idnits 2.17.1 draft-ietf-nsis-qos-nslp-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1.a on line 20. -- Found old boilerplate from RFC 3978, Section 5.5 on line 3209. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 3186. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 3193. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 3199. ** The document seems to lack an RFC 3978 Section 5.1 IPR Disclosure Acknowledgement. ** This document has an original RFC 3978 Section 5.4 Copyright Line, instead of the newer IETF Trust Copyright according to RFC 4748. ** This document has an original RFC 3978 Section 5.5 Disclaimer, instead of the newer disclaimer which includes the IETF Trust according to RFC 4748. ** The document uses RFC 3667 boilerplate or RFC 3978-like boilerplate instead of verbatim RFC 3978 boilerplate. After 6 May 2005, submission of drafts without verbatim RFC 3978 boilerplate is not accepted. The following non-3978 patterns matched text found in the document. That text should be removed or replaced: This document is an Internet-Draft and is subject to all provisions of Section 3 of RFC 3667. By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** There is 1 instance of too long lines in the document, the longest one being 2 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the RFC 3978 Section 5.4 Copyright Line does not match the current year == Line 2999 has weird spacing: '...d usage excha...' -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (February 20, 2005) is 7004 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'RFC2119' is defined on line 2934, but no explicit reference was found in the text == Unused Reference: 'I-D.westberg-rmd-framework' is defined on line 2986, but no explicit reference was found in the text == Unused Reference: 'RFC2475' is defined on line 3017, but no explicit reference was found in the text == Unused Reference: 'RFC3583' is defined on line 3035, but no explicit reference was found in the text == Outdated reference: A later version (-20) exists of draft-ietf-nsis-ntlp-04 ** Downref: Normative reference to an Experimental draft: draft-ietf-nsis-ntlp (ref. 'I-D.ietf-nsis-ntlp') ** Obsolete normative reference: RFC 2234 (Obsoleted by RFC 4234) == Outdated reference: A later version (-24) exists of draft-ietf-nsis-qspec-03 == Outdated reference: A later version (-20) exists of draft-ietf-nsis-rmd-01 == Outdated reference: A later version (-14) exists of draft-kappler-nsis-qosmodel-controlledload-00 -- Obsolete informational reference (is this intentional?): RFC 2434 (Obsoleted by RFC 5226) Summary: 8 errors (**), 0 flaws (~~), 11 warnings (==), 8 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Next Steps in Signalling S. Van den Bosch 3 Internet-Draft Alcatel 4 Expires: August 24, 2005 G. Karagiannis 5 University of Twente/Ericsson 6 A. McDonald 7 Siemens/Roke Manor Research 8 February 20, 2005 10 NSLP for Quality-of-Service signalling 11 draft-ietf-nsis-qos-nslp-06.txt 13 Status of this Memo 15 This document is an Internet-Draft and is subject to all provisions 16 of Section 3 of RFC 3667. By submitting this Internet-Draft, each 17 author represents that any applicable patent or other IPR claims of 18 which he or she is aware have been or will be disclosed, and any of 19 which he or she become aware will be disclosed, in accordance with 20 RFC 3668. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF), its areas, and its working groups. Note that 24 other groups may also distribute working documents as 25 Internet-Drafts. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 The list of current Internet-Drafts can be accessed at 33 http://www.ietf.org/ietf/1id-abstracts.txt. 35 The list of Internet-Draft Shadow Directories can be accessed at 36 http://www.ietf.org/shadow.html. 38 This Internet-Draft will expire on August 24, 2005. 40 Copyright Notice 42 Copyright (C) The Internet Society (2005). 44 Abstract 46 This draft describes an NSIS Signalling Layer Protocol (NSLP) for 47 signalling QoS reservations in the Internet. It is in accordance 48 with the framework and requirements developed in NSIS. 50 Together with GIMPS, it provides functionality similar to RSVP and 51 extends it. The QoS-NSLP is independent of the underlying QoS 52 specification or architecture and provides support for different 53 reservation models. It is simplified by the elimination of support 54 for multicast flows. 56 This draft explains the overall protocol approach, design decisions 57 made and provides examples. It specifies object and message formats 58 and processing rules. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 5 63 1.1 Scope and background . . . . . . . . . . . . . . . . . . . 5 64 1.2 Model of operation . . . . . . . . . . . . . . . . . . . . 6 65 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . 9 66 3. Protocol Overview . . . . . . . . . . . . . . . . . . . . . 10 67 3.1 Overall approach . . . . . . . . . . . . . . . . . . . . . 10 68 3.1.1 GIMPS Interactions . . . . . . . . . . . . . . . . . . 10 69 3.1.2 Protocol messages . . . . . . . . . . . . . . . . . . 10 70 3.1.3 QoS Models and QoS Specifications . . . . . . . . . . 11 71 3.1.4 Policy control . . . . . . . . . . . . . . . . . . . . 12 72 3.2 Design decisions . . . . . . . . . . . . . . . . . . . . . 14 73 3.2.1 Soft-state . . . . . . . . . . . . . . . . . . . . . . 14 74 3.2.2 Sender-receiver initiation . . . . . . . . . . . . . . 14 75 3.2.3 Message sequencing . . . . . . . . . . . . . . . . . . 14 76 3.2.4 Explicit state installation confirmation and 77 responses . . . . . . . . . . . . . . . . . . . . . . 15 78 3.2.5 Summary refreshes . . . . . . . . . . . . . . . . . . 15 79 3.2.6 Message scoping . . . . . . . . . . . . . . . . . . . 15 80 3.2.7 Session binding . . . . . . . . . . . . . . . . . . . 16 81 3.2.8 Layering . . . . . . . . . . . . . . . . . . . . . . . 16 82 3.2.9 Priority . . . . . . . . . . . . . . . . . . . . . . . 18 83 3.2.10 Rerouting . . . . . . . . . . . . . . . . . . . . . 19 84 4. Examples of QoS NSLP Operation . . . . . . . . . . . . . . . 21 85 4.1 Basic sender-initiated reservation . . . . . . . . . . . . 21 86 4.2 Sending a Query . . . . . . . . . . . . . . . . . . . . . 22 87 4.3 Basic receiver-initiated reservation . . . . . . . . . . . 23 88 4.4 Bidirectional Reservations . . . . . . . . . . . . . . . . 25 89 4.5 Use of Local QoS Models . . . . . . . . . . . . . . . . . 26 90 4.6 Aggregate Reservations . . . . . . . . . . . . . . . . . . 27 91 4.7 Reduced State or Stateless Interior Nodes . . . . . . . . 29 92 4.8 Re-routing scenario . . . . . . . . . . . . . . . . . . . 32 93 4.9 Authorization Model Examples . . . . . . . . . . . . . . . 33 94 4.9.1 Authorization for the two party approach . . . . . . . 33 95 4.9.2 Token based three party approach . . . . . . . . . . . 33 96 4.9.3 Generic three party approach . . . . . . . . . . . . . 34 97 5. QoS NSLP Functional specification . . . . . . . . . . . . . 