the peer does not 4608 know of any servers, in which case the process gets repeated with 4609 some other random Resource-ID. As long as the ratio of servers 4610 relative to peers is not too low, this approach will result in 4611 finding a server relatively quickly. 4613 9. Chord Algorithm 4615 This algorithm is assigned the name chord-reload to indicate it is an 4616 adaptation of the basic Chord DHT algorithm. 4618 This algorithm differs from the originally presented Chord algorithm 4619 [Chord]. It has been updated based on more recent research results 4620 and implementation experiences, and to adapt it to the RELOAD 4621 protocol. A short list of differences: 4623 o The original Chord algorithm specified that a single predecessor 4624 and a successor list be stored. The chord-reload algorithm 4625 attempts to have more than one predecessor and successor. The 4626 predecessor sets help other neighbors learn their successor list. 4627 o The original Chord specification and analysis called for iterative 4628 routing. RELOAD specifies recursive routing. In addition to the 4629 performance implications, the cost of NAT traversal dictates 4630 recursive routing. 4631 o Finger table entries are indexed in opposite order. Original 4632 Chord specifies finger[0] as the immediate successor of the peer. 4633 chord-reload specifies finger[0] as the peer 180 degrees around 4634 the ring from the peer. This change was made to simplify 4635 discussion and implementation of variable sized finger tables. 4636 However, with either approach no more than O(log N) entries should 4637 typically be stored in a finger table. 4638 o The stabilize() and fix_fingers() algorithms in the original Chord 4639 algorithm are merged into a single periodic process. 4640 Stabilization is implemented slightly differently because of the 4641 larger neighborhood, and fix_fingers is not as aggressive to 4642 reduce load, nor does it search for optimal matches of the finger 4643 table entries. 4644 o RELOAD uses a 128 bit hash instead of a 160 bit hash, as RELOAD is 4645 not designed to be used in networks with close to or more than 4646 2^128 nodes (and it is hard to see how one would assemble such a 4647 network). 4648 o RELOAD uses randomized finger entries as described in 4649 Section 9.7.4.2. 4650 o This algorithm allows the use of either reactive or periodic 4651 recovery. The original Chord paper used periodic recovery. 4652 Reactive recovery provides better performance in small overlays, 4653 but is believed to be unstable in large (>1000) overlays with high 4654 levels of churn [handling-churn-usenix04]. The overlay 4655 configuration file specifies a "chord-reactive" element that 4656 indicates whether reactive recovery should be used. 4658 9.1. Overview 4660 The algorithm described here is a modified version of the Chord 4661 algorithm. Each peer keeps track of a finger table and a neighbor 4662 table. The neighbor table contains at least the three peers before 4663 and after this peer in the DHT ring. There may not be three entries 4664 in all cases such as small rings or while the ring topology is 4665 changing. The first entry in the finger table contains the peer 4666 half-way around the ring from this peer; the second entry contains 4667 the peer that is 1/4 of the way around; the third entry contains the 4668 peer that is 1/8th of the way around, and so on. Fundamentally, the 4669 chord data structure can be thought of a doubly-linked list formed by 4670 knowing the successors and predecessor peers in the neighbor table, 4671 sorted by the Node-ID. As long as the successor peers are correct, 4672 the DHT will return the correct result. The pointers to the prior 4673 peers are kept to enable the insertion of new peers into the list 4674 structure. Keeping multiple predecessor and successor pointers makes 4675 it possible to maintain the integrity of the data structure even when 4676 consecutive peers simultaneously fail. The finger table forms a skip 4677 list, so that entries in the linked list can be found in O(log(N)) 4678 time instead of the typical O(N) time that a linked list would 4679 provide. 4681 A peer, n, is responsible for a particular Resource-ID k if k is less 4682 than or equal to n and k is greater than p, where p is the Node-ID of 4683 the previous peer in the neighbor table. Care must be taken when 4684 computing to note that all math is modulo 2^128. 4686 9.2. Hash Function 4688 For this Chord topology plugin, the size of the Resource-ID is 128 4689 bits. The hash of a Resource-ID is computed using SHA-1 4690 [RFC3174]then truncating the SHA-1 result to the most significant 128 4691 bits. 4693 9.3. Routing 4695 The routing table is the union of the neighbor table and the finger 4696 table. 4698 If a peer is not responsible for a Resource-ID k, but is directly 4699 connected to a node with Node-ID k, then it routes the message to 4700 that node. Otherwise, it routes the request to the peer in the 4701 routing table that has the largest Node-ID that is in the interval 4702 between the peer and k. If no such node is found, it finds the 4703 smallest Node-Id that is greater than k and routes the message to 4704 that node. 4706 9.4. Redundancy 4708 When a peer receives a Store request for Resource-ID k, and it is 4709 responsible for Resource-ID k, it stores the data and returns a 4710 success response. It then sends a Store request to its successor in 4711 the neighbor table and to that peer's successor. Note that these 4712 Store requests are addressed to those specific peers, even though the 4713 Resource-ID they are being asked to store is outside the range that 4714 they are responsible for. The peers receiving these check they came 4715 from an appropriate predecessor in their neighbor table and that they 4716 are in a range that this predecessor is responsible for, and then 4717 they store the data. They do not themselves perform further Stores 4718 because they can determine that they are not responsible for the 4719 Resource-ID. 4721 Managing replicas as the overlay changes is described in 4722 Section 9.7.3. 4724 The sequential replicas used in this overlay algorithm protect 4725 against peer failure but not against malicious peers. Additional 4726 replication from the Usage is required to protect resources from such 4727 attacks, as discussed in Section 12.5.4. 4729 9.5. Joining 4731 The join process for a joining party (JP) with Node-ID n is as 4732 follows. 4734 1. JP MUST connect to its chosen bootstrap node. 4735 2. JP SHOULD send an Attach request to the admitting peer (AP) for 4736 Node-ID n. The "send_update" flag should be used to acquire the 4737 routing table for AP. 4738 3. JP SHOULD send Attach requests to initiate connections to each of 4739 the peers in the neighbor table as well as to the desired finger 4740 table entries. Note that this does not populate their routing 4741 tables, but only their connection tables, so JP will not get 4742 messages that it is expected to route to other nodes. 4743 4. JP MUST enter all the peers it has contacted into its routing 4744 table. 4745 5. JP MUST send a Join to AP. The AP sends the response to the 4746 Join. 4747 6. AP MUST do a series of Store requests to JP to store the data 4748 that JP will be responsible for. 4749 7. AP MUST send JP an Update explicitly labeling JP as its 4750 predecessor. At this point, JP is part of the ring and 4751 responsible for a section of the overlay. AP can now forget any 4752 data which is assigned to JP and not AP. 4754 8. The AP MUST send an Update to all of its neighbors with the new 4755 values of its neighbor set (including JP). 4756 9. The JP MUST send Updates to all the peers in its neighbor table. 4758 If JP sends an Attach to AP with send_update, it immediately knows 4759 most of its expected neighbors from AP's routing table update and can 4760 directly connect to them. This is the RECOMMENDED procedure. 4762 If for some reason JP does not get AP's routing table, it can still 4763 populate its neighbor table incrementally. It sends a Ping directed 4764 at Resource-ID n+1 (directly after its own Resource-ID). This allows 4765 it to discover its own successor. Call that node p0. It then sends 4766 a ping to p0+1 to discover its successor (p1). This process can be 4767 repeated to discover as many successors as desired. The values for 4768 the two peers before p will be found at a later stage when n receives 4769 an Update. An alternate procedure is to send Attaches to those nodes 4770 rather than pings, which forms the connections immediately but may be 4771 slower if the nodes need to collect ICE candidates, thus reducing 4772 parallelism. 4774 In order to set up its finger table entry for peer i, JP simply sends 4775 an Attach to peer (n+2^(128-i). This will be routed to a peer in 4776 approximately the right location around the ring. 4778 The joining peer MUST NOT send any Update message placing itself in 4779 the overlay until it has successfully completed an Attach with each 4780 peer that should be in its neighbor table. 4782 9.6. Routing Attaches 4784 When a peer needs to Attach to a new peer in its neighbor table, it 4785 MUST source-route the Attach request through the peer from which it 4786 learned the new peer's Node-ID. Source-routing these requests allows 4787 the overlay to recover from instability. 4789 All other Attach requests, such as those for new finger table 4790 entries, are routed conventionally through the overlay. 4792 9.7. Updates 4794 A chord Update is defined as 4795 enum { reserved (0), 4796 peer_ready(1), neighbors(2), full(3), (255) } 4797 ChordUpdateType; 4799 struct { 4800 uint32 uptime; 4801 ChordUpdateType type; 4802 select(type){ 4803 case peer_ready: /* Empty */ 4804 ; 4806 case neighbors: 4807 NodeId predecessors<0..2^16-1>; 4808 NodeId successors<0..2^16-1>; 4810 case full: 4811 NodeId predecessors<0..2^16-1>; 4812 NodeId successors<0..2^16-1>; 4813 NodeId fingers<0..2^16-1>; 4814 }; 4815 } ChordUpdate; 4817 The "uptime" field contains the time this peer has been up in 4818 seconds. 4820 The "type" field contains the type of the update, which depends on 4821 the reason the update was sent. 4823 peer_ready: this peer is ready to receive messages. This message 4824 is used to indicate that a node which has Attached is a peer and 4825 can be routed through. It is also used as a connectivity check to 4826 non-neighbor peers. 4828 neighbors: this version is sent to members of the Chord neighbor 4829 table. 4831 full: this version is sent to peers which request an Update with a 4832 RouteQueryReq. 4834 If the message is of type "neighbors", then the contents of the 4835 message will be: 4837 predecessors 4838 The predecessor set of the Updating peer. 4840 successors 4841 The successor set of the Updating peer. 4843 If the message is of type "full", then the contents of the message 4844 will be: 4846 predecessors 4847 The predecessor set of the Updating peer. 4849 successors 4850 The successor set of the Updating peer. 4852 fingers 4853 The finger table of the Updating peer, in numerically ascending 4854 order. 4856 A peer MUST maintain an association (via Attach) to every member of 4857 its neighbor set. A peer MUST attempt to maintain at least three 4858 predecessors and three successors, even though this will not be 4859 possible if the ring is very small. It is RECOMMENDED that O(log(N)) 4860 predecessors and successors be maintained in the neighbor set. 4862 9.7.1. Handling Neighbor Failures 4864 Every time a connection to a peer in the neighbor table is lost (as 4865 determined by connectivity pings or the failure of some request), the 4866 peer MUST remove the entry from its neighbor table and replace it 4867 with the best match it has from the other peers in its routing table. 4868 If using reactive recovery, it then sends an immediate Update to all 4869 nodes in its Neighbor Table. The update will contain all the Node- 4870 IDs of the current entries of the table (after the failed one has 4871 been removed). Note that when replacing a successor the peer SHOULD 4872 delay the creation of new replicas for successor replacement hold- 4873 down time (30 seconds) after removing the failed entry from its 4874 neighbor table in order to allow a triggered update to inform it of a 4875 better match for its neighbor table. 4877 If the neighbor failure effects the peer's range of responsible IDs, 4878 then the Update MUST be sent to all nodes in its Connection Table. 4880 A peer MAY attempt to reestablish connectivity with a lost neighbor 4881 either by waiting additional time to see if connectivity returns or 4882 by actively routing a new Attach to the lost peer. Details for these 4883 procedures are beyond the scope of this document. In no event does 4884 an attempt to reestablish connectivity with a lost neighbor allow the 4885 peer to remain in the neighbor table. Such a peer is returned to the 4886 neighbor table once connectivity is reestablished. 4888 If connectivity is lost to all successor peers in the neighbor table, 4889 then this peer should behave as if it is joining the network and use 4890 Pings to find a peer and send it a Join. If connectivity is lost to 4891 all the peers in the finger table, this peer should assume that it 4892 has been disconnected from the rest of the network, and it should 4893 periodically try to join the DHT. 4895 9.7.2. Handling Finger Table Entry Failure 4897 If a finger table entry is found to have failed, all references to 4898 the failed peer are removed from the finger table and replaced with 4899 the closest preceding peer from the finger table or neighbor table. 4901 If using reactive recovery, the peer initiates a search for a new 4902 finger table entry as described below. 4904 9.7.3. Receiving Updates 4906 When a peer, N, receives an Update request, it examines the Node-IDs 4907 in the UpdateReq and at its neighbor table and decides if this 4908 UpdateReq would change its neighbor table. This is done by taking 4909 the set of peers currently in the neighbor table and comparing them 4910 to the peers in the update request. There are two major cases: 4912 o The UpdateReq contains peers that match N's neighbor table, so no 4913 change is needed to the neighbor set. 4914 o The UpdateReq contains peers N does not know about that should be 4915 in N's neighbor table, i.e. they are closer than entries in the 4916 neighbor table. 