36 98 5.1 QoS NSLP Message and Object Formats . . . . . . . . . . . 36 99 5.1.1 Common header . . . . . . . . . . . . . . . . . . . . 36 100 5.1.2 Message formats . . . . . . . . . . . . . . . . . . . 37 101 5.1.3 Object Formats . . . . . . . . . . . . . . . . . . . . 40 102 5.2 General Processing Rules . . . . . . . . . . . . . . . . . 44 103 5.2.1 State Manipulation . . . . . . . . . . . . . . . . . . 44 104 5.2.2 Message Forwarding . . . . . . . . . . . . . . . . . . 45 105 5.2.3 Standard Message Processing Rules . . . . . . . . . . 46 106 5.3 Object Processing . . . . . . . . . . . . . . . . . . . . 46 107 5.3.1 Reservation Sequence Number (RSN) . . . . . . . . . . 46 108 5.3.2 Request Identification Information (RII) . . . . . . . 46 109 5.3.3 BOUND_SESSION_ID . . . . . . . . . . . . . . . . . . . 47 110 5.3.4 REFRESH_PERIOD . . . . . . . . . . . . . . . . . . . . 47 111 5.3.5 ERROR_SPEC . . . . . . . . . . . . . . . . . . . . . . 49 112 5.3.6 QSPEC . . . . . . . . . . . . . . . . . . . . . . . . 49 113 5.4 Message Processing Rules . . . . . . . . . . . . . . . . . 50 114 5.4.1 RESERVE Messages . . . . . . . . . . . . . . . . . . . 50 115 5.4.2 QUERY Messages . . . . . . . . . . . . . . . . . . . . 53 116 5.4.3 RESPONSE Messages . . . . . . . . . . . . . . . . . . 54 117 5.4.4 NOTIFY Messages . . . . . . . . . . . . . . . . . . . 54 118 6. IANA considerations . . . . . . . . . . . . . . . . . . . . 56 119 7. QoS use of GIMPS service interface . . . . . . . . . . . . . 57 120 7.1 Example sender-initiated reservation . . . . . . . . . . . 57 121 7.2 Session identification . . . . . . . . . . . . . . . . . . 58 122 7.3 Support for bypassing intermediate nodes . . . . . . . . . 58 123 7.4 Support for peer change identification . . . . . . . . . . 59 124 7.5 Support for stateless operation . . . . . . . . . . . . . 59 125 7.6 Last node detection . . . . . . . . . . . . . . . . . . . 59 126 7.7 Re-routing detection . . . . . . . . . . . . . . . . . . . 60 127 7.8 Priority of signalling messages . . . . . . . . . . . . . 60 128 7.9 Knowledge of intermediate QoS NSLP unaware nodes . . . . . 60 129 7.10 NSLP Data Size . . . . . . . . . . . . . . . . . . . . . 61 130 7.11 Notification of GIMPS 'D' flag value . . . . . . . . . . 61 131 7.12 NAT Traversal . . . . . . . . . . . . . . . . . . . . . 61 132 8. Assumptions on the QoS Model . . . . . . . . . . . . . . . . 62 133 8.1 Resource sharing . . . . . . . . . . . . . . . . . . . . . 62 134 8.2 Reserve/commit support . . . . . . . . . . . . . . . . . . 62 135 9. Open issues . . . . . . . . . . . . . . . . . . . . . . . . 63 136 9.1 Peering agreements on interdomain links . . . . . . . . . 63 137 9.2 Protocol Operating Environment Assumptions . . . . . . . . 63 138 9.3 Authorization components in QoS NSLP . . . . . . . . . . . 64 139 10. Security Considerations . . . . . . . . . . . . . . . . . . 65 140 10.1 Introduction and Threat Overview . . . . . . . . . . . . 65 141 10.2 Trust Model . . . . . . . . . . . . . . . . . . . . . . 66 142 10.3 Computing the authorization decision . . . . . . . . . . 68 143 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 69 144 12. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 70 145 13. References . . . . . . . . . . . . . . . . . . . . . . . . . 71 146 13.1 Normative References . . . . . . . . . . . . . . . . . . 71 147 13.2 Informative References . . . . . . . . . . . . . . . . . 71 148 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . 73 149 A. Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . 75 150 B. Change History . . . . . . . . . . . . . . . . . . . . . . . 76 151 Intellectual Property and Copyright Statements . . . . . . . 78 153 1. Introduction 155 1.1 Scope and background 157 This document defines a Quality of Service (QoS) NSIS Signalling 158 Layer Protocol (NSLP), henceforth referred to as the "QoS-NSLP". 159 This protocol establishes and maintains state at nodes along the path 160 of a data flow for the purpose of providing some forwarding resources 161 for that flow. It is intended to satisfy the QoS-related 162 requirements of RFC 3726 [RFC3726]. This QoS-NSLP is part of a 163 larger suite of signalling protocols, whose structure is outlined in 164 the NSIS framework [I-D.ietf-nsis-fw]; this defines a common NSIS 165 Transport Layer Protocol (NTLP) which QoS-NSLP uses to carry out many 166 aspects of signalling message delivery. A specification of the NTLP, 167 GIMPS [I-D.ietf-nsis-ntlp] is done in another document. 169 The design of QoS-NSLP is conceptually similar to RSVP, RFC 2205 170 [RFC2205], and uses soft-state peer-to-peer refresh messages as the 171 primary state management mechanism (i.e. state installation/refresh 172 is performed between pairs of adjacent NSLP nodes, rather than in an 173 end-to-end fashion along the complete signalling path). Although 174 there is no backwards compatibility at the level of protocol 175 messages, interworking with RSVP at a signalling application gateway 176 would be possible in some circumstances. QoS-NSLP extends the set of 177 reservation mechanisms to meet the requirements of RFC 3726 178 [RFC3726], in particular support of sender or receiver-initiated 179 reservations, as well as a type of bi-directional reservation and 180 support of reservations between arbitrary nodes, e.g. edge-to-edge, 181 end-to-access, etc. On the other hand, there is no support for IP 182 multicast. 184 A distinction is made between the operation of the signalling 185 protocol and the information required for the operation of the 186 Resource Management Function (RMF). This document describes the 187 signalling protocol, whilst [I-D.ietf-nsis-qspec] describes the 188 RMF-related information carried in the QSpec (QoS Specification) 189 object in QoS-NSLP messages. This is similar to the decoupling 190 between RSVP and the IntServ architecture, RFC 1633 [RFC1633]. The 191 QSpec carries information on resources available, resources required, 192 traffic descriptions and other information required by the RMF. 194 This document is structured as follows. The overall approach to 195 protocol design is outlined in Section 3.1. The operation and use of 196 QoS NSLP is then clarified by means of a number of examples in 197 Section 4. These sections should be read by readers interested in 198 the protocol capabilities. The functional specification Section 5 199 contains more detailed object and message formats and processing 200 rules and should be the basis for implementers. The subsequent 201 sections describe extensibility (IANA), requirements on GIMPS API and 202 security considerations. 204 1.2 Model of operation 206 This section presents a logical model for the operation of the 207 QoS-NSLP and associated provisioning mechanisms within a single node. 208 The model is shown in Figure 1. 210 +---------------+ 211 | Local | 212 |Applications or| 213 |Management (e.g| 214 |for aggregates)| 215 +---------------+ 216 ^ 217 ^ 218 V 219 V 220 +----------+ +----------+ +---------+ 221 | QoS-NSLP | | Resource | | Policy | 222 |Processing|<<<<<<>>>>>>>|Management|<<<>>>| Control | 223 +----------+ +----------+ +---------+ 224 . ^ | * ^ 225 | V . * ^ 226 +----------+ * ^ 227 | NTLP | * ^ 228 |Processing| * V 229 +----------+ * V 230 | | * V 231 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 232 . . * V 233 | | * ............................. 234 . . * . Traffic Control . 235 | | * . +---------+. 236 . . * . |Admission|. 237 | | * . | Control |. 238 +----------+ +------------+ . +---------+. 239 <-.-| Input | | Outgoing |-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-> 240 | Packet | | Interface | .+----------+ +---------+. 241 ===>|Processing|====| Selection |===.| Packet |====| Packet |.==> 242 | | |(Forwarding)| .|Classifier| Scheduler|. 243 +----------+ +------------+ .+----------+ +---------+. 244 ............................. 245 <.-.-> = signalling flow 246 =====> = data flow (sender --> receiver) 247 <<<>>> = control and configuration operations 248 ****** = routing table manipulation 250 Figure 1: QoS-NSLP in a Node 252 This diagram shows an example implementation scenario where QoS 253 conditioning is performed on the output interface. However, this 254 does not limit the possible implementations. For example, in some 255 cases traffic conditioning may be performed on the incoming 256 interface, or it may be split over the input and output interfaces. 258 From the perspective of a single node, the request for QoS may result 259 from a local application request, or from processing an incoming QoS- 260 NSLP message. 261 o The 'local application case' includes not only user applications 262 (e.g. multimedia applications) but also network management (e.g. 263 initiating a tunnel to handle an aggregate, or interworking with 264 some other reservation protocol - such as RSVP) and the policy 265 control module (e.g. for explicit teardown triggered by AAA). In 266 this sense, the model does not distinguish between hosts and 267 routers. 268 o The 'incoming message' case requires NSIS messages to be captured 269 during input packet processing and handled by GIMPS. Only 270 messages related to QoS are passed to the QoS-NSLP. GIMPS may 271 also generate triggers to the QoS-NSLP (e.g. indications that a 272 route change has occurred). 274 The QoS request is handled by a local 'resource management' function, 275 which coordinates the activities required to grant and configure the 276 resource. It also handles policy-specific aspects of QoS signaling. 277 o The grant processing involves two local decision modules, 'policy 278 control' and 'admission control'. Policy control determines 279 whether the user has administrative permission to make the 280 reservation. Admission control determines whether the node has 281 sufficient available resources to supply the requested QoS. 282 o If both checks succeed, parameters are set in the packet 283 classifier and in the link layer interface (e.g., in the packet 284 scheduler) to obtain the desired QoS. Error notifications are 285 passed back to the request originator. The resource management 286 function may also manipulate the forwarding tables at this stage, 287 to select (or at least pin) a route; this must be done before 288 interface-dependent actions are carried out (including forwarding 289 outgoing messages over any new route), and is in any case 290 invisible to the operation of the protocol. 292 Policy control is expected to make use of a AAA service external to 293 the node itself. Some discussion can be found in a separate document 294 on AAA issues [I-D.tschofenig-nsis-aaa-issues] and one on 295 auhorization issues [I-D.tschofenig-nsis-qos-authz-issues]. More 296 generally, the processing of policy and resource management functions 297 may be outsourced to an external node leaving only 'stubs' co-located 298 with the NSLP; this is not visible to the protocol operation, 299 although it may have some influence on the detailed design of 300 protocol messages to allow the stub to be minimally complex. A more 301 detailed discussion on authentication and authorization can be found 302 in Section 3.1.4. 304 The group of user plane functions, which implement QoS for a flow 305 (admission control, packet classification, and scheduling) is 306 sometimes known as 'traffic control'. 308 Admission control, packet scheduling, and any part of policy control 309 beyond simple authentication have to be implemented using specific 310 definitions for types and levels of QoS; Our assumption is that the 311 QoS-NSLP is independent of the QoS parameters (e.g. IntServ service 312 elements). These are captured in a QoS Model and interpreted only by 313 the resource management and associated functions, and are opaque to 314 the QoS-NSLP itself. QoS Models are discussed further in 315 Section 3.1.3. 317 The final stage of processing for a resource request is to indicate 318 to the QoS-NSLP protocol processing that the required resources have 319 been configured. The QoS-NSLP may generate an acknowledgement 320 message in one direction, and may forward the resource request in the 321 other. Message routing is (by default) carried out by GIMPS module. 322 Note that while Figure 1 shows a unidirectional data flow, the 323 signalling messages can pass in both directions through the node, 324 depending on the particular message and orientation of the 325 reservation. 327 2. Terminology 329 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 330 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 331 document are to be interpreted as described in RFC 2119. 333 The terminology defined by GIMPS [I-D.ietf-nsis-ntlp] applies to this 334 draft. 336 In addition, the following terms are used: 337 QNE: an NSIS Entity (NE), which supports the QoS-NSLP. 338 QNI: the first node in the sequence of QNEs that issues a reservation 339 request for a session. 340 QNR: the last node in the sequence of QNEs that receives a 341 reservation request for a session. 342 Source or message source: The one of two adjacent NSLP peers that is 343 sending a signalling message (maybe the upstream or the downstream 344 peer). NB: this is not necessarily the QNI. 345 QoS NSLP operation state: state used/kept by QoS NSLP processing to 346 handle messaging aspects. 347 QoS reservation state: state used/kept by Resource Management 348 Function to describe reserved resources for a session. 