4918 In the first case, no change is needed. 4920 In the second case, N MUST attempt to Attach to the new peers and if 4921 it is successful it MUST adjust its neighbor set accordingly. Note 4922 that it can maintain the now inferior peers as neighbors, but it MUST 4923 remember the closer ones. 4925 After any Pings and Attaches are done, if the neighbor table changes 4926 and the peer is using reactive recovery, the peer sends an Update 4927 request to each member of its Connection Table. These Update 4928 requests are what end up filling in the predecessor/successor tables 4929 of peers that this peer is a neighbor to. A peer MUST NOT enter 4930 itself in its successor or predecessor table and instead should leave 4931 the entries empty. 4933 If peer N is responsible for a Resource-ID R, and N discovers that 4934 the replica set for R (the next two nodes in its successor set) has 4935 changed, it MUST send a Store for any data associated with R to any 4936 new node in the replica set. It SHOULD NOT delete data from peers 4937 which have left the replica set. 4939 When a peer N detects that it is no longer in the replica set for a 4940 resource R (i.e., there are three predecessors between N and R), it 4941 SHOULD delete all data associated with R from its local store. 4943 When a peer discovers that its range of responsible IDs have changed, 4944 it MUST send an Update to all entries in its connection table. 4946 9.7.4. Stabilization 4948 There are four components to stabilization: 4949 1. exchange Updates with all peers in its neighbor table to exchange 4950 state. 4951 2. search for better peers to place in its finger table. 4952 3. search to determine if the current finger table size is 4953 sufficiently large. 4954 4. search to determine if the overlay has partitioned and needs to 4955 recover. 4957 9.7.4.1. Updating neighbor table 4959 A peer MUST periodically send an Update request to every peer in its 4960 Connection Table. The purpose of this is to keep the predecessor and 4961 successor lists up to date and to detect failed peers. The default 4962 time is about every ten minutes, but the configuration server SHOULD 4963 set this in the configuration document using the "chord-update- 4964 interval" element (denominated in seconds.) A peer SHOULD randomly 4965 offset these Update requests so they do not occur all at once. 4967 9.7.4.2. Refreshing finger table 4969 A peer MUST periodically search for new peers to replace invalid 4970 entries in the finger table. A finger table entry i is valid if it 4971 is in the range [ n+2^( 128-i ) , n+2^( 128-(i-1) )-1 ]. Invalid 4972 entries occur in the finger table when a previous finger table entry 4973 has failed or when no peer has been found in that range. 4975 A peer SHOULD NOT send Ping requests looking for new finger table 4976 entries more often than the configuration element "chord-ping- 4977 interval", which defaults to 3600 seconds (one per hour). 4979 Two possible methods for searching for new peers for the finger table 4980 entries are presented: 4982 Alternative 1: A peer selects one entry in the finger table from 4983 among the invalid entries. It pings for a new peer for that finger 4984 table entry. The selection SHOULD be exponentially weighted to 4985 attempt to replace earlier (lower i) entries in the finger table. A 4986 simple way to implement this selection is to search through the 4987 finger table entries from i=0 and each time an invalid entry is 4988 encountered, send a Ping to replace that entry with probability 0.5. 4990 Alternative 2: A peer monitors the Update messages received from its 4991 connections to observe when an Update indicates a peer that would be 4992 used to replace in invalid finger table entry, i, and flags that 4993 entry in the finger table. Every "chord-ping-interval" seconds, the 4994 peer selects from among those flagged candidates using an 4995 exponentially weighted probability as above. 4997 When searching for a better entry, the peer SHOULD send the Ping to a 4998 Node-ID selected randomly from that range. Random selection is 4999 preferred over a search for strictly spaced entries to minimize the 5000 effect of churn on overlay routing [minimizing-churn-sigcomm06]. An 5001 implementation or subsequent specification MAY choose a method for 5002 selecting finger table entries other than choosing randomly within 5003 the range. Any such alternate methods SHOULD be employed only on 5004 finger table stabilization and not for the selection of initial 5005 finger table entries unless the alternative method is faster and 5006 imposes less overhead on the overlay. 5008 A peer MAY choose to keep connections to multiple peers that can act 5009 for a given finger table entry. 5011 9.7.4.3. Adjusting finger table size 5013 If the finger table has less than 16 entries, the node SHOULD attempt 5014 to discover more fingers to grow the size of the table to 16. The 5015 value 16 was chosen to ensure high odds of a node maintaining 5016 connectivity to the overlay even with strange network partitions. 5018 For many overlays, 16 finger table entries will be enough, but as an 5019 overlay grows very large, more than 16 entries may be required in the 5020 finger table for efficient routing. An implementation SHOULD be 5021 capable of increasing the number of entries in the finger table to 5022 128 entries. 5024 Note to implementers: Although log(N) entries are all that are 5025 required for optimal performance, careful implementation of 5026 stabilization will result in no additional traffic being generated 5027 when maintaining a finger table larger than log(N) entries. 5028 Implementers are encouraged to make use of RouteQuery and algorithms 5029 for determining where new finger table entries may be found. 5030 Complete details of possible implementations are outside the scope of 5031 this specification. 5033 A simple approach to sizing the finger table is to ensure the finger 5034 table is large enough to contain at least the final successor in the 5035 peer's neighbor table. 5037 9.7.4.4. Detecting partitioning 5039 To detect that a partitioning has occurred and to heal the overlay, a 5040 peer P MUST periodically repeat the discovery process used in the 5041 initial join for the overlay to locate an appropriate bootstrap node, 5042 B. P should then send a Ping for its own Node-ID routed through B. If 5043 a response is received from a peer S', which is not P's successor, 5044 then the overlay is partitioned and P should send an Attach to S' 5045 routed through B, followed by an Update sent to S'. (Note that S' 5046 may not be in P's neighbor table once the overlay is healed, but the 5047 connection will allow S' to discover appropriate neighbor entries for 5048 itself via its own stabilization.) 5050 Future specifications may describe alternative mechanisms for 5051 determining when to repeat the discovery process. 5053 9.8. Route query 5055 For this topology plugin, the RouteQueryReq contains no additional 5056 information. The RouteQueryAns contains the single node ID of the 5057 next peer to which the responding peer would have routed the request 5058 message in recursive routing: 5060 struct { 5061 NodeId next_peer; 5062 } ChordRouteQueryAns; 5064 The contents of this structure are as follows: 5066 next_peer 5067 The peer to which the responding peer would route the message in 5068 order to deliver it to the destination listed in the request. 5070 If the requester has set the send_update flag, the responder SHOULD 5071 initiate an Update immediately after sending the RouteQueryAns. 5073 9.9. Leaving 5075 To support extensions, such as [I-D.maenpaa-p2psip-self-tuning], 5076 Peers SHOULD send a Leave request to all members of their neighbor 5077 table prior to exiting the Overlay Instance. The 5078 overlay_specific_data field MUST contain the ChordLeaveData structure 5079 defined below: 5081 enum { reserved (0), 5082 from_succ(1), from_pred(2), (255) } 5083 ChordLeaveType; 5085 struct { 5086 ChordLeaveType type; 5088 select(type) { 5089 case from_succ: 5090 NodeId successors<0..2^16-1>; 5091 case from_pred: 5092 NodeId predecessors<0..2^16-1>; 5093 }; 5094 } ChordLeaveData; 5096 The 'type' field indicates whether the Leave request was sent by a 5097 predecessor or a successor of the recipient: 5099 from_succ 5100 The Leave request was sent by a successor. 5102 from_pred 5103 The Leave request was sent by a predecessor. 5105 If the type of the request is 'from_succ', the contents will be: 5107 successors 5108 The sender's successor list. 5110 If the type of the request is 'from_pred', the contents will be: 5112 predecessors 5113 The sender's predecessor list. 5115 Any peer which receives a Leave for a peer n in its neighbor set 5116 follows procedures as if it had detected a peer failure as described 5117 in Section 9.7.1. 5119 10. Enrollment and Bootstrap 5121 The section defines the format of the configuration data as well the 5122 process to join a new overlay. 5124 10.1. Overlay Configuration 5126 This specification defines a new content type "application/ 5127 p2p-overlay+xml" for an MIME entity that contains overlay 5128 information. An example document is shown below. 5130 5131 5134 5136 CHORD-RELOAD 5137 16 5138 5139 MIIDJDCCAo2gAwIBAgIBADANBgkqhkiG9w0BAQUFADBwMQswCQYDVQQGEwJVUzET 5140 MBEGA1UECBMKQ2FsaWZvcm5pYTERMA8GA1UEBxMIU2FuIEpvc2UxDjAMBgNVBAoT 5141 BXNpcGl0MSkwJwYDVQQLEyBTaXBpdCBUZXN0IENlcnRpZmljYXRlIEF1dGhvcml0 5142 eTAeFw0wMzA3MTgxMjIxNTJaFw0xMzA3MTUxMjIxNTJaMHAxCzAJBgNVBAYTAlVT 5143 MRMwEQYDVQQIEwpDYWxpZm9ybmlhMREwDwYDVQQHEwhTYW4gSm9zZTEOMAwGA1UE 5144 ChMFc2lwaXQxKTAnBgNVBAsTIFNpcGl0IFRlc3QgQ2VydGlmaWNhdGUgQXV0aG9y 5145 aXR5MIGfMA0GCSqGSIb3DQEBAQUAA4GNADCBiQKBgQDDIh6DkcUDLDyK9BEUxkud 5146 +nJ4xrCVGKfgjHm6XaSuHiEtnfELHM+9WymzkBNzZpJu30yzsxwfKoIKugdNUrD4 5147 N3viCicwcN35LgP/KnbN34cavXHr4ZlqxH+OdKB3hQTpQa38A7YXdaoz6goW2ft5 5148 Mi74z03GNKP/G9BoKOGd5QIDAQABo4HNMIHKMB0GA1UdDgQWBBRrRhcU6pR2JYBU 5149 bhNU2qHjVBShtjCBmgYDVR0jBIGSMIGPgBRrRhcU6pR2JYBUbhNU2qHjVBShtqF0 5150 pHIwcDELMAkGA1UEBhMCVVMxEzARBgNVBAgTCkNhbGlmb3JuaWExETAPBgNVBAcT 5151 CFNhbiBKb3NlMQ4wDAYDVQQKEwVzaXBpdDEpMCcGA1UECxMgU2lwaXQgVGVzdCBD 5152 ZXJ0aWZpY2F0ZSBBdXRob3JpdHmCAQAwDAYDVR0TBAUwAwEB/zANBgkqhkiG9w0B 5153 AQUFAAOBgQCWbRvv1ZGTRXxbH8/EqkdSCzSoUPrs+rQqR0xdQac9wNY/nlZbkR3O 5154 qAezG6Sfmklvf+DOg5RxQq/+Y6I03LRepc7KeVDpaplMFGnpfKsibETMipwzayNQ 5155 QgUf4cKBiF+65Ue7hZuDJa2EMv8qW4twEhGDYclpFU9YozyS1OhvUg== 5156 5157 YmFkIGNlcnQK 5158 https://example.org 5159 https://example.net 5160 false 5162 5163 5164 5165 20 5166 5167 5168 false 5169 false 5170 5171 400 5172 30 5173 true 5174 password 5175 4000 5176 30 5177 TLS 5178 47112162e84c69ba 5179 47112162e84c69ba 5180 6eba45d31a900c06 5181 6ebc45d31a900c06 5182 6ebc45d31a900ca6 5184 foo 5186 5187 urn:ietf:params:xml:ns:p2p:config-ext1 5188 5190 5191 5192 5193 SINGLE 5194 USER-MATCH 5195 1 5196 100 5197 5198 5199 VGhpcyBpcyBub3QgcmlnaHQhCg== 5200 5201 5202 5203 5204 ARRAY 5205 NODE-MULTIPLE 5206 3 5207 22 5208 4 5209 1 5210 5211 5212 5213 VGhpcyBpcyBub3QgcmlnaHQhCg== 5214 5216 5217 5218 5219 VGhpcyBpcyBub3QgcmlnaHQhCg== 5221 5222 5223 VGhpcyBpcyBub3QgcmlnaHQhCg== 5225 5227 The file MUST be a well formed XML document and it SHOULD contain an 5228 encoding declaration in the XML declaration. The file MUST use the 5229 UTF-8 character encoding. The namespace for the elements defined in 5230 this specification is urn:ietf:params:xml:ns:p2p:config-base and 5231 urn:ietf:params:xml:ns:p2p:config-chord". 5233 The file can contain multiple "configuration" elements where each one 5234 contains the configuration information for a different overlay. Each 5235 configuration element may be followed by signature elements that 5236 provides a signature over the preceding configuration element. Each 5237 configuration element has the following attributes: 5239 instance-name: name of the overlay 5240 expiration: time in the future at which this overlay configuration 5241 is no longer valid. The node SHOULD retrieve a new copy of the 5242 configuration at a randomly selected time that is before the 5243 expiration time. Note that if the certificates expire before a 5244 new configuration is retried, the node will not be able to 5245 validate the configuration file. 5246 sequence: a monotonically increasing sequence number between 1 and 5247 2^16-2 5249 Inside each overlay element, the following elements can occur: 5251 topology-plugin This element defines the overlay algorithm being 5252 used. If missing the default is "CHORD-RELOAD". 5253 node-id-length This element contains the length of a NodeId 5254 (NodeIdLength) in bytes. This value MUST be between 16 (128 bits) 5255 and 20 (160 bits). If this element is not present, the default of 5256 16 is used. 5257 root-cert This element contains a base-64 encoded X.509v3 5258 certificate that is a root trust anchor used to sign all 5259 certificates in this overlay. There can be more than one root- 5260 cert element. 5262 enrollment-server This element contains the URL at which the 5263 enrollment server can be reached in a "url" element. This URL 5264 MUST be of type "https:". More than one enrollment-server element 5265 may be present. 5266 self-signed-permitted This element indicates whether self-signed 5267 certificates are permitted. If it is set to "true", then self- 5268 signed certificates are allowed, in which case the enrollment- 5269 server and root-cert elements may be absent. Otherwise, it SHOULD 5270 be absent, but MAY be set to "false". This element also contains 5271 an attribute "digest" which indicates the digest to be used to 5272 compute the Node-ID. Valid values for this parameter are "sha1" 5273 and "sha256" representing SHA-1 [RFC3174] and SHA-256 [RFC6234] 5274 respectively. Implementations MUST support both of these 5275 algorithms. 