350 QoS NSLP nodes 351 IP address (QoS unware NSIS nodes are IP address 352 = Flow not shown) = Flow 353 Source | | | Destination 354 Address | | | Address 355 V V V 356 +--------+ Data +------+ +------+ +------+ +--------+ 357 | Flow |-------|------|------|------|-------|------|---->| Flow | 358 | Sender | Flow | | | | | | |Receiver| 359 +--------+ | QNI | | QNE | | QNR | +--------+ 360 | | | | | | 361 +------+ +------+ +------+ 362 =====================> 363 <===================== 364 Signalling 365 Flow 367 A glossary of terms and abbreviations used in this document can be 368 found in Appendix A. 370 3. Protocol Overview 372 3.1 Overall approach 374 3.1.1 GIMPS Interactions 376 The QoS NSLP uses GIMPS for delivery of all its messages. Messages 377 are normally passed from the NSLP to the GIMPS via an API (defined in 378 Appendix D of [I-D.ietf-nsis-ntlp]), which also specifies additional 379 information, including an identifier for the signalling application 380 (e.g. 'QoS-NSLP'), the flow/session identifier, and an indication of 381 the intended direction - towards data sender or receiver. On 382 reception, GIMPS provides the same information to the QoS-NSLP. In 383 addition to the NSLP message data itself, other meta-data (e.g. 384 session identifier, flow routing information) can be transferred 385 across this interface. 387 The QoS NSLP does not provide any method of interacting with 388 firewalls or Network Address Translators (NATs). It assumes that a 389 basic NAT traversal service is provided by the GIMPS. 391 3.1.2 Protocol messages 393 The QoS NSLP uses four message types: 394 RESERVE: The RESERVE message is the only message that manipulates 395 QoS NSLP reservation state. It is used to create, refresh, modify 396 and remove such state. The RESERVE message is idempotent; the 397 resultant effect is the same whether a message is received once or 398 many times. 400 QUERY: A QUERY message is used to request information about the data 401 path without making a reservation. This functionality can be used 402 to 'probe' the network for path characteristics, for 403 receiver-initiated reservations or for support of certain QoS 404 models. The information obtained from a QUERY may be used in the 405 admission control process of a QNE (e.g. in case of 406 measurement-based admission control). Note that a QUERY does not 407 change existing reservation state. It does not cause QoS NSLP 408 state to be installed in nodes other than the one that generated 409 the QUERY. 411 RESPONSE: The RESPONSE message is used to provide information about 412 the result of a previous QoS-NSLP message. This includes explicit 413 confirmation of the state manipulation signaled in the RESERVE 414 message, the response to a QUERY message or an error code if the 415 QNE or QNR is unable to provide the requested information or if 416 the response is negative. The RESPONSE message is impotent, it 417 does not cause any reservation state to be installed or modified. 419 NOTIFY: NOTIFY messages are used to convey information to a QNE. 420 They differ from RESPONSE messages in that they are sent 421 asynchronously and need not refer to any particular state or 422 previously received message. The information conveyed by a NOTIFY 423 message is typically related to error conditions. Examples would 424 be notification to an upstream peer about state being torn down or 425 to indicate when a reservation has been pre-empted. 427 QoS-NSLP messages are sent peer-to-peer. This means that a QNE 428 considers its adjacent upstream or downstream peer to be the source 429 of the each message. 431 Each protocol message has a common header which indicates the message 432 type and contains flags. Message formats are defined in 433 Section 5.1.2. Message processing rules are defined in Section 5.4. 435 QoS NSLP messages contain three types of objects: 436 Control Information: Control information objects carry general 437 information for the QoS NSLP processing, such as sequence numbers 438 or whether a response is required. 440 QoS specifications (QSPECs): QSPEC objects describe the actual 441 resources that are required and depend on the QoS model being 442 used. Besides any resource description they may also contain 443 other control information used by the RMF's processing. 445 Policy objects: Policy objects contain data used to authorise the 446 reservation of resources. 448 Object formats are defined in Section 5.1.3. Object processing rules 449 are defined in Section 5.3. 451 3.1.3 QoS Models and QoS Specifications 453 QoS-NSLP provides flexibility over the exact patterns of signalling 454 messages that are exchanged. The decoupling of QoS-NSLP and QSpec 455 allows the QoS-NSLP to be ignorant about the ways in which traffic, 456 resources, etc are described, and it can treat the QSpec as an opaque 457 object. 459 The QSpec fulfils a similar purpose to the TSpec, RSpec and AdSpec 460 objects used with RSVP and specified in RFC 2205 [RFC2205] and RFC 461 2210 [RFC2210]. At each QNE, its content is interpreted by the 462 Resource Management Function and the Policy Control Function for the 463 purposes of policy control and traffic control (including admission 464 control and configuration of the packet classifier and scheduler). 466 QoS-NSLP does not mandate any particular behaviour for the RMF, 467 instead demanding interoperability at the signalling protocol whilst 468 leaving the validation of RMF behaviour to SLAs or contracts external 469 to the protocol itself. 471 The QSpec carries a collection of objects that can describe QoS 472 specifications in a number of different ways. A QSpec will usually 473 contain some objects which need to be understood by all 474 implementations, and it can also be enhanced with additional objects 475 to provide a more exact definition to the RMF, which may be better 476 able to use its specific resource management mechanisms (which may, 477 for example, be link specific) as a result. 479 A QoS Model defines the behavior of the RMF, including inputs and 480 outputs, and how QSpec information is used to describe resources 481 available, resources required, traffic descriptions, and control 482 information required by the RMF. A QoS Model also describes the 483 minimum set of parameters QNEs should use in the QSpec when signaling 484 about this QoS Model. 486 QoS Models may be local (private to one network), 487 implementation/vendor specific, or global (implementable by different 488 networks and vendors). The authors are currently aware of three 489 efforts related to QoS Model specification: IntServ Controlled Load 490 [I-D.kappler-nsis-qosmodel-controlledload], based on ITU Y.1541 491 [I-D.ash-nsis-y1541-qsp] and Resource Management for DiffServ (RMD) 492 [I-D.ietf-nsis-rmd]. 494 The definition of a QoS model may also have implications on how local 495 behaviour should be implemented in the areas where the QoS NSLP gives 496 freedom to implementors. For example, it may be useful to identify 497 recommended behaviour for how a RESERVE message that is forwarded 498 relates to that received, or when additional signalling sessions 499 should be started based on existing sessions, such as required for 500 aggregate reservations. In some cases, suggestions may be made on 501 whether state that may optionally be retained should be held in 502 particular scenarios. 