5276 bootstrap-node This element represents the address of one of the 5277 bootstrap nodes. It has an attribute called "address" that 5278 represents the IP address (either IPv4 or IPv6, since they can be 5279 distinguished) and an optional attribute called "port" that 5280 represents the port and defaults to 6084. The IP address is in 5281 typical hexadecimal form using standard period and colon 5282 separators as specified in [RFC5952]. More than one bootstrap- 5283 peer element may be present. 5284 turn-density This element is a positive integer that represents the 5285 approximate reciprocal of density of nodes that can act as TURN 5286 servers. For example, if 5% of the nodes can act as TURN servers, 5287 this would be set to 20. If it is not present, the default value 5288 is 1. If there are no TURN servers in the overlay, it is set to 5289 zero. 5290 multicast-bootstrap This element represents the address of a 5291 multicast, broadcast, or anycast address and port that may be used 5292 for bootstrap. Nodes SHOULD listen on the address. It has an 5293 attributed called "address" that represents the IP address and an 5294 optional attribute called "port" that represents the port and 5295 defaults to 6084. More than one "multicast-bootstrap" element may 5296 be present. 5297 clients-permitted This element represents whether clients are 5298 permitted or whether all nodes must be peers. If it is set to 5299 "true" or absent, this indicates that clients are permitted. If 5300 it is set to "false" then nodes are not allowed to remain clients 5301 after the initial join. There is currently no way for the overlay 5302 to enforce this. 5303 no-ice This element represents whether nodes are required to use 5304 the "No-ICE" Overlay Link protocols in this overlay. If it is 5305 absent, it is treated as if it were set to "false". 5307 chord-update-interval The update frequency for the Chord-reload 5308 topology plugin (see Section 9). 5309 chord-ping-interval The ping frequency for the Chord-reload 5310 topology plugin (see Section 9). 5311 chord-reactive Whether reactive recovery should be used for this 5312 overlay. Set to "true" or "false". Default if missing is "true". 5313 (see Section 9). 5314 shared-secret If shared secret mode is used, this contains the 5315 shared secret. 5316 max-message-size Maximum size in bytes of any message in the 5317 overlay. If this value is not present, the default is 5000. 5318 initial-ttl Initial default TTL (time to live, see Section 5.3.2) 5319 for messages. If this value is not present, the default is 100. 5320 overlay-link-protocol Indicates a permissible overlay link protocol 5321 (see Section 5.6.1 for requirements for such protocols). An 5322 arbitrary number of these elements may appear. If none appear, 5323 then this implies the default value, "TLS", which refers to the 5324 use of TLS and DTLS. If one or more elements appear, then no 5325 default value applies. 5326 kind-signer This contains a single Node-ID in hexadecimal and 5327 indicates that the certificate with this Node-ID is allowed to 5328 sign kinds. Identifying kind-signer by Node-ID instead of 5329 certificate allows the use of short lived certificates without 5330 constantly having to provide an updated configuration file. 5331 configuration-signer This contains a single Node-ID in hexadecimal 5332 and indicates that the certificate with this Node-ID is allowed to 5333 sign configurations for this instance-name. Identifying the 5334 signer by Node-ID instead of certificate allows the use of short 5335 lived certificates without constantly having to provide an updated 5336 configuration file. 5337 bad-node This contains a single Node-ID in hexadecimal and 5338 indicates that the certificate with this Node-ID MUST NOT be 5339 considered valid. This allows certificate revocation. An 5340 arbitrary number of these elements can be provided. Note that 5341 because certificates may expire, bad-node entries need only be 5342 present for the lifetime of the certificate. Technically 5343 speaking, bad node-ids may be reused once their certificates have 5344 expired, the requirement for node-ids to be pseudo randomly 5345 generated gives this event a vanishing probability. 5346 mandatory-extension This element contains the name of an XML 5347 namespace that a node joining the overlay MUST support. The 5348 presence of a mandatory-extension element does not require the 5349 extension to be used in the current configuration file, but can 5350 indicate that it may be used in the future. Note that the 5351 namespace is case-sensitive, as specified in [w3c-xml-namespaces] 5352 Section 2.3. More than one mandatory-extension element may be 5353 present. 5355 Inside each overlay element, the required-kinds elements can also 5356 occur. This element indicates the kinds that members must support 5357 and contains multiple kind-block elements that each define a single 5358 kind that MUST be supported by nodes in the overlay. Each kind-block 5359 consists of a single kind element and a kind-signature. The kind 5360 element defines the kind. The kind-signature is the signature 5361 computed over the kind element. 5363 Each kind has either an id attribute or a name attribute. The name 5364 attribute is a string representing the kind (the name registered to 5365 IANA) while the id is an integer kind-id allocated out of private 5366 space. 5368 In addition, the kind element contains the following elements: 5369 max-count: the maximum number of values which members of the overlay 5370 must support. 5371 data-model: the data model to be used. 5372 max-size: the maximum size of individual values. 5373 access-control: the access control model to be used. 5374 max-node-multiple: This is optional and only used when the access 5375 control is NODE-MULTIPLE. This indicates the maximum value for 5376 the i counter. This is an integer greater than 0. 5378 All of the non optional values MUST be provided. If the kind is 5379 registered with IANA, the data-model and access-control elements MUST 5380 match those in the kind registration, and clients MUST ignore them in 5381 favor of the IANA versions. Multiple required-kinds elements MAY be 5382 present. 5384 The kind-block element also MUST contain a "kind-signature" element. 5385 This signature is computed across the kind from the beginning of the 5386 first < of the kind to the end of the last > of the kind in the same 5387 way as the signature element described later in this section. 5389 The configuration file is a binary file and cannot be changed - 5390 including whitespace changes - or the signature will break. The 5391 signature is computed by taking each configuration element and 5392 starting from, and including, the first < at the start of 5393 up to and including the > in and 5394 treating this as a binary blob that is signed using the standard 5395 SecurityBlock defined in Section 5.3.4. The SecurityBlock is base 64 5396 encoded using the base64 alphabet from RFC[RFC4648] and put in the 5397 signature element following the configuration object in the 5398 configuration file. 5400 When a node receives a new configuration file, it MUST change its 5401 configuration to meet the new requirements. This may require the 5402 node to exit the DHT and re-join. If a node is not capable of 5403 supporting the new requirements, it MUST exit the overlay. If some 5404 information about a particular kind changes from what the node 5405 previously knew about the kind (for example the max size), the new 5406 information in the configuration files overrides any previously 5407 learned information. If any kind data was signed by a node that is 5408 no longer allowed to sign kinds, that kind MUST be discarded along 5409 with any stored information of that kind. Note that forcing an 5410 avalanche restart of the overlay with a configuration change that 5411 requires re-joining the overlay may result in serious performance 5412 problems, including total collapse of the network if configuration 5413 parameters are not properly considered. Such an event may be 5414 necessary in case of a compromised CA or similar problem, but for 5415 large overlays should be avoided in almost all circumstances. 5417 10.1.1. Relax NG Grammar 5419 The grammar for the configuration data is: 5421 namespace chord = "urn:ietf:params:xml:ns:p2p:config-chord" 5422 namespace local = "" 5423 default namespace p2pcf = "urn:ietf:params:xml:ns:p2p:config-base" 5424 namespace rng = "http://relaxng.org/ns/structure/1.0" 5426 anything = 5427 (element * { anything } 5428 | attribute * { text } 5429 | text)* 5431 foreign-elements = element * - (p2pcf:* | local:* | chord:*) 5432 { anything }* 5433 foreign-attributes = attribute * - (p2pcf:*|local:*|chord:*) 5434 { text }* 5435 foreign-nodes = (foreign-attributes | foreign-elements)* 5437 start = element p2pcf:overlay { 5438 overlay-element 5439 } 5441 overlay-element &= element configuration { 5442 attribute instance-name { xsd:string }, 5443 attribute expiration { xsd:dateTime }?, 5444 attribute sequence { xsd:long }?, 5445 foreign-attributes*, 5446 parameter 5447 }+ 5448 overlay-element &= element signature { 5449 attribute algorithm { signature-algorithm-type }?, 5450 xsd:base64Binary 5452 }* 5454 signature-algorithm-type |= "rsa-sha1" 5455 signature-algorithm-type |= xsd:string # signature alg extensions 5457 parameter &= element topology-plugin { topology-plugin-type }? 5458 topology-plugin-type |= xsd:string # topo plugin extensions 5459 parameter &= element max-message-size { xsd:unsignedInt }? 5460 parameter &= element initial-ttl { xsd:int }? 5461 parameter &= element root-cert { xsd:base64Binary }* 5462 parameter &= element required-kinds { kind-block* }? 5463 parameter &= element enrollment-server { xsd:anyURI }* 5464 parameter &= element kind-signer { xsd:string }* 5465 parameter &= element configuration-signer { xsd:string }* 5466 parameter &= element bad-node { xsd:string }* 5467 parameter &= element no-ice { xsd:boolean }? 5468 parameter &= element shared-secret { xsd:string }? 5469 parameter &= element overlay-link-protocol { xsd:string }* 5470 parameter &= element clients-permitted { xsd:boolean }? 5471 parameter &= element turn-density { xsd:unsignedByte }? 5472 parameter &= element node-id-length { xsd:int }? 5473 parameter &= element mandatory-extension { xsd:string }* 5474 parameter &= foreign-elements* 5476 parameter &= 5477 element self-signed-permitted { 5478 attribute digest { self-signed-digest-type }, 5479 xsd:boolean 5480 }? 5481 self-signed-digest-type |= "sha1" 5482 self-signed-digest-type |= xsd:string # signature digest extensions 5484 parameter &= element bootstrap-node { 5485 attribute address { xsd:string }, 5486 attribute port { xsd:int }? 5487 }* 5489 parameter &= element multicast-bootstrap { 5490 attribute address { xsd:string }, 5491 attribute port { xsd:int }? 5492 }* 5494 kind-block = element kind-block { 5495 element kind { 5496 ( attribute name { kind-names } 5497 | attribute id { xsd:unsignedInt } ), 5498 kind-parameter 5499 } & 5500 element kind-signature { 5501 attribute algorithm { signature-algorithm-type }?, 5502 xsd:base64Binary 5503 }? 5504 } 5506 kind-parameter &= element max-count { xsd:int } 5507 kind-parameter &= element max-size { xsd:int } 5508 kind-parameter &= element max-node-multiple { xsd:int }? 5510 kind-parameter &= element data-model { data-model-type } 5511 data-model-type |= "SINGLE" 5512 data-model-type |= "ARRAY" 5513 data-model-type |= "DICTIONARY" 5514 data-model-type |= xsd:string # data model extensions 5516 kind-parameter &= element access-control { access-control-type } 5517 access-control-type |= "USER-MATCH" 5518 access-control-type |= "NODE-MATCH" 5519 access-control-type |= "USER-NODE-MATCH" 5520 access-control-type |= "NODE-MULTIPLE" 5521 access-control-type |= xsd:string # access control extensions 5523 kind-parameter &= foreign-elements* 5525 kind-names |= "TURN-SERVICE" 5526 kind-names |= "CERTIFICATE_BY_NODE" 5527 kind-names |= "CERTIFICATE_BY_USER" 5528 kind-names |= xsd:string # kind extensions 5530 # Chord specific parameters 5531 topology-plugin-type |= "CHORD-RELOAD" 5532 parameter &= element chord:chord-ping-interval { xsd:int }? 5533 parameter &= element chord:chord-update-interval { xsd:int }? 5534 parameter &= element chord:chord-reactive { xsd:boolean }? 5536 10.2. Discovery Through Configuration Server 5538 When a node first enrolls in a new overlay, it starts with a 5539 discovery process to find a configuration server. 5541 The node first determines the overlay name. This value is provided 5542 by the user or some other out of band provisioning mechanism. The 5543 out of band mechanisms may also provide an optional URL for the 5544 configuration server. If a URL for the configuration server is not 5545 provided, the node MUST do a DNS SRV query using a Service name of 5546 "p2psip-enroll" and a protocol of TCP to find a configuration server 5547 and form the URL by appending a path of "/.well-known/p2psip-enroll" 5548 to the overlay name. This uses the "well known URI" framework 5549 defined in [RFC5785]. For example, if the overlay name was 5550 example.com, the URL would be 5551 "https://example.com//.well-known/p2psip-enroll". 5553 Once an address and URL for the configuration server is determined, 5554 the peer forms an HTTPS connection to that IP address. The 5555 certificate MUST match the overlay name as described in [RFC2818]. 5556 Then the node MUST fetch a new copy of the configuration file. To do 5557 this, the peer performs a GET to the URL. The result of the HTTP GET 5558 is an XML configuration file described above, which replaces any 5559 previously learned configuration file for this overlay. 5561 For overlays that do not use a configuration server, nodes obtain the 5562 configuration information needed to join the overlay through some out 5563 of band approach such an XML configuration file sent over email. 