504 An ongoing effort attempts to specify a QSPEC template 505 [I-D.ietf-nsis-qspec]. The QSPEC template contains object formats 506 for generally useful elements of the QoS description, which is 507 designed to ensure interoperability when using the basic set of 508 objects. 510 3.1.4 Policy control 512 Getting access to network resources typically involves some kind of 513 policy control. One example of this is authorisation of the resource 514 requester. Policy control for QoS NSLP resource reservation 515 signalling is conceptually organised as illustrated below. 517 +-------------+ 518 | Policy | 519 | Decision | 520 | Point (PDP) | 521 +------+------+ 522 | 523 | 524 /-\-----+-----/\ 525 //// \\\\ 526 || || 527 | Policy transport | 528 || || 529 \\\\ //// 530 \-------+------/ 531 | 532 +-------------+ QoS signalling +------+------+ 533 | Entity |<==============>| QNE = Policy|<=========> 534 | requesting | Data Flow | Enforcement | 535 | resource |----------------|-Point (PEP)-|----------> 536 +-------------+ +-------------+ 538 From QoS NSLP point of view, the policy control model is essentially 539 a two-party model between neighbouring QNEs. The actual policy 540 decision may depend on the involvement of a third entity (the policy 541 decision point), but this happens outside of the QoS NSLP protocol by 542 means of existing policy infrastructure (COPS, Diameter, etc). The 543 policy control model for the entire end-to-end chain of QNEs is 544 therefore one of transitivity, where each of the QNEs exchanges 545 policy information with its QoS NSLP policy peer. 547 The input to policy control is referred to as "Policy data", some of 548 which QoS NSLP carries in the POLICY_DATA object while other 549 information is provided across the GIMPS API. Policy data itself is 550 opaque to the QoS NSLP, which simply passes it to policy control when 551 required. The policy data is independent from the QoS model in use. 553 Three options have currently been suggested for support by QoS-NSLP 554 policy control: 555 a. Reuse of GIMPS channel security mechanisms 556 b. Carrying (authorisation) tokens 557 c. Encapsulating EAP exchanges in the QoS NSLP protocol 559 Option a is the preferred option. Option b is also supported in this 560 version of the specification. Option c is not considered at this 561 moment in time, but could potentially be added at a later date 562 through the use of extension mechanisms already available in the 563 protocol.. 565 It is generally assumed that policy enforcement is likely to 566 concentrate on border nodes between administrative domains. In some 567 cases policy objects transmitted across the domain traverse an 568 intermediate Policy Ignorant Node (PIN) that is allowed to process 569 QoS NSLP message but does not handle policy information. The policy 570 peering between ingress and egress edge of a domain therefore relies 571 on the internal chain of trust between the nodes in the domain. If 572 this is not acceptable, a separate signalling session can be set up 573 between the edge node in order to exchange policy information. This 574 is similar to the aggregation mechanism. 576 3.2 Design decisions 578 QoS NSLP was designed according to the principles and supports the 579 functionality outlined below. 581 3.2.1 Soft-state 583 The NSIS protocol suite takes a soft-state approach to state 584 management. This means that reservation state in QNEs must be 585 periodically refreshed. The frequency with which state installation 586 is refreshed is expressed in the REFRESH_PERIOD object. This object 587 contains a value in milliseconds indicating how long the state that 588 is signalled for remains valid. Maintaining the reservation beyond 589 this lifetime can be done by sending a ("refreshing") RESERVE 590 message. 592 3.2.2 Sender-receiver initiation 594 QoS NSLP supports both sender-initiated and receiver-initiated 595 reservations. For a sender-initiated reservation, RESERVE messages 596 travel in the same direction as the dataflow that is being signalled 597 for (the QNI is at the side of the source of the dataflow). For a 598 receiver-initiated reservation, RESERVE messages travel in the 599 opposite direction (the QNI is at the side of the receiver of the 600 data flow) 602 3.2.3 Message sequencing 604 RESERVE messages affect the installed reservation state. Unlike 605 NOTIFY, QUERY and RESPONSE messages, the order in which RESERVE 606 messages are received influences the eventual reservation state that 607 will be stored at a QNE. Therefore, QoS NSLP supports detection of 608 RESERVE message re-ordering or duplication with Reservation Sequence 609 Number (RSN). 611 The RSN has local significance only, i.e. between QNEs. Attempting 612 to make an identifier that was unique in the context of a SESSION_ID 613 but the same along the complete path would be very hard. Since 614 RESERVE messages can be sent by any node on the path that maintains 615 reservation state (e.g. for path repair) we would have the difficult 616 task of attempting to keep the identifier synchronized along the 617 whole path. Since message ordering only ever matters between a pair 618 of peer QNEs, we can make the RSN unique just between a pair of 619 neighbouring stateful QNEs. By managing the sequence numbers in this 620 manner, the source of the RESERVE does not need to determine how the 621 next QNE will process the message. 623 Note that, since the RSN is unique within a SESSION_ID, it can be 624 used together with a SESSION_ID to refer to particular installed 625 state. 627 3.2.4 Explicit state installation confirmation and responses 629 A QoS-NSLP instance MAY request an explicit confirmation of its state 630 installation actions from the immediate upstream or downstream peer. 631 This is achieved by using an ACKNOWLEDGE (A) flag in the message 632 header. 634 In addition to this, a QNE may require other information such as a 635 confirmation that the end-to-end reservation is in place or a reply 636 to a query along the path. For such requests, it must be able to 637 keep track of which request each response refers to. This is 638 supported by including a Request Identification Information (RII) 639 object in a QoS NSLP message. 641 3.2.5 Summary refreshes 643 For scalability, QoS NSLP supports an abbreviated form of refreshing 644 RESERVE message ("summary refresh"). In this case, the refreshing 645 RESERVE references the reservation using the RSN and the SESSION_ID, 646 rather than including the full reservation specification (including 647 QSPEC, ...). Summary refreshes require an explicit acknowledgment of 648 state installation to ensure that the RSN reference will be 649 understood. It is up to a QNE that receives a message containing an 650 RII to decide whether it wants to accept summary refreshes and 651 provide this explicit acknowledgment. 653 3.2.6 Message scoping 655 A QNE may use local policy when deciding whether to propagate a 656 message or not. The QoS NSLP also includes an explicit mechanism to 657 restrict message propagation by means of a scoping mechanism. 659 For a RESERVE or a QUERY message, a SCOPING flag limits the part of 660 the path on which state is installed or the downstream nodes that can 661 respond. When set to zero, it indicates that the scope is "whole 662 path" (default). When set to one, the scope is "single hop". 664 The propagation of a RESPONSE message is limited by the RII object, 665 which ensures that it is not forwarded back along the path further 666 than the node that requested the RESPONSE. 668 This specification does not support an explicit notion of a region 669 scope or "to a mobility-related path branch/merge point". If needed, 670 this can be easily proposed as an extension later on,e.g. based on 671 LRSVP [I-D.manner-lrsvp]. 673 3.2.7 Session binding 675 Session binding is defined as the enforcement of a relation between 676 different QoS NSLP sessions (i.e. signalling flows with different 677 SESSION_ID (SID) as defined in GIMPS [I-D.ietf-nsis-ntlp]). 679 Session binding indicates a (possibly asymmetric) dependency relation 680 between two or more sessions by including a BOUND_SESSION_ID object. 681 A session with SID_A (the binding session) can express its relation 682 to another session with SID_B (the bound session) by including a 683 BOUND_SESSION_ID object containing SID_B in its messages. The 684 dependency is asymmetric if the session with SID_B does not carry a 685 BOUND_SESSION_ID object containing SID_A. 687 The concept of session binding is used to indicate the dependency 688 between the end-to-end session and the aggregate session in case of 689 aggregate reservations. In case of bidirectional reservations, it is 690 used to express the dependency between the sessions used for forward 691 and reverse reservation. Note that the dependency indicated by 692 session binding is purely informative in nature and does not 693 automatically trigger any action in a QNE. However, a QNE may use 694 the information for local resource optimisation or to tear down 695 reservations that are no longer useful. 697 3.2.8 Layering 699 QoS NSLP supports layered reservations. Layered reservations may 700 occur when certain parts of the network (domains) implement one or 701 more local QoS models, or when they locally apply specific control 702 plane characteristics (e.g. GIMPS unreliable transfer mode instead 703 of reliable transfer mode). They may also occur when several 704 per-flow reservations are locally combined into an aggregate 705 reservation. 707 3.2.8.1 Local QoS models 709 A domain may have local policies regarding QoS model implementation, 710 i.e. it may map incoming traffic to its own locally defined QoS 711 models. QoS NSLP supports this by allowing QSPEC objects to be 712 stacked. 714 When a domain wants to apply a certain QoS model to an incoming 715 per-flow reservation request, each edge of the domain is configured 716 to map the incoming QSPEC object to a local QSPEC object and push 717 that object onto the stack of QSPEC objects (typically immediately 718 following the Common Control Information, i.e. the first QSPEC that 719 is found in the message). QNEs inside the domain look at the top of 720 the QSPEC object stack to determine which QoS model to apply for the 721 reservation. 723 The position of the local QSPEC object in the stack implies a 724 tradeoff between the speed with which incoming messages can be 725 processed and the time it takes to construct the outgoing message (if 726 any). By mandating the locally valid object to be on top of the 727 stack we value ease of processing over ease of message construction. 729 3.2.8.2 Local control plane properties 731 The way signalling messages are handled is mainly determined by the 732 parameters that are sent over GIMPS-NSLP API and by the Common 733 Control Information. A domain may have a policy to implement local 734 control plane behaviour. It may, for instance, elect to use an 735 unreliable transport locally in the domain while still keeping 736 end-to-end reliability intact. 738 The QoS NSLP supports this situation by allowing two sessions to be 739 set up for the same reservation. The local session has the desired 740 local control plane properties and is interpreted in internal QNEs. 741 This solution poses two requirements: the end-to-end session must be 742 able to bypass intermediate nodes and the egress QNE needs to bind 743 both sessions together. 745 Intermediate node bypass is achieved with GIMPS. The local session 746 and the end-to-end session are bound at the egress QNE by means of 747 the BOUND_SESSION_ID object. 749 3.2.8.3 Aggregate reservations 751 In some cases it is desirable to create reservations for an 752 aggregate, rather than on a per-flow basis, in order to reduce the 753 amount of reservation state needed as well as the processing load for 754 signalling messages. The QoS NSLP, therefore, provides aggregation 755 facilities similar to RFC 3175 [RFC3175]. However, the aggregation 756 scenarios supported are wider than that proposed there. Note that 757 QoS NSLP does not specify how reservations need to be combined in an 758 aggregate or how end-to-end properties need to be computed but only 759 provides signalling support for it. 761 The essential difference with the layering approaches described in 762 Section 3.2.8.1 and Section 3.2.8.2 is that the aggregate reservation 763 needs a FlowID that describes all traffic carried in the aggregate 764 (e.g. a DSCP in case of IntServ over DiffServ). The need for a 765 different FlowID mandates the use of two different sessions, similar 766 to Section 3.2.8.2 and to the RSVP aggregation solution RFC 3175 767 [RFC3175]. 769 Edge QNEs of the aggregation domain that want to maintain some 770 end-to-end properties may establish a peering relation by sending the 771 end-to-end message transparantly over the domain (using the 772 intermediate node bypass capability described above). Updating the 773 end-to-end properties in this message may require some knowledge of 774 the aggregated session (e.g. for updating delay values). For this 775 purpose, the end-to-end session contains a BOUND_SESSION_ID carrying 776 the SESSION_ID of the aggregate session. 778 3.2.9 Priority 780 This specification acknowledges the fact that in some situations, 781 some messages or some reservations may be more important than others 782 and therefore foresees mechanisms to give these messages or 783 reservations priority. 785 Priority of certain signalling messages over others may be required 786 in mobile scenarios when a message loss during call set-up is less 787 harmful then during handover. This situation only occurs when GIMPS 788 or QoS NSLP processing is the congested part or scarce resource. 