5565 10.3. Credentials 5567 If the configuration document contains a enrollment-server element, 5568 credentials are required to join the Overlay Instance. A peer which 5569 does not yet have credentials MUST contact the enrollment server to 5570 acquire them. 5572 RELOAD defines its own trivial certificate request protocol. We 5573 would have liked to have used an existing protocol but were concerned 5574 about the implementation burden of even the simplest of those 5575 protocols, such as [RFC5272] and [RFC5273]. Our objective was to 5576 have a protocol which could be easily implemented in a Web server 5577 which the operator did not control (e.g., in a hosted service) and 5578 was compatible with the existing certificate handling tooling as used 5579 with the Web certificate infrastructure. This means accepting bare 5580 PKCS#10 requests and returning a single bare X.509 certificate. 5581 Although the MIME types for these objects are defined, none of the 5582 existing protocols support exactly this model. 5584 The certificate request protocol is performed over HTTPS. The 5585 request is an HTTP POST with the following properties: 5587 o If authentication is required, there is an URL parameter of 5588 "password" and "username" containing the user's name and password 5589 in the clear (hence the need for HTTPS) 5590 o If more than one Node-ID is required, there is an URL parameter of 5591 "nodeids" containing the number of Node-IDs required. 5592 o The body is of content type "application/pkcs10", as defined in 5593 [RFC2311]. 5595 o The Accept header contains the type "application/pkix-cert", 5596 indicating the type that is expected in the response. 5598 The enrollment server MUST authenticate the request using the 5599 provided user name and password. If the authentication succeeds and 5600 the requested user name is acceptable, the server generates and 5601 returns a certificate. The SubjectAltName field in the certificate 5602 contains the following values: 5604 o One or more Node-IDs which MUST be cryptographically random 5605 [RFC4086]. Each MUST be chosen by the enrollment server in such a 5606 way that they are unpredictable to the requesting user. E.g., the 5607 user MUST NOT be informed of potential (random) Node-IDs prior to 5608 authenticating. Each is placed in the subjectAltName using the 5609 uniformResourceIdentifier type and MUST contain RELOAD URIs as 5610 described in Section 13.15 and MUST contain a Destination list 5611 with a single entry of type "node_id". 5612 o A single name this user is allowed to use in the overlay, using 5613 type rfc822Name. 5615 The certificate is returned as type "application/pkix-cert" as 5616 defined in [RFC2585], with an HTTP status code of 200 OK. 5617 Certificate processing errors should be treated as HTTP errors and 5618 have appropriate HTTP status codes. 5620 The client MUST check that the certificate returned was signed by one 5621 of the certificates received in the "root-cert" list of the overlay 5622 configuration data. The node then reads the certificate to find the 5623 Node-IDs it can use. 5625 10.3.1. Self-Generated Credentials 5627 If the "self-signed-permitted" element is present in the 5628 configuration and set to "true", then a node MUST generate its own 5629 self-signed certificate to join the overlay. The self-signed 5630 certificate MAY contain any user name of the users choice. 5632 The Node-ID MUST be computed by applying the digest specified in the 5633 self-signed-permitted element to the DER representation of the user's 5634 public key (more specifically the subjectPublicKeyInfo) and taking 5635 the high order bits. When accepting a self-signed certificate, nodes 5636 MUST check that the Node-ID and public keys match. This prevents 5637 Node-ID theft. 5639 Once the node has constructed a self-signed certificate, it MAY join 5640 the overlay. Before storing its certificate in the overlay 5641 (Section 7) it SHOULD look to see if the user name is already taken 5642 and if so choose another user name. Note that this only provides 5643 protection against accidental name collisions. Name theft is still 5644 possible. If protection against name theft is desired, then the 5645 enrollment service must be used. 5647 10.4. Searching for a Bootstrap Node 5649 If no cached bootstrap nodes are available and the configuration file 5650 has an multicast-bootstrap element, then the node SHOULD send a Ping 5651 request over UDP to the address and port found to each multicast- 5652 bootstrap element found in the configuration document. This MAY be a 5653 multicast, broadcast, or anycast address. The Ping should use the 5654 wildcard Node-ID as the destination Node-ID. 5656 The responder node that receives the Ping request SHOULD check that 5657 the overlay name is correct and that the requester peer sending the 5658 request has appropriate credentials for the overlay before responding 5659 to the Ping request even if the response is only an error. 5661 10.5. Contacting a Bootstrap Node 5663 In order to join the overlay, the joining node MUST contact a node in 5664 the overlay. Typically this means contacting the bootstrap nodes, 5665 since they are reachable by the local peer or have public IP 5666 addresses. If the joining node has cached a list of peers it has 5667 previously been connected with in this overlay, as an optimization it 5668 MAY attempt to use one or more of them as bootstrap nodes before 5669 falling back to the bootstrap nodes listed in the configuration file. 5671 When contacting a bootstrap node, the joining node first forms the 5672 DTLS or TLS connection to the bootstrap node and then sends an Attach 5673 request over this connection with the destination Node-ID set to the 5674 joining node's Node-ID. 5676 When the requester node finally does receive a response from some 5677 responding node, it can note the Node-ID in the response and use this 5678 Node-ID to start sending requests to join the Overlay Instance as 5679 described in Section 5.4. 5681 After a node has successfully joined the overlay network, it will 5682 have direct connections to several peers. Some MAY be added to the 5683 cached bootstrap nodes list and used in future boots. Peers that are 5684 not directly connected MUST NOT be cached. The suggested number of 5685 peers to cache is 10. Algorithms for determining which peers to 5686 cache are beyond the scope of this specification. 5688 11. Message Flow Example 5690 The following abbreviation are used in the message flow diagrams: JP 5691 = joining peer, AP = admitting peer, NP = next peer after the AP, NNP 5692 = next next peer which is the peer after NP, PP = previous peer 5693 before the AP, PPP = previous previous peer which is the peer before 5694 the PP, BP = bootstrap peer. 5696 In the following example, we assume that JP has formed a connection 5697 to one of the bootstrap nodes. JP then sends an Attach through that 5698 peer to a resource ID of itself (JP). It gets routed to the 5699 admitting peer (AP) because JP is not yet part of the overlay. When 5700 AP responds, JP and AP use ICE to set up a connection and then set up 5701 TLS. Once AP has connected to JP, AP sends to JP an Update to 5702 populate its Routing Table. The following example shows the Update 5703 happening after the TLS connection is formed but it could also happen 5704 before in which case the Update would often be routed through other 5705 nodes. 5707 JP PPP PP AP NP NNP BP 5708 | | | | | | | 5709 | | | | | | | 5710 | | | | | | | 5711 |Attach Dest=JP | | | | | 5712 |---------------------------------------------------------->| 5713 | | | | | | | 5714 | | | | | | | 5715 | | |Attach Dest=JP | | | 5716 | | |<--------------------------------------| 5717 | | | | | | | 5718 | | | | | | | 5719 | | |Attach Dest=JP | | | 5720 | | |-------->| | | | 5721 | | | | | | | 5722 | | | | | | | 5723 | | |AttachAns | | | 5724 | | |<--------| | | | 5725 | | | | | | | 5726 | | | | | | | 5727 | | |AttachAns | | | 5728 | | |-------------------------------------->| 5729 | | | | | | | 5730 | | | | | | | 5731 |AttachAns | | | | | 5732 |<----------------------------------------------------------| 5733 | | | | | | | 5734 | | | | | | | 5735 |TLS | | | | | | 5736 |.............................| | | | 5737 | | | | | | | 5738 | | | | | | | 5739 | | | | | | | 5740 |Update | | | | | | 5741 |<----------------------------| | | | 5742 | | | | | | | 5743 | | | | | | | 5744 |UpdateAns| | | | | | 5745 |---------------------------->| | | | 5746 | | | | | | | 5747 | | | | | | | 5748 | | | | | | | 5750 The JP then forms connections to the appropriate neighbors, such as 5751 NP, by sending an Attach which gets routed via other nodes. When NP 5752 responds, JP and NP use ICE and TLS to set up a connection. 5754 JP PPP PP AP NP NNP BP 5755 | | | | | | | 5756 | | | | | | | 5757 | | | | | | | 5758 |Attach NP | | | | | 5759 |---------------------------->| | | | 5760 | | | | | | | 5761 | | | | | | | 5762 | | | |Attach NP| | | 5763 | | | |-------->| | | 5764 | | | | | | | 5765 | | | | | | | 5766 | | | |AttachAns| | | 5767 | | | |<--------| | | 5768 | | | | | | | 5769 | | | | | | | 5770 |AttachAns | | | | | 5771 |<----------------------------| | | | 5772 | | | | | | | 5773 | | | | | | | 5774 |Attach | | | | | | 5775 |-------------------------------------->| | | 5776 | | | | | | | 5777 | | | | | | | 5778 |TLS | | | | | | 5779 |.......................................| | | 5780 | | | | | | | 5781 | | | | | | | 5782 | | | | | | | 5783 | | | | | | | 5785 JP also needs to populate its finger table (for Chord). It issues an 5786 Attach to a variety of locations around the overlay. The diagram 5787 below shows it sending an Attach halfway around the Chord ring to the 5788 JP + 2^127. 5790 JP NP XX TP 5791 | | | | 5792 | | | | 5793 | | | | 5794 |Attach JP+2<<126 | | 5795 |-------->| | | 5796 | | | | 5797 | | | | 5798 | |Attach JP+2<<126 | 5799 | |-------->| | 5800 | | | | 5801 | | | | 5802 | | |Attach JP+2<<126 5803 | | |-------->| 5804 | | | | 5805 | | | | 5806 | | |AttachAns| 5807 | | |<--------| 5808 | | | | 5809 | | | | 5810 | |AttachAns| | 5811 | |<--------| | 5812 | | | | 5813 | | | | 5814 |AttachAns| | | 5815 |<--------| | | 5816 | | | | 5817 | | | | 5818 |TLS | | | 5819 |.............................| 5820 | | | | 5821 | | | | 5822 | | | | 5823 | | | | 5825 Once JP has a reasonable set of connections, it is ready to take its 5826 place in the DHT. It does this by sending a Join to AP. AP does a 5827 series of Store requests to JP to store the data that JP will be 5828 responsible for. AP then sends JP an Update explicitly labeling JP 5829 as its predecessor. At this point, JP is part of the ring and 5830 responsible for a section of the overlay. AP can now forget any data 5831 which is assigned to JP and not AP. 5833 JP PPP PP AP NP NNP BP 5834 | | | | | | | 5835 | | | | | | | 5836 | | | | | | | 5837 |JoinReq | | | | | | 5838 |---------------------------->| | | | 5839 | | | | | | | 5840 | | | | | | | 5841 |JoinAns | | | | | | 5842 |<----------------------------| | | | 5843 | | | | | | | 5844 | | | | | | | 5845 |StoreReq Data A | | | | | 5846 |<----------------------------| | | | 5847 | | | | | | | 5848 | | | | | | | 5849 |StoreAns | | | | | | 5850 |---------------------------->| | | | 5851 | | | | | | | 5852 | | | | | | | 5853 |StoreReq Data B | | | | | 5854 |<----------------------------| | | | 5855 | | | | | | | 5856 | | | | | | | 5857 |StoreAns | | | | | | 5858 |---------------------------->| | | | 5859 | | | | | | | 5860 | | | | | | | 5861 |UpdateReq| | | | | | 5862 |<----------------------------| | | | 5863 | | | | | | | 5864 | | | | | | | 5865 |UpdateAns| | | | | | 5866 |---------------------------->| | | | 5867 | | | | | | | 5868 | | | | | | | 5869 | | | | | | | 5870 | | | | | | | 5872 In Chord, JP's neighbor table needs to contain its own predecessors. 5873 It couldn't connect to them previously because it did not yet know 5874 their addresses. However, now that it has received an Update from 5875 AP, it has AP's predecessors, which are also its own, so it sends 5876 Attaches to them. Below it is shown connecting to AP's closest 5877 predecessor, PP. 5879 JP PPP PP AP NP NNP BP 5880 | | | | | | | 5881 | | | | | | | 5882 | | | | | | | 5883 |Attach Dest=PP | | | | | 5884 |---------------------------->| | | | 5885 | | | | | | | 5886 | | | | | | | 5887 | | |Attach Dest=PP | | | 5888 | | |<--------| | | | 5889 | | | | | | | 5890 | | | | | | | 5891 | | |AttachAns| | | | 5892 | | |-------->| | | | 5893 | | | | | | | 5894 | | | | | | | 5895 |AttachAns| | | | | | 5896 |<----------------------------| | | | 5897 | | | | | | | 5898 | | | | | | | 5899 |TLS | | | | | | 5900 |...................| | | | | 5901 | | | | | | | 5902 | | | | | | | 5903 |UpdateReq| | | | | | 5904 |------------------>| | | | | 5905 | | | | | | | 5906 | | | | | | | 5907 |UpdateAns| | | | | | 5908 |<------------------| | | | | 5909 | | | | | | | 5910 | | | | | | | 5911 |UpdateReq| | | | | | 5912 |---------------------------->| | | | 5913 | | | | | | | 5914 | | | | | | | 5915 |UpdateAns| | | | | | 5916 |<----------------------------| | | | 5917 | | | | | | | 5918 | | | | | | | 5919 |UpdateReq| | | | | | 5920 |-------------------------------------->| | | 5921 | | | | | | | 5922 | | | | | | | 5923 |UpdateAns| | | | | | 5924 |<--------------------------------------| | | 5925 | | | | | | | 5926 | | | | | | | 5928 Finally, now that JP has a copy of all the data and is ready to route 5929 messages and receive requests, it sends Updates to everyone in its 5930 Routing Table to tell them it is ready to go. Below, it is shown 5931 sending such an update to TP. 5933 JP NP XX TP 5934 | | | | 5935 | | | | 5936 | | | | 5937 |Update | | | 5938 |---------------------------->| 5939 | | | | 5940 | | | | 5941 |UpdateAns| | | 5942 |<----------------------------| 5943 | | | | 5944 | | | | 5945 | | | | 5946 | | | | 5948 12. Security Considerations 5950 12.1. Overview 5952 RELOAD provides a generic storage service, albeit one designed to be 5953 useful for P2PSIP. In this section we discuss security issues that 5954 are likely to be relevant to any usage of RELOAD. More background 5955 information can be found in [RFC5765]. 5957 In any Overlay Instance, any given user depends on a number of peers 5958 with which they have no well-defined relationship except that they 5959 are fellow members of the Overlay Instance. In practice, these other 5960 nodes may be friendly, lazy, curious, or outright malicious. No 5961 security system can provide complete protection in an environment 5962 where most nodes are malicious. The goal of security in RELOAD is to 5963 provide strong security guarantees of some properties even in the 5964 face of a large number of malicious nodes and to allow the overlay to 5965 function correctly in the face of a modest number of malicious nodes. 5967 P2PSIP deployments require the ability to authenticate both peers and 5968 resources (users) without the active presence of a trusted entity in 5969 the system. We describe two mechanisms. The first mechanism is 5970 based on public key certificates and is suitable for general 5971 deployments. The second is an admission control mechanism based on 5972 an overlay-wide shared symmetric key. 5974 12.2. Attacks on P2P Overlays 5976 The two basic functions provided by overlay nodes are storage and 5977 routing: some node is responsible for storing a peer's data and for 5978 allowing a third peer to fetch this stored data. Other nodes are 5979 responsible for routing messages to and from the storing nodes. Each 5980 of these issues is covered in the following sections. 5982 P2P overlays are subject to attacks by subversive nodes that may 5983 attempt to disrupt routing, corrupt or remove user registrations, or 5984 eavesdrop on signaling. The certificate-based security algorithms we 5985 describe in this specification are intended to protect overlay 5986 routing and user registration information in RELOAD messages. 5988 To protect the signaling from attackers pretending to be valid peers 5989 (or peers other than themselves), the first requirement is to ensure 5990 that all messages are received from authorized members of the 5991 overlay. For this reason, RELOAD transports all messages over a 5992 secure channel (TLS and DTLS are defined in this document) which 5993 provides message integrity and authentication of the directly 5994 communicating peer. In addition, messages and data are digitally 5995 signed with the sender's private key, providing end-to-end security 5996 for communications. 5998 12.3. Certificate-based Security 6000 This specification stores users' registrations and possibly other 6001 data in an overlay network. This requires a solution to securing 6002 this data as well as securing, as well as possible, the routing in 6003 the overlay. Both types of security are based on requiring that 6004 every entity in the system (whether user or peer) authenticate 6005 cryptographically using an asymmetric key pair tied to a certificate. 6007 When a user enrolls in the Overlay Instance, they request or are 6008 assigned a unique name, such as "alice@dht.example.net". These names 6009 are unique and are meant to be chosen and used by humans much like a 6010 SIP Address of Record (AOR) or an email address. The user is also 6011 assigned one or more Node-IDs by the central enrollment authority. 6012 Both the name and the Node-ID are placed in the certificate, along 6013 with the user's public key. 6015 Each certificate enables an entity to act in two sorts of roles: 6017 o As a user, storing data at specific Resource-IDs in the Overlay 6018 Instance corresponding to the user name. 6019 o As a overlay peer with the Node-ID(s) listed in the certificate. 6021 Note that since only users of this Overlay Instance need to validate 6022 a certificate, this usage does not require a global PKI. Instead, 6023 certificates are signed by a central enrollment authority which acts 6024 as the certificate authority for the Overlay Instance. This 6025 authority signs each peer's certificate. Because each peer possesses 6026 the CA's certificate (which they receive on enrollment) they can 6027 verify the certificates of the other entities in the overlay without 6028 further communication. Because the certificates contain the user/ 6029 peer's public key, communications from the user/peer can be verified 6030 in turn. 6032 If self-signed certificates are used, then the security provided is 6033 significantly decreased, since attackers can mount Sybil attacks. In 6034 addition, attackers cannot trust the user names in certificates 6035 (though they can trust the Node-IDs because they are 6036 cryptographically verifiable). This scheme may be appropriate for 6037 some small deployments, such as a small office or an ad hoc overlay 6038 set up among participants in a meeting where all hosts on the network 6039 are trusted. Some additional security can be provided by using the 6040 shared secret admission control scheme as well. 6042 Because all stored data is signed by the owner of the data the 6043 storing peer can verify that the storer is authorized to perform a 6044 store at that Resource-ID and also allow any consumer of the data to 6045 verify the provenance and integrity of the data when it retrieves it. 6047 Note that RELOAD does not itself provide a revocation/status 6048 mechanism (though certificates may of course include OCSP responder 6049 information). Thus, certificate lifetimes should be chosen to 6050 balance the compromise window versus the cost of certificate renewal. 6051 Because RELOAD is already designed to operate in the face of some 6052 fraction of malicious peers, this form of compromise is not fatal. 6054 All implementations MUST implement certificate-based security. 6056 12.4. Shared-Secret Security 6058 RELOAD also supports a shared secret admission control scheme that 6059 relies on a single key that is shared among all members of the 6060 overlay. It is appropriate for small groups that wish to form a 6061 private network without complexity. In shared secret mode, all the 6062 peers share a single symmetric key which is used to key TLS-PSK 6063 [RFC4279] or TLS-SRP [RFC5054] mode. A peer which does not know the 6064 key cannot form TLS connections with any other peer and therefore 6065 cannot join the overlay. 6067 One natural approach to a shared-secret scheme is to use a user- 6068 entered password as the key. The difficulty with this is that in 6069 TLS-PSK mode, such keys are very susceptible to dictionary attacks. 6071 If passwords are used as the source of shared-keys, then TLS-SRP is a 6072 superior choice because it is not subject to dictionary attacks. 6074 12.5. Storage Security 6076 When certificate-based security is used in RELOAD, any given 6077 Resource-ID/Kind-ID pair is bound to some small set of certificates. 6078 In order to write data, the writer must prove possession of the 6079 private key for one of those certificates. Moreover, all data is 6080 stored, signed with the same private key that was used to authorize 6081 the storage. This set of rules makes questions of authorization and 6082 data integrity - which have historically been thorny for overlays - 6083 relatively simple. 6085 12.5.1. Authorization 6087 When a client wants to store some value, it first digitally signs the 6088 value with its own private key. It then sends a Store request that 6089 contains both the value and the signature towards the storing peer 6090 (which is defined by the Resource Name construction algorithm for 6091 that particular kind of value). 6093 When the storing peer receives the request, it must determine whether 6094 the storing client is authorized to store at this Resource-ID/Kind-ID 6095 pair. Determining this requires comparing the user's identity to the 6096 requirements of the access control model (see Section 6.3). If it 6097 satisfies those requirements the user is authorized to write, pending 6098 quota checks as described in the next section. 6100 For example, consider the certificate with the following properties: 6102 User name: alice@dht.example.com 6103 Node-ID: 013456789abcdef 6104 Serial: 1234 6106 If Alice wishes to Store a value of the "SIP Location" kind, the 6107 Resource Name will be the SIP AOR "sip:alice@dht.example.com". The 6108 Resource-ID will be determined by hashing the Resource Name. Because 6109 SIP Location uses the USER-NODE-MATCH policy, it first verifies that 6110 the user name in the certificate hashes to the requested Resource-ID. 6111 It then verifies that the Node-Id in the certificate matches the 6112 dictionary key being used for the store. If both of these checks 6113 succeed, the Store is authorized. Note that because the access 6114 control model is different for different kinds, the exact set of 6115 checks will vary. 6117 12.5.2. Distributed Quota 6119 Being a peer in an Overlay Instance carries with it the 6120 responsibility to store data for a given region of the Overlay 6121 Instance. However, allowing clients to store unlimited amounts of 6122 data would create unacceptable burdens on peers and would also enable 6123 trivial denial of service attacks. RELOAD addresses this issue by 6124 requiring configurations to define maximum sizes for each kind of 6125 stored data. Attempts to store values exceeding this size MUST be 6126 rejected (if peers are inconsistent about this, then strange 6127 artifacts will happen when the zone of responsibility shifts and a 6128 different peer becomes responsible for overlarge data). Because each 6129 Resource-ID/Kind-ID pair is bound to a small set of certificates, 6130 these size restrictions also create a distributed quota mechanism, 6131 with the quotas administered by the central configuration server. 6133 Allowing different kinds of data to have different size restrictions 6134 allows new usages the flexibility to define limits that fit their 6135 needs without requiring all usages to have expansive limits. 6137 12.5.3. Correctness 6139 Because each stored value is signed, it is trivial for any retrieving 6140 peer to verify the integrity of the stored value. Some more care 6141 needs to be taken to prevent version rollback attacks. Rollback 6142 attacks on storage are prevented by the use of store times and 6143 lifetime values in each store. A lifetime represents the latest time 6144 at which the data is valid and thus limits (though does not 6145 completely prevent) the ability of the storing node to perform a 6146 rollback attack on retrievers. In order to prevent a rollback attack 6147 at the time of the Store request, we require that storage times be 6148 monotonically increasing. Storing peers MUST reject Store requests 6149 with storage times smaller than or equal to those they are currently 6150 storing. In addition, a fetching node which receives a data value 6151 with a storage time older than the result of the previous fetch knows 6152 a rollback has occurred. 6154 12.5.4. Residual Attacks 6156 The mechanisms described here provides a high degree of security, but 6157 some attacks remain possible. Most simply, it is possible for 6158 storing nodes to refuse to store a value (i.e., reject any request). 6159 In addition, a storing node can deny knowledge of values which it has 6160 previously accepted. To some extent these attacks can be ameliorated 6161 by attempting to store to/retrieve from replicas, but a retrieving 6162 client does not know whether it should try this or not, since there 6163 is a cost to doing so. 6165 The certificate-based authentication scheme prevents a single peer 6166 from being able to forge data owned by other peers. Furthermore, 6167 although a subversive peer can refuse to return data resources for 6168 which it is responsible, it cannot return forged data because it 6169 cannot provide authentication for such registrations. Therefore 6170 parallel searches for redundant registrations can mitigate most of 6171 the effects of a compromised peer. The ultimate reliability of such 6172 an overlay is a statistical question based on the replication factor 6173 and the percentage of compromised peers. 6175 In addition, when a kind is multivalued (e.g., an array data model), 6176 the storing node can return only some subset of the values, thus 6177 biasing its responses. This can be countered by using single values 6178 rather than sets, but that makes coordination between multiple 6179 storing agents much more difficult. This is a trade off that must be 6180 made when designing any usage. 6182 12.6. Routing Security 6184 Because the storage security system guarantees (within limits) the 6185 integrity of the stored data, routing security focuses on stopping 6186 the attacker from performing a DOS attack that misroutes requests in 6187 the overlay. There are a few obvious observations to make about 6188 this. First, it is easy to ensure that an attacker is at least a 6189 valid peer in the Overlay Instance. Second, this is a DOS attack 6190 only. Third, if a large percentage of the peers on the Overlay 6191 Instance are controlled by the attacker, it is probably impossible to 6192 perfectly secure against this. 6194 12.6.1. Background 6196 In general, attacks on DHT routing are mounted by the attacker 6197 arranging to route traffic through one or two nodes it controls. In 6198 the Eclipse attack [Eclipse] the attacker tampers with messages to 6199 and from nodes for which it is on-path with respect to a given victim 6200 node. This allows it to pretend to be all the nodes that are 6201 reachable through it. In the Sybil attack [Sybil], the attacker 6202 registers a large number of nodes and is therefore able to capture a 6203 large amount of the traffic through the DHT. 6205 Both the Eclipse and Sybil attacks require the attacker to be able to 6206 exercise control over her Node-IDs. The Sybil attack requires the 6207 creation of a large number of peers. The Eclipse attack requires 6208 that the attacker be able to impersonate specific peers. In both 6209 cases, these attacks are limited by the use of centralized, 6210 certificate-based admission control. 