789 This specification requests GIMPS design to foresee a mechanism to 790 support a number of levels of message priority that can be requested 791 over the NSLP-GIMPS API. 793 Priority of certain reservations over others may be required when QoS 794 resources are oversubscribed. In that case, existing reservations 795 may be preempted in order to make room for new higher-priority 796 reservations. A typical approach to deal with priority and 797 preemption is through the specification of a setup priority and 798 holding priority for each reservation. The resource management 799 function at each QNE then keeps track of the resource consumption at 800 each priority level. Reservations are established when resources, at 801 their setup priority level, are still available. They may cause 802 preemption of reservations with a lower holding priority than their 803 setup priority. 805 Support of reservation priority is provided by a QSpec parameter and 806 therefore outside the scope of this specification. 808 3.2.10 Rerouting 810 QoS NSLP needs to adapt to route changes in the data path. This 811 assumes the capability to detect rerouting events, perform QoS 812 reservation on the new path and optionally tear down reservations on 813 the old path. 815 Rerouting detection can be performed at three levels. First, routing 816 modules may detect route changes through their interaction with 817 routing protocols. Certain QNEs or GIMPS implementations may 818 interact with local routing module to receive quick notification of 819 route changes. This is largely implementation-specific and outside 820 of the scope of NSIS. Second, route changes may be detected at GIMPS 821 layer. This specification requests GIMPS design to foresee 822 notification of this information over the API. This is outside the 823 scope of the QoS NSLP specification. Third, rerouting may be 824 detected at the NSLP layer. A QoS NSLP node is able to detect 825 changes in its QoS NSLP peers by keeping track of a Source 826 Identification Information (SII) object that is similar in nature to 827 the RSVP_HOP object described in RFC 2205 [RFC2205]. When a RESERVE 828 message with an existing SESSION_ID and a different SII is received, 829 the QNE knows its upstream peer has changed. 831 Reservation on the new path happens when a refreshing RESERVE message 832 arrives at the QNE where the old and the new path diverge. The 833 refreshing RESERVE will be interpreted as a new RESERVE on the new 834 path. Depending on the transfer mode, this may require installation 835 of a new messaging association. Rapid recovery at the NSLP layer 836 therefore requires short refresh periods. Detection before the next 837 RESERVE message arrives is only possible at the IP layer or through 838 monitoring of GIMPS peering relations (e.g. by TTL counting the 839 number of GIMPS hops between NSLP peers or the observing changes in 840 the outgoing interface towards GIMPS peer). These mechanisms can 841 provide implementation specific optimisations, and are outside the 842 scope of this specification. 844 When the QoS NSLP is aware of the route change, it needs to set up 845 the reservation on the new path. This is done by incrementing the 846 RSN and then sending a new RESERVE message. On links that are common 847 to the old and the new path, this RESERVE message is interpreted as a 848 refreshing RESERVE. On new links, it creates the reservation. 850 After the reservation on the new path is set up, the branching node 851 or the merging node may want to tear down the reservation on the old 852 path (faster than what would result from normal soft-state time-out). 853 This functionality is supported by keeping track of the old SII. 854 This specification requests GIMPS design to provide support for an 855 SII that is interpreted as a random identifier at the QoS NSLP but 856 that allows, when passed over the API, to forward QoS NSLP messages 857 to the QNE identified by that SII. 859 A QNI or a branch node may wish to keep the reservation on the old 860 branch. This could for instance be the case when a mobile node has 861 experienced a mobility event and wishes to keep reservation to its 862 old attachment point in case it moves back there. For this purpose, 863 a REPLACE flag is foreseen in the common header, which, when set to 864 FALSE, indicates that the reservation on the old branch should be 865 kept. 867 4. Examples of QoS NSLP Operation 869 The QoS NSLP can be used in a number of ways. The examples given 870 here give an indication of some of the basic processing. However, 871 they are not exhaustive and do not attempt to cover the details of 872 the protocol processing. 874 4.1 Basic sender-initiated reservation 876 QNI QNE QNE QNR 877 | | | | 878 | RESERVE | | | 879 +--------->| | | 880 | | RESERVE | | 881 | +--------->| | 882 | | | RESERVE | 883 | | +--------->| 884 | | | | 885 | | | RESPONSE | 886 | | |<---------+ 887 | | RESPONSE | | 888 | |<---------+ | 889 | RESPONSE | | | 890 |<---------+ | | 891 | | | | 892 | | | | 894 Figure 4: Basic Sender Initiated Reservation 896 To make a new reservation, the QNI constructs a RESERVE message 897 containing a QSPEC object, from its chosen QoS model, which describes 898 the required QoS parameters. 900 The RESERVE message is passed to GIMPS which transports it to the 901 next QNE. There it is delivered to the QoS NSLP processing which 902 examines the message. Policy control and admission control decisions 903 are made. The exact processing also takes into account the QoS model 904 being used. The node performs appropriate actions (e.g. installing 905 reservation) based on the QSPEC object in the message. 907 The QoS NSLP then generates a new RESERVE message (usually based on 908 the one received). This is passed to GIMPS, which forwards it to the 909 next QNE. 911 The same processing is performed at further QNEs along the path, up 912 to the QNR. The determination that a node is the QNR may be made 913 directly (e.g. that node is the destination for the data flow), or 914 using some GIMPS functionality to determine that there are no more 915 QNEs between this node and the data flow destination. 917 A node can ask a confirmation of the installed state from its 918 immediate peer. It does so by setting the A flag, which causes a 919 RESPONSE message to be sent by the immediate peer. One use of this 920 is to confirm the installation of state, which allows the use of 921 summary refreshes that later refer to that state. A RESPONSE message 922 can also indicate an error when, for example, a reservation has 923 failed to be installed. 925 Any node may include a request for a RESPONSE in its RESERVE 926 messages. It does so by including a Request Identification 927 Information (RII) object in the RESERVE message. The RESPONSE is 928 forwarded peer-to-peer along the reverse of the path that the RESERVE 929 message took (using GIMPS path state), and so is seen by all the QNEs 930 on the reverse-path. It is only forwarded as far as the node which 931 requested the RESPONSE. 