6212 12.6.2. Admissions Control 6214 Admission to a RELOAD Overlay Instance is controlled by requiring 6215 that each peer have a certificate containing its Node-Id. The 6216 requirement to have a certificate is enforced by using certificate- 6217 based mutual authentication on each connection. (Note: the 6218 following only applies when self-signed certificates are not used.) 6219 Whenever a peer connects to another peer, each side automatically 6220 checks that the other has a suitable certificate. These Node-Ids are 6221 randomly assigned by the central enrollment server. This has two 6222 benefits: 6224 o It allows the enrollment server to limit the number of Node-IDs 6225 issued to any individual user. 6226 o It prevents the attacker from choosing specific Node-Ids. 6228 The first property allows protection against Sybil attacks (provided 6229 the enrollment server uses strict rate limiting policies). The 6230 second property deters but does not completely prevent Eclipse 6231 attacks. Because an Eclipse attacker must impersonate peers on the 6232 other side of the attacker, he must have a certificate for suitable 6233 Node-Ids, which requires him to repeatedly query the enrollment 6234 server for new certificates, which will match only by chance. From 6235 the attacker's perspective, the difficulty is that if he only has a 6236 small number of certificates, the region of the Overlay Instance he 6237 is impersonating appears to be very sparsely populated by comparison 6238 to the victim's local region. 6240 12.6.3. Peer Identification and Authentication 6242 In general, whenever a peer engages in overlay activity that might 6243 affect the routing table it must establish its identity. This 6244 happens in two ways. First, whenever a peer establishes a direct 6245 connection to another peer it authenticates via certificate-based 6246 mutual authentication. All messages between peers are sent over this 6247 protected channel and therefore the peers can verify the data origin 6248 of the last hop peer for requests and responses without further 6249 cryptography. 6251 In some situations, however, it is desirable to be able to establish 6252 the identity of a peer with whom one is not directly connected. The 6253 most natural case is when a peer Updates its state. At this point, 6254 other peers may need to update their view of the overlay structure, 6255 but they need to verify that the Update message came from the actual 6256 peer rather than from an attacker. To prevent this, all overlay 6257 routing messages are signed by the peer that generated them. 6259 Replay is typically prevented for messages that impact the topology 6260 of the overlay by having the information come directly, or be 6261 verified by, the nodes that claimed to have generated the update. 6262 Data storage replay detection is done by signing time of the node 6263 that generated the signature on the store request thus providing a 6264 time based replay protection but the time synchronization is only 6265 needed between peers that can write to the same location. 6267 12.6.4. Protecting the Signaling 6269 The goal here is to stop an attacker from knowing who is signaling 6270 what to whom. An attacker is unlikely to be able to observe the 6271 activities of a specific individual given the randomization of IDs 6272 and routing based on the present peers discussed above. Furthermore, 6273 because messages can be routed using only the header information, the 6274 actual body of the RELOAD message can be encrypted during 6275 transmission. 6277 There are two lines of defense here. The first is the use of TLS or 6278 DTLS for each communications link between peers. This provides 6279 protection against attackers who are not members of the overlay. The 6280 second line of defense is to digitally sign each message. This 6281 prevents adversarial peers from modifying messages in flight, even if 6282 they are on the routing path. 6284 12.6.5. Residual Attacks 6286 The routing security mechanisms in RELOAD are designed to contain 6287 rather than eliminate attacks on routing. It is still possible for 6288 an attacker to mount a variety of attacks. In particular, if an 6289 attacker is able to take up a position on the overlay routing between 6290 A and B it can make it appear as if B does not exist or is 6291 disconnected. It can also advertise false network metrics in an 6292 attempt to reroute traffic. However, these are primarily DOS 6293 attacks. 6295 The certificate-based security scheme secures the namespace, but if 6296 an individual peer is compromised or if an attacker obtains a 6297 certificate from the CA, then a number of subversive peers can still 6298 appear in the overlay. While these peers cannot falsify responses to 6299 resource queries, they can respond with error messages, effecting a 6300 DoS attack on the resource registration. They can also subvert 6301 routing to other compromised peers. To defend against such attacks, 6302 a resource search must still consist of parallel searches for 6303 replicated registrations. 6305 13. IANA Considerations 6307 This section contains the new code points registered by this 6308 document. [NOTE TO IANA/RFC-EDITOR: Please replace RFC-AAAA with 6309 the RFC number for this specification in the following list.] 6311 13.1. Well-Known URI Registration 6313 IANA will make the following "Well Known URI" registration as 6314 described in [RFC5785]: 6316 [[Note to RFC Editor - this paragraph can be removed before 6317 publication. ]] A review request was sent to 6318 wellknown-uri-review@ietf.org on October 12, 2010. 6320 +----------------------------+----------------------+ 6321 | URI suffix: | p2psip-enroll | 6322 | Change controller: | IETF | 6323 | Specification document(s): | [RFC-AAAA] | 6324 | Related information: | None | 6325 +----------------------------+----------------------+ 6327 13.2. Port Registrations 6329 [[Note to RFC Editor - this paragraph can be removed before 6330 publication. ]] IANA has already allocated a TCP port for the main 6331 peer to peer protocol. This port has the name p2p-sip and the port 6332 number of 6084. IANA needs to update this registration to be defined 6333 for UDP as well as TCP. 6335 IANA will make the following port registration: 6337 +------------------------------+------------------------------------+ 6338 | Registration Technical | Cullen Jennings | 6339 | Contact | | 6340 | Registration Owner | IETF | 6341 | Transport Protocol | TCP & UDP | 6342 | Port Number | 6084 | 6343 | Service Name | p2psip-enroll | 6344 | Description | Peer to Peer Infrastructure | 6345 | | Enrollment | 6346 | Reference | [RFC-AAAA] | 6347 +------------------------------+------------------------------------+ 6349 13.3. Overlay Algorithm Types 6351 IANA SHALL create a "RELOAD Overlay Algorithm Type" Registry. 6352 Entries in this registry are strings denoting the names of overlay 6353 algorithms. The registration policy for this registry is RFC 5226 6354 IETF Review. The initial contents of this registry are: 6356 +----------------+----------+ 6357 | Algorithm Name | RFC | 6358 +----------------+----------+ 6359 | CHORD-RELOAD | RFC-AAAA | 6360 +----------------+----------+ 6362 13.4. Access Control Policies 6364 IANA SHALL create a "RELOAD Access Control Policy" Registry. Entries 6365 in this registry are strings denoting access control policies, as 6366 described in Section 6.3. New entries in this registry SHALL be 6367 registered via RFC 5226 Standards Action. The initial contents of 6368 this registry are: 6370 +-----------------+----------+ 6371 | Access Policy | RFC | 6372 +-----------------+----------+ 6373 | USER-MATCH | RFC-AAAA | 6374 | NODE-MATCH | RFC-AAAA | 6375 | USER-NODE-MATCH | RFC-AAAA | 6376 | NODE-MULTIPLE | RFC-AAAA | 6377 +-----------------+----------+ 6379 13.5. Application-ID 6381 IANA SHALL create a "RELOAD Application-ID" Registry. Entries in 6382 this registry are 16-bit integers denoting application kinds. Code 6383 points in the range 0x0001 to 0x7fff SHALL be registered via RFC 5226 6384 Standards Action. Code points in the range 0x8000 to 0xf000 SHALL be 6385 registered via RFC 5226 Expert Review. Code points in the range 6386 0xf001 to 0xfffe are reserved for private use. The initial contents 6387 of this registry are: 6389 +-------------+----------------+-------------------------------+ 6390 | Application | Application-ID | Specification | 6391 +-------------+----------------+-------------------------------+ 6392 | INVALID | 0 | RFC-AAAA | 6393 | SIP | 5060 | Reserved for use by SIP Usage | 6394 | SIP | 5061 | Reserved for use by SIP Usage | 6395 | Reserved | 0xffff | RFC-AAAA | 6396 +-------------+----------------+-------------------------------+ 6398 13.6. Data Kind-ID 6400 IANA SHALL create a "RELOAD Data Kind-ID" Registry. Entries in this 6401 registry are 32-bit integers denoting data kinds, as described in 6402 Section 4.2. Code points in the range 0x00000001 to 0x7fffffff SHALL 6403 be registered via RFC 5226 Standards Action. Code points in the 6404 range 0x8000000 to 0xf0000000 SHALL be registered via RFC 5226 Expert 6405 Review. Code points in the range 0xf0000001 to 0xfffffffe are 6406 reserved for private use via the kind description mechanism described 6407 in Section 10. The initial contents of this registry are: 6409 +---------------------+------------+----------+ 6410 | Kind | Kind-ID | RFC | 6411 +---------------------+------------+----------+ 6412 | INVALID | 0 | RFC-AAAA | 6413 | TURN_SERVICE | 2 | RFC-AAAA | 6414 | CERTIFICATE_BY_NODE | 3 | RFC-AAAA | 6415 | CERTIFICATE_BY_USER | 16 | RFC-AAAA | 6416 | Reserved | 0x7fffffff | RFC-AAAA | 6417 | Reserved | 0xfffffffe | RFC-AAAA | 6418 +---------------------+------------+----------+ 6420 13.7. Data Model 6422 IANA SHALL create a "RELOAD Data Model" Registry. Entries in this 6423 registry denoting data models, as described in Section 6.2. Code 6424 points in this registry SHALL be registered via RFC 5226 Standards 6425 Action. The initial contents of this registry are: 6427 +------------+----------+ 6428 | Data Model | RFC | 6429 +------------+----------+ 6430 | INVALID | RFC-AAAA | 6431 | SINGLE | RFC-AAAA | 6432 | ARRAY | RFC-AAAA | 6433 | DICTIONARY | RFC-AAAA | 6434 | RESERVED | RFC-AAAA | 6435 +------------+----------+ 6437 13.8. Message Codes 6439 IANA SHALL create a "RELOAD Message Code" Registry. Entries in this 6440 registry are 16-bit integers denoting method codes as described in 6441 Section 5.3.3. These codes SHALL be registered via RFC 5226 6442 Standards Action. The initial contents of this registry are: 6444 +---------------------------------+----------------+----------+ 6445 | Message Code Name | Code Value | RFC | 6446 +---------------------------------+----------------+----------+ 6447 | invalid | 0 | RFC-AAAA | 6448 | probe_req | 1 | RFC-AAAA | 6449 | probe_ans | 2 | RFC-AAAA | 6450 | attach_req | 3 | RFC-AAAA | 6451 | attach_ans | 4 | RFC-AAAA | 6452 | unused | 5 | | 6453 | unused | 6 | | 6454 | store_req | 7 | RFC-AAAA | 6455 | store_ans | 8 | RFC-AAAA | 6456 | fetch_req | 9 | RFC-AAAA | 6457 | fetch_ans | 10 | RFC-AAAA | 6458 | unused (was remove_req) | 11 | RFC-AAAA | 6459 | unused (was remove_ans) | 12 | RFC-AAAA | 6460 | find_req | 13 | RFC-AAAA | 6461 | find_ans | 14 | RFC-AAAA | 6462 | join_req | 15 | RFC-AAAA | 6463 | join_ans | 16 | RFC-AAAA | 6464 | leave_req | 17 | RFC-AAAA | 6465 | leave_ans | 18 | RFC-AAAA | 6466 | update_req | 19 | RFC-AAAA | 6467 | update_ans | 20 | RFC-AAAA | 6468 | route_query_req | 21 | RFC-AAAA | 6469 | route_query_ans | 22 | RFC-AAAA | 6470 | ping_req | 23 | RFC-AAAA | 6471 | ping_ans | 24 | RFC-AAAA | 6472 | stat_req | 25 | RFC-AAAA | 6473 | stat_ans | 26 | RFC-AAAA | 6474 | unused (was attachlite_req) | 27 | RFC-AAAA | 6475 | unused (was attachlite_ans) | 28 | RFC-AAAA | 6476 | app_attach_req | 29 | RFC-AAAA | 6477 | app_attach_ans | 30 | RFC-AAAA | 6478 | unused (was app_attachlite_req) | 31 | RFC-AAAA | 6479 | unused (was app_attachlite_ans) | 32 | RFC-AAAA | 6480 | config_update_req | 33 | RFC-AAAA | 6481 | config_update_ans | 34 | RFC-AAAA | 6482 | reserved | 0x8000..0xfffe | RFC-AAAA | 6483 | error | 0xffff | RFC-AAAA | 6484 +---------------------------------+----------------+----------+ 6486 13.9. Error Codes 6488 IANA SHALL create a "RELOAD Error Code" Registry. Entries in this 6489 registry are 16-bit integers denoting error codes. New entries SHALL 6490 be defined via RFC 5226 Standards Action. The initial contents of 6491 this registry are: 6493 +-------------------------------------+----------------+----------+ 6494 | Error Code Name | Code Value | RFC | 6495 +-------------------------------------+----------------+----------+ 6496 | invalid | 0 | RFC-AAAA | 6497 | Unused | 1 | RFC-AAAA | 6498 | Error_Forbidden | 2 | RFC-AAAA | 6499 | Error_Not_Found | 3 | RFC-AAAA | 6500 | Error_Request_Timeout | 4 | RFC-AAAA | 6501 | Error_Generation_Counter_Too_Low | 5 | RFC-AAAA | 6502 | Error_Incompatible_with_Overlay | 6 | RFC-AAAA | 6503 | Error_Unsupported_Forwarding_Option | 7 | RFC-AAAA | 6504 | Error_Data_Too_Large | 8 | RFC-AAAA | 6505 | Error_Data_Too_Old | 9 | RFC-AAAA | 6506 | Error_TTL_Exceeded | 10 | RFC-AAAA | 6507 | Error_Message_Too_Large | 11 | RFC-AAAA | 6508 | Error_Unknown_Kind | 12 | RFC-AAAA | 6509 | Error_Unknown_Extension | 13 | RFC-AAAA | 6510 | Error_Response_Too_Large | 14 | RFC-AAAA | 6511 | Error_Config_Too_Old | 15 | RFC-AAAA | 6512 | Error_Config_Too_New | 16 | RFC-AAAA | 6513 | Error_In_Progress | 17 | RFC-AAAA | 6514 | reserved | 0x8000..0xfffe | RFC-AAAA | 6515 +-------------------------------------+----------------+----------+ 6517 13.10. Overlay Link Types 6519 IANA shall create a "RELOAD Overlay Link." New entries SHALL be 6520 defined via RFC 5226 Standards Action. This registry SHALL be 6521 initially populated with the following values: 6523 +--------------------+------+---------------+ 6524 | Protocol | Code | Specification | 6525 +--------------------+------+---------------+ 6526 | reserved | 0 | RFC-AAAA | 6527 | DTLS-UDP-SR | 1 | RFC-AAAA | 6528 | DTLS-UDP-SR-NO-ICE | 3 | RFC-AAAA | 6529 | TLS-TCP-FH-NO-ICE | 4 | RFC-AAAA | 6530 | reserved | 255 | RFC-AAAA | 6531 +--------------------+------+---------------+ 6533 13.11. Overlay Link Protocols 6535 IANA shall create an "Overlay Link Protocol Registry". Entries in 6536 this registry SHALL be defined via RFC 5226 Standards Action. This 6537 registry SHALL be initially populated with the following value: 6538 "TLS". 6540 13.12. Forwarding Options 6542 IANA shall create a "Forwarding Option Registry". Entries in this 6543 registry between 1 and 127 SHALL be defined via RFC 5226 Standards 6544 Action. Entries in this registry between 128 and 254 SHALL be 6545 defined via RFC 5226 Specification Required. This registry SHALL be 6546 initially populated with the following values: 6548 +-------------------+------+---------------+ 6549 | Forwarding Option | Code | Specification | 6550 +-------------------+------+---------------+ 6551 | invalid | 0 | RFC-AAAA | 6552 | reserved | 255 | RFC-AAAA | 6553 +-------------------+------+---------------+ 6555 13.13. Probe Information Types 6557 IANA shall create a "RELOAD Probe Information Type Registry". 6558 Entries in this registry SHALL be defined via RFC 5226 Standards 6559 Action. This registry SHALL be initially populated with the 6560 following values: 6562 +-----------------+------+---------------+ 6563 | Probe Option | Code | Specification | 6564 +-----------------+------+---------------+ 6565 | invalid | 0 | RFC-AAAA | 6566 | responsible_set | 1 | RFC-AAAA | 6567 | num_resources | 2 | RFC-AAAA | 6568 | uptime | 3 | RFC-AAAA | 6569 | reserved | 255 | RFC-AAAA | 6570 +-----------------+------+---------------+ 6572 13.14. Message Extensions 6574 IANA shall create a "RELOAD Extensions Registry". Entries in this 6575 registry SHALL be defined via RFC 5226 Specification Required. This 6576 registry SHALL be initially populated with the following values: 6578 +-----------------+--------+---------------+ 6579 | Extensions Name | Code | Specification | 6580 +-----------------+--------+---------------+ 6581 | invalid | 0 | RFC-AAAA | 6582 | reserved | 0xFFFF | RFC-AAAA | 6583 +-----------------+--------+---------------+ 6585 13.15. reload URI Scheme 6587 This section describes the scheme for a reload URI, which can be used 6588 to refer to either: 6590 o A peer. 6591 o A resource inside a peer. 6593 The reload URI is defined using a subset of the URI schema specified 6594 in Appendix A of RFC 3986 [RFC3986] and the associated URI Guidelines 6595 [RFC4395] per the following ABNF syntax: 6597 RELOAD-URI = "reload://" destination "@" overlay "/" 6598 [specifier] 6600 destination = 1 * HEXDIG 6601 overlay = reg-name 6602 specifier = 1*HEXDIG 6604 The definitions of these productions are as follows: 6606 destination: a hex-encoded Destination List object (i.e., multiple 6607 concatenated Destination objects with no length prefix prior to 6608 the object as a whole.) 6610 overlay: the name of the overlay. 6612 specifier : a hex-encoded StoredDataSpecifier indicating the data 6613 element. 6615 If no specifier is present then this URI addresses the peer which can 6616 be reached via the indicated destination list at the indicated 6617 overlay name. If a specifier is present, then the URI addresses the 6618 data value. 6620 13.15.1. URI Registration 6622 [[ Note to RFC Editor - please remove this paragraph before 6623 publication. ]] A review request was sent to uri-review@ietf.org on 6624 Oct 7, 2010. 6626 The following summarizes the information necessary to register the 6627 reload URI. 6629 URI Scheme Name: reload 6630 Status: permanent 6631 URI Scheme Syntax: see Section 13.15 of RFC-AAAA 6632 URI Scheme Semantics: The reload URI is intended to be used as a 6633 reference to a RELOAD peer or resource. 6634 Encoding Considerations: The reload URI is not intended to be human- 6635 readable text, so it is encoded entirely in US-ASCII. 6636 Applications/protocols that use this URI scheme: The RELOAD protocol 6637 described in RFC-AAAA. 6638 Interoperability considerations: See RFC-AAAA. 6639 Security considerations: See RFC-AAAA 6640 Contact: Cullen Jennings 6641 Author/Change controller: IESG 6642 References: RFC-AAAA 6644 13.16. Media Type Registration 6646 [[ Note to RFC Editor - please remove this paragraph before 6647 publication. ]] A review request was sent to ietf-types@iana.org on 6648 May 27, 2011. 6650 To: ietf-types@iana.org 6652 Subject: Registration of media type application/p2p-overlay+xml 6654 Type name: application 6656 Subtype name: p2p-overlay+xml 6658 Required parameters: none 6660 Optional parameters: none 6662 Encoding considerations: Must be binary encoded. The contents MUST 6663 be valid XML compliant with the relax NG grammar specified in RFC- 6664 AAAA and use the UTF-8[RFC3629] character encoding. 6666 Security considerations: This media type is typically not used to 6667 transport information that typically needs to be kept confidential 6668 however there are cases where it is integrity of the information is 6669 important. For these cases using a digital signature is RECOMMENDED. 6670 One way of doing this is specified in RFC-AAAA. In the case when the 6671 media includes a "shared-secret" element, then the contents of the 6672 file need to be kept confidential or else anyone that can see the 6673 shared-secret and effect the RELOAD overlay network. 6675 Interoperability considerations: Same as application/xml as defined 6676 in [RFC3023]. 6678 Published specification: RFC-AAAA 6680 Applications that use this media type: The type is used to configure 6681 the peer to peer overlay networks defined in RFC-AAAA. 6683 Additional information: The syntax for this media type is specified 6684 in Section 10.1 of RFC-AAAA. 6686 Magic number(s): none 6688 File extension(s): relo 6690 Macintosh file type code(s): none 6692 Person & email address to contact for further information: Cullen 6693 Jennings 6695 Intended usage: COMMON 6697 Restrictions on usage: None 6699 Author: Cullen Jennings 6701 Change controller: IESG 6703 14. Acknowledgments 6705 This specification is a merge of the "REsource LOcation And Discovery 6706 (RELOAD)" draft by David A. Bryan, Marcia Zangrilli and Bruce B. 6707 Lowekamp, the "Address Settlement by Peer to Peer" draft by Cullen 6708 Jennings, Jonathan Rosenberg, and Eric Rescorla, the "Security 6709 Extensions for RELOAD" draft by Bruce B. Lowekamp and James Deverick, 6710 the "A Chord-based DHT for Resource Lookup in P2PSIP" by Marcia 6711 Zangrilli and David A. Bryan, and the Peer-to-Peer Protocol (P2PP) 6712 draft by Salman A. Baset, Henning Schulzrinne, and Marcin 6713 Matuszewski. Thanks to the authors of RFC 5389 for text included 6714 from that. Vidya Narayanan provided many comments and improvements. 6716 The ideas and text for the Chord specific extension data to the Leave 6717 mechanisms was provided by J. Maenpaa, G. Camarillo, and J. 6718 Hautakorpi. 6720 Thanks to the many people who contributed including Ted Hardie, 6721 Michael Chen, Dan York, Das Saumitra, Lyndsay Campbell, Brian Rosen, 6722 David Bryan, Dave Craig, and Julian Cain. Extensive working last 6723 call comments were provided by: Jouni Maenpaa, Roni Even, Ari 6724 Keranen, John Buford, Michael Chen, Frederic-Philippe Met, and David 6725 Bryan. Special thanks to Marc Petit-Huguenin who provied an amazing 6726 amount to detailed review. 6728 15. References 6730 15.1. Normative References 6732 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 6733 Requirement Levels", BCP 14, RFC 2119, March 1997. 6735 [RFC2585] Housley, R. and P. Hoffman, "Internet X.509 Public Key 6736 Infrastructure Operational Protocols: FTP and HTTP", 6737 RFC 2585, May 1999. 6739 [RFC2818] Rescorla, E., "HTTP Over TLS", RFC 2818, May 2000. 6741 [RFC2988] Paxson, V. and M. Allman, "Computing TCP's Retransmission 6742 Timer", RFC 2988, November 2000. 6744 [RFC3023] Murata, M., St. Laurent, S., and D. Kohn, "XML Media 6745 Types", RFC 3023, January 2001. 6747 [RFC3174] Eastlake, D. and P. Jones, "US Secure Hash Algorithm 1 6748 (SHA1)", RFC 3174, September 2001. 6750 [RFC3447] Jonsson, J. and B. Kaliski, "Public-Key Cryptography 6751 Standards (PKCS) #1: RSA Cryptography Specifications 6752 Version 2.1", RFC 3447, February 2003. 6754 [RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO 6755 10646", STD 63, RFC 3629, November 2003. 6757 [RFC3986] Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform 6758 Resource Identifier (URI): Generic Syntax", STD 66, 6759 RFC 3986, January 2005. 6761 [RFC4279] Eronen, P. and H. Tschofenig, "Pre-Shared Key Ciphersuites 6762 for Transport Layer Security (TLS)", RFC 4279, 6763 December 2005. 6765 [RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 6766 Security", RFC 4347, April 2006. 6768 [RFC4395] Hansen, T., Hardie, T., and L. Masinter, "Guidelines and 6769 Registration Procedures for New URI Schemes", BCP 35, 6770 RFC 4395, February 2006. 6772 [RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data 6773 Encodings", RFC 4648, October 2006. 6775 [RFC5245] Rosenberg, J., "Interactive Connectivity Establishment 6776 (ICE): A Protocol for Network Address Translator (NAT) 6777 Traversal for Offer/Answer Protocols", RFC 5245, 6778 April 2010. 6780 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 6781 (TLS) Protocol Version 1.2", RFC 5246, August 2008. 6783 [RFC5272] Schaad, J. and M. Myers, "Certificate Management over CMS 6784 (CMC)", RFC 5272, June 2008. 6786 [RFC5273] Schaad, J. and M. Myers, "Certificate Management over CMS 6787 (CMC): Transport Protocols", RFC 5273, June 2008. 6789 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 6790 Friendly Rate Control (TFRC): Protocol Specification", 6791 RFC 5348, September 2008. 6793 [RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, 6794 "Session Traversal Utilities for NAT (STUN)", RFC 5389, 6795 October 2008. 6797 [RFC5766] Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using 6798 Relays around NAT (TURN): Relay Extensions to Session 6799 Traversal Utilities for NAT (STUN)", RFC 5766, April 2010. 6801 [RFC5952] Kawamura, S. and M. Kawashima, "A Recommendation for IPv6 6802 Address Text Representation", RFC 5952, August 2010. 6804 [RFC6091] Mavrogiannopoulos, N. and D. Gillmor, "Using OpenPGP Keys 6805 for Transport Layer Security (TLS) Authentication", 6806 RFC 6091, February 2011. 6808 [RFC6234] Eastlake, D. and T. Hansen, "US Secure Hash Algorithms 6809 (SHA and SHA-based HMAC and HKDF)", RFC 6234, May 2011. 6811 [w3c-xml-namespaces] 6812 Bray, T., Hollander, D., Layman, A., Tobin, R., and Henry 6813 S. , "Namespaces in XML 1.0 (Third Edition)". 6815 15.2. Informative References 6817 [Chord] Stoica, I., Morris, R., Liben-Nowell, D., Karger, D., 6818 Kaashoek, M., Dabek, F., and H. Balakrishnan, "Chord: A 6819 Scalable Peer-to-peer Lookup Protocol for Internet 6820 Applications", IEEE/ACM Transactions on Networking Volume 6821 11, Issue 1, 17-32, Feb 2003. 6823 [Eclipse] Singh, A., Ngan, T., Druschel, T., and D. Wallach, 6824 "Eclipse Attacks on Overlay Networks: Threats and 6825 Defenses", INFOCOM 2006, April 2006. 6827 [I-D.baset-tsvwg-tcp-over-udp] 6828 Baset, S. and H. Schulzrinne, "TCP-over-UDP", 6829 draft-baset-tsvwg-tcp-over-udp-01 (work in progress), 6830 June 2009. 6832 [I-D.ietf-hip-reload-instance] 6833 Keranen, A., Camarillo, G., and J. Maenpaa, "Host Identity 6834 Protocol-Based Overlay Networking Environment (HIP BONE) 6835 Instance Specification for REsource LOcation And Discovery 6836 (RELOAD)", draft-ietf-hip-reload-instance-03 (work in 6837 progress), January 2011. 6839 [I-D.ietf-mmusic-ice-tcp] 6840 Rosenberg, J., Keranen, A., Lowekamp, B., and A. Roach, 6841 "TCP Candidates with Interactive Connectivity 6842 Establishment (ICE)", draft-ietf-mmusic-ice-tcp-13 (work 6843 in progress), April 2011. 6845 [I-D.ietf-p2psip-concepts] 6846 Bryan, D., Matthews, P., Shim, E., Willis, D., and S. 6847 Dawkins, "Concepts and Terminology for Peer to Peer SIP", 6848 draft-ietf-p2psip-concepts-03 (work in progress), 6849 October 2010. 6851 [I-D.ietf-p2psip-sip] 6852 Jennings, C., Lowekamp, B., Rescorla, E., Baset, S., and 6853 H. Schulzrinne, "A SIP Usage for RELOAD", 6854 draft-ietf-p2psip-sip-05 (work in progress), July 2010. 6856 [I-D.jiang-p2psip-relay] 6857 Jiang, X., Zong, N., Even, R., and Y. Zhang, "An extension 6858 to RELOAD to support Direct Response and Relay Peer 6859 routing", draft-jiang-p2psip-relay-05 (work in progress), 6860 March 2011. 6862 [I-D.maenpaa-p2psip-self-tuning] 6863 Maenpaa, J., Camarillo, G., and J. Hautakorpi, "A Self- 6864 tuning Distributed Hash Table (DHT) for REsource LOcation 6865 And Discovery (RELOAD)", 6866 draft-maenpaa-p2psip-self-tuning-01 (work in progress), 6867 October 2009. 6869 [I-D.maenpaa-p2psip-service-discovery] 6870 Maenpaa, J. and G. Camarillo, "Service Discovery Usage for 6871 REsource LOcation And Discovery (RELOAD)", 6872 draft-maenpaa-p2psip-service-discovery-00 (work in 6873 progress), October 2009. 6875 [I-D.pascual-p2psip-clients] 6876 Pascual, V., Matuszewski, M., Shim, E., Zhang, H., and S. 6877 Yongchao, "P2PSIP Clients", 6878 draft-pascual-p2psip-clients-01 (work in progress), 6879 February 2008. 6881 [RFC1122] Braden, R., "Requirements for Internet Hosts - 6882 Communication Layers", STD 3, RFC 1122, October 1989. 6884 [RFC2311] Dusse, S., Hoffman, P., Ramsdell, B., Lundblade, L., and 6885 L. Repka, "S/MIME Version 2 Message Specification", 6886 RFC 2311, March 1998. 6888 [RFC4086] Eastlake, D., Schiller, J., and S. Crocker, "Randomness 6889 Requirements for Security", BCP 106, RFC 4086, June 2005. 6891 [RFC4145] Yon, D. and G. Camarillo, "TCP-Based Media Transport in 6892 the Session Description Protocol (SDP)", RFC 4145, 6893 September 2005. 6895 [RFC4787] Audet, F. and C. Jennings, "Network Address Translation 6896 (NAT) Behavioral Requirements for Unicast UDP", BCP 127, 6897 RFC 4787, January 2007. 6899 [RFC4828] Floyd, S. and E. Kohler, "TCP Friendly Rate Control 6900 (TFRC): The Small-Packet (SP) Variant", RFC 4828, 6901 April 2007. 6903 [RFC5054] Taylor, D., Wu, T., Mavrogiannopoulos, N., and T. Perrin, 6904 "Using the Secure Remote Password (SRP) Protocol for TLS 6905 Authentication", RFC 5054, November 2007. 