933 The reservation can subsequently be refreshed by sending further 934 RESERVE messages containing the complete reservation information, as 935 for the initial reservation. The reservation can also be modified in 936 the same way, by changing the QSpec data to indicate a different set 937 of resources to reserve. 939 The overhead required to perform refreshes can be reduced, in a 940 similar way to that proposed for RSVP in RFC 2961 [RFC2961]. Once a 941 RESPONSE message has been received indicating the successful 942 installation of a reservation, subsequent refreshing RESERVE messages 943 can simply refer to the existing reservation, rather than including 944 the complete reservation specification. 946 4.2 Sending a Query 948 QUERY messages can be used to gather information from QNEs along to 949 path. For example, it can be used to find out what resources are 950 available before a reservation is made. 952 In order to perform a query along a path, the QNE constructs a QUERY 953 message. This message includes a QSpec containing the actual query 954 to be performed at QNEs along the path. It also contains an object 955 used to match the response back to the query, and an indicator of the 956 query scope (next node, whole path). 958 The QUERY message is passed to GIMPS to forward it along the path. 960 A QNE (including the QNR) receiving a QUERY message should inspect it 961 and create a new message, based on that received with the query 962 objects modified as required. For example, the query may request 963 information on whether a flow can be admitted, and so a node 964 processing the query might record the available bandwidth. The new 965 message is then passed to GIMPS for further forwarding (unless it 966 knows it is the QNR, or is the limit for the scope in the QUERY). 968 At the QNR, a RESPONSE message must be generated if the QUERY message 969 includes a Request Identification Information (RII) object. Into 970 this is copied various objects from the received QUERY message. It 971 is then passed to GIMPS to be forwarded peer-to-peer back along the 972 path. 974 Each QNE receiving the RESPONSE message should inspect the RII object 975 to see if it 'belongs' to it (i.e. it was the one that originally 976 created it). If it does not then it simply passes the message back 977 to GIMPS to be forwarded back down the path. 979 4.3 Basic receiver-initiated reservation 981 As described in the NSIS framework [I-D.ietf-nsis-fw] in some 982 signalling applications, a node at one end of the data flow takes 983 responsibility for requesting special treatment - such as a resource 984 reservation - from the network. Both ends then agree whether sender 985 or receiver-initiated reservation is to be done. In case of a 986 receiver initiated reservation, both ends agree whether a "One Pass 987 With Advertising" (OPWA) [_XREF_OPWA95] model is being used. This 988 negotiation can be accomplished using mechanisms that are outside the 989 scope of NSIS, see Section 9.2. 991 To make a receiver-initiated reservation, the QNI constructs a QUERY 992 message, which may contain a QSPEC object from its chosen QoS model 993 (see Figure 5). This QUERY message does not need to trigger a 994 RESPONSE message and therefore, the QNI must not include the RII 995 object (Section 5.4.2), into the QUERY message. The QUERY message 996 may be used to gather information along the path, which is carried by 997 the QSPEC object. An example of such information is the "One Pass 998 With Advertising" (OPWA) [_XREF_OPWA95]. This QUERY message causes 999 GIMPS reverse-path state to be installed. 1001 QNR QNE QNE QNI 1002 sender receiver 1003 | | | | 1004 | QUERY | | | 1005 +--------->| | | 1006 | | QUERY | | 1007 | +--------->| | 1008 | | | QUERY | 1009 | | +--------->| 1010 | | | | 1011 | | | RESERVE | 1012 | | |<---------+ 1013 | | RESERVE | | 1014 | |<---------+ | 1015 | RESERVE | | | 1016 |<---------+ | | 1017 | | | | 1018 | RESPONSE | | | 1019 +--------->| | | 1020 | | RESPONSE | | 1021 | +--------->| | 1022 | | | RESPONSE | 1023 | | +--------->| 1024 | | | | 1026 Figure 5: Basic Receiver Initiated Reservation 1028 The QUERY message is transported by GIMPS to the next downstream QoS 1029 NSLP node. There it is delivered to the QoS NSLP processing which 1030 examines the message. The exact processing also takes into account 1031 the QoS model being used and may include gathering information on 1032 path characteristics that may be used to predict the end-to-end QoS. 1034 The QoS NSLP then generates a new QUERY message (usually based on the 1035 one received). This is passed to GIMPS, which forwards it to the 1036 next QNE. The same processing is performed at further QNEs along the 1037 path, up to the receiver, which in this situation is the QNR. The 1038 QNR detects that this QUERY message does not carry an RII object and 1039 by using the information contained in the received QUERY message, 1040 such as the QSPEC, constructs a RESERVE message. 1042 The RESERVE is forwarded peer-to-peer along the reverse of the path 1043 that the QUERY message took (using GIMPS reverse path state). 1044 Similar to the sender-initiated approach, any node may include an RII 1045 in its RESERVE messages. 1047 The reservation can subsequently be refreshed in the same way as for 1048 the sender-initiated approach. This RESERVE message may be also used 1049 to refresh GIMPS reverse path state. Alternatively, refreshing GIMPS 1050 reverse path state could be performed by sending periodic QUERY 1051 messages, which are needed in case of route changes anyway. 1053 4.4 Bidirectional Reservations 1055 Bidirectional reservations are supported by binding two 1056 uni-directional sessions together. We distinguish two cases: 1057 o Binding two sender-initiated reservations, e.g. one 1058 sender-initiated reservation from QNE A to QNE B and another one 1059 from QNE B to QNE A. 1060 o Binding a sender-intiated and a receiver-initiated reservation, 1061 e.g. a sender-initiated reservation from QNE A towards QNE B, and 1062 a receiver-initiated reservation from QNE A towards QNE B for the 1063 data flow in the opposite direction (from QNE B to QNE A). This 1064 case is particularly useful when one end of the communication has 1065 all required information to set up both sessions. 1067 Both ends have to agree on which bi-directional reservation type they 1068 need to use. This negotiation/agreement can be accomplished using 1069 mechanisms that are outside the scope of NSIS, see Section 9.2. 1071 The scenario with two sender-initiated reservation is shown on 1072 Figure 6. Note that RESERVE messages for both directions may visit 1073 different QNEs along the path because of asymmetric routing. Both 1074 directions of the flows are bound by inserting the BOUND_SESSION_ID 1075 object at the QNI and QNR. RESPONSE messages are optional and not 1076 shown on the picture for simplicity. 1078 A QNE QNE B 1079 | | FLOW-1 | | 1080 |===============================>| 1081 |RESERVE-1 | | | 1082 QNI+--------->|