6907 [RFC5201] Moskowitz, R., Nikander, P., Jokela, P., and T. Henderson, 6908 "Host Identity Protocol", RFC 5201, April 2008. 6910 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 6911 Housley, R., and W. Polk, "Internet X.509 Public Key 6912 Infrastructure Certificate and Certificate Revocation List 6913 (CRL) Profile", RFC 5280, May 2008. 6915 [RFC5765] Schulzrinne, H., Marocco, E., and E. Ivov, "Security 6916 Issues and Solutions in Peer-to-Peer Systems for Realtime 6917 Communications", RFC 5765, February 2010. 6919 [RFC5785] Nottingham, M. and E. Hammer-Lahav, "Defining Well-Known 6920 Uniform Resource Identifiers (URIs)", RFC 5785, 6921 April 2010. 6923 [RFC6079] Camarillo, G., Nikander, P., Hautakorpi, J., Keranen, A., 6924 and A. Johnston, "HIP BONE: Host Identity Protocol (HIP) 6925 Based Overlay Networking Environment (BONE)", RFC 6079, 6926 January 2011. 6928 [Sybil] Douceur, J., "The Sybil Attack", IPTPS 02, March 2002. 6930 [UnixTime] 6931 Wikipedia, "Unix Time", . 6934 [bryan-design-hotp2p08] 6935 Bryan, D., Lowekamp, B., and M. Zangrilli, "The Design of 6936 a Versatile, Secure P2PSIP Communications Architecture for 6937 the Public Internet", Hot-P2P'08. 6939 [handling-churn-usenix04] 6940 Rhea, S., Geels, D., Roscoe, T., and J. Kubiatowicz, 6941 "Handling Churn in a DHT", In Proc. of the USENIX Annual 6942 Technical Conference June 2004 USENIX 2004. 6944 [lookups-churn-p2p06] 6945 Wu, D., Tian, Y., and K. Ng, "Analytical Study on 6946 Improving DHT Lookup Performance under Churn", IEEE 6947 P2P'06. 6949 [minimizing-churn-sigcomm06] 6950 Godfrey, P., Shenker, S., and I. Stoica, "Minimizing Churn 6951 in Distributed Systems", SIGCOMM 2006. 6953 [non-transitive-dhts-worlds05] 6954 Freedman, M., Lakshminarayanan, K., Rhea, S., and I. 6955 Stoica, "Non-Transitive Connectivity and DHTs", 6956 WORLDS'05. 6958 [opendht-sigcomm05] 6959 Rhea, S., Godfrey, B., Karp, B., Kubiatowicz, J., 6960 Ratnasamy, S., Shenker, S., Stoica, I., and H. Yu, 6961 "OpenDHT: A Public DHT and its Uses", SIGCOMM'05. 6963 [vulnerabilities-acsac04] 6964 Srivatsa, M. and L. Liu, "Vulnerabilities and Security 6965 Threats in Structured Peer-to-Peer Systems: A Quantitative 6966 Analysis", ACSAC 2004. 6968 Appendix A. Routing Alternatives 6970 Significant discussion has been focused on the selection of a routing 6971 algorithm for P2PSIP. This section discusses the motivations for 6972 selecting symmetric recursive routing for RELOAD and describes the 6973 extensions that would be required to support additional routing 6974 algorithms. 6976 A.1. Iterative vs Recursive 6978 Iterative routing has a number of advantages. It is easier to debug, 6979 consumes fewer resources on intermediate peers, and allows the 6980 querying peer to identify and route around misbehaving peers 6981 [non-transitive-dhts-worlds05]. However, in the presence of NATs, 6982 iterative routing is intolerably expensive because a new connection 6983 must be established for each hop (using ICE) [bryan-design-hotp2p08]. 6985 Iterative routing is supported through the RouteQuery mechanism and 6986 is primarily intended for debugging. It also allows the querying 6987 peer to evaluate the routing decisions made by the peers at each hop, 6988 consider alternatives, and perhaps detect at what point the 6989 forwarding path fails. 6991 A.2. Symmetric vs Forward response 6993 An alternative to the symmetric recursive routing method used by 6994 RELOAD is Forward-Only routing, where the response is routed to the 6995 requester as if it were a new message initiated by the responder (in 6996 the previous example, Z sends the response to A as if it were sending 6997 a request). Forward-only routing requires no state in either the 6998 message or intermediate peers. 7000 The drawback of forward-only routing is that it does not work when 7001 the overlay is unstable. For example, if A is in the process of 7002 joining the overlay and is sending a Join request to Z, it is not yet 7003 reachable via forward routing. Even if it is established in the 7004 overlay, if network failures produce temporary instability, A may not 7005 be reachable (and may be trying to stabilize its network connectivity 7006 via Attach messages). 7008 Furthermore, forward-only responses are less likely to reach the 7009 querying peer than symmetric recursive ones are, because the forward 7010 path is more likely to have a failed peer than is the request path 7011 (which was just tested to route the request) 7013 [non-transitive-dhts-worlds05]. 7015 An extension to RELOAD that supports forward-only routing but relies 7016 on symmetric responses as a fallback would be possible, but due to 7017 the complexities of determining when to use forward-only and when to 7018 fallback to symmetric, we have chosen not to include it as an option 7019 at this point. 7021 A.3. Direct Response 7023 Another routing option is Direct Response routing, in which the 7024 response is returned directly to the querying node. In the previous 7025 example, if A encodes its IP address in the request, then Z can 7026 simply deliver the response directly to A. In the absence of NATs or 7027 other connectivity issues, this is the optimal routing technique. 7029 The challenge of implementing direct response is the presence of 7030 NATs. There are a number of complexities that must be addressed. In 7031 this discussion, we will continue our assumption that A issued the 7032 request and Z is generating the response. 7034 o The IP address listed by A may be unreachable, either due to NAT 7035 or firewall rules. Therefore, a direct response technique must 7036 fallback to symmetric response [non-transitive-dhts-worlds05]. 7037 The hop-by-hop ACKs used by RELOAD allow Z to determine when A has 7038 received the message (and the TLS negotiation will provide earlier 7039 confirmation that A is reachable), but this fallback requires a 7040 timeout that will increase the response latency whenever A is not 7041 reachable from Z. 7042 o Whenever A is behind a NAT it will have multiple candidate IP 7043 addresses, each of which must be advertised to ensure 7044 connectivity; therefore Z will need to attempt multiple 7045 connections to deliver the response. 7046 o One (or all) of A's candidate addresses may route from Z to a 7047 different device on the Internet. In the worst case these nodes 7048 may actually be running RELOAD on the same port. Therefore, it is 7049 absolutely necessary to establish a secure connection to 7050 authenticate A before delivering the response. This step 7051 diminishes the efficiency of direct response because multiple 7052 roundtrips are required before the message can be delivered. 7053 o If A is behind a NAT and does not have a connection already 7054 established with Z, there are only two ways the direct response 7055 will work. The first is that A and Z both be behind the same NAT, 7056 in which case the NAT is not involved. In the more common case, 7057 when Z is outside A's NAT, the response will only be received if 7058 A's NAT implements endpoint-independent filtering. As the choice 7059 of filtering mode conflates application transparency with security 7060 [RFC4787], and no clear recommendation is available, the 7061 prevalence of this feature in future devices remains unclear. 7063 An extension to RELOAD that supports direct response routing but 7064 relies on symmetric responses as a fallback would be possible, but 7065 due to the complexities of determining when to use direct response 7066 and when to fallback to symmetric, and the reduced performance for 7067 responses to peers behind restrictive NATs, we have chosen not to 7068 include it as an option at this point. 7070 A.4. Relay Peers 7072 [I-D.jiang-p2psip-relay] has proposed implementing a form of direct 7073 response by having A identify a peer, Q, that will be directly 7074 reachable by any other peer. A uses Attach to establish a connection 7075 with Q and advertises Q's IP address in the request sent to Z. Z 7076 sends the response to Q, which relays it to A. This then reduces the 7077 latency to two hops, plus Z negotiating a secure connection to Q. 7079 This technique relies on the relative population of nodes such as A 7080 that require relay peers and peers such as Q that are capable of 7081 serving as a relay peer. It also requires nodes to be able to 7082 identify which category they are in. This identification problem has 7083 turned out to be hard to solve and is still an open area of 7084 exploration. 7086 An extension to RELOAD that supports relay peers is possible, but due 7087 to the complexities of implementing such an alternative, we have not 7088 added such a feature to RELOAD at this point. 7090 A concept similar to relay peers, essentially choosing a relay peer 7091 at random, has previously been suggested to solve problems of 7092 pairwise non-transitivity [non-transitive-dhts-worlds05], but 7093 deterministic filtering provided by NATs makes random relay peers no 7094 more likely to work than the responding peer. 7096 A.5. Symmetric Route Stability 7098 A common concern about symmetric recursive routing has been that one 7099 or more peers along the request path may fail before the response is 7100 received. The significance of this problem essentially depends on 7101 the response latency of the overlay. An overlay that produces slow 7102 responses will be vulnerable to churn, whereas responses that are 7103 delivered very quickly are vulnerable only to failures that occur 7104 over that small interval. 7106 The other aspect of this issue is whether the request itself can be 7107 successfully delivered. Assuming typical connection maintenance 7108 intervals, the time period between the last maintenance and the 7109 request being sent will be orders of magnitude greater than the delay 7110 between the request being forwarded and the response being received. 7111 Therefore, if the path was stable enough to be available to route the 7112 request, it is almost certainly going to remain available to route 7113 the response. 7115 An overlay that is unstable enough to suffer this type of failure 7116 frequently is unlikely to be able to support reliable functionality 7117 regardless of the routing mechanism. However, regardless of the 7118 stability of the return path, studies show that in the event of high 7119 churn, iterative routing is a better solution to ensure request 7120 completion [lookups-churn-p2p06] [non-transitive-dhts-worlds05] 7122 Finally, because RELOAD retries the end-to-end request, that retry 7123 will address the issues of churn that remain. 7125 Appendix B. Why Clients? 7127 There are a wide variety of reasons a node may act as a client rather 7128 than as a peer [I-D.pascual-p2psip-clients]. This section outlines 7129 some of those scenarios and how the client's behavior changes based 7130 on its capabilities. 7132 B.1. Why Not Only Peers? 7134 For a number of reasons, a particular node may be forced to act as a 7135 client even though it is willing to act as a peer. These include: 7137 o The node does not have appropriate network connectivity, typically 7138 because it has a low-bandwidth network connection. 7139 o The node may not have sufficient resources, such as computing 7140 power, storage space, or battery power. 7141 o The overlay algorithm may dictate specific requirements for peer 7142 selection. These may include participating in the overlay to 7143 determine trustworthiness; controlling the number of peers in the 7144 overlay to reduce overly-long routing paths; or ensuring minimum 7145 application uptime before a node can join as a peer. 7147 The ultimate criteria for a node to become a peer are determined by 7148 the overlay algorithm and specific deployment. A node acting as a 7149 client that has a full implementation of RELOAD and the appropriate 7150 overlay algorithm is capable of locating its responsible peer in the 7151 overlay and using Attach to establish a direct connection to that 7152 peer. In that way, it may elect to be reachable under either of the 7153 routing approaches listed above. Particularly for overlay algorithms 7154 that elect nodes to serve as peers based on trustworthiness or 7155 population, the overlay algorithm may require such a client to locate 7156 itself at a particular place in the overlay. 7158 B.2. Clients as Application-Level Agents 7160 SIP defines an extensive protocol for registration and security 7161 between a client and its registrar/proxy server(s). Any SIP device 7162 can act as a client of a RELOAD-based P2PSIP overlay if it contacts a 7163 peer that implements the server-side functionality required by the 7164 SIP protocol. In this case, the peer would be acting as if it were 7165 the user's peer, and would need the appropriate credentials for that 7166 user. 7168 Application-level support for clients is defined by a usage. A usage 7169 offering support for application-level clients should specify how the 7170 security of the system is maintained when the data is moved between 7171 the application and RELOAD layers. 7173 Appendix C. Change Log 7175 C.1. Changes since draft-ietf-p2psip-reload-13 7177 Note to RFC Editor: Please remove this section. 7179 o Fixed finger formula. 7180 o Support for getting a certificate with multiple Node-IDs. 7181 o Added configuration signer. 7182 o Added way to specify mandatory extensions in configuration file. 7184 Authors' Addresses 7186 Cullen Jennings 7187 Cisco 7188 170 West Tasman Drive 7189 MS: SJC-21/2 7190 San Jose, CA 95134 7191 USA 7193 Phone: +1 408 421-9990 7194 Email: fluffy@cisco.com 7195 Bruce B. Lowekamp (editor) 7196 Skype 7197 Palo Alto, CA 7198 USA 7200 Email: bbl@lowekamp.net 7202 Eric Rescorla 7203 RTFM, Inc. 7204 2064 Edgewood Drive 7205 Palo Alto, CA 94303 7206 USA 7208 Phone: +1 650 678 2350 7209 Email: ekr@rtfm.com 7211 Salman A. Baset 7212 Columbia University 7213 1214 Amsterdam Avenue 7214 New York, NY 7215 USA 7217 Email: salman@cs.columbia.edu 7219 Henning Schulzrinne 7220 Columbia University 7221 1214 Amsterdam Avenue 7222 New York, NY 7223 USA 7225 Email: hgs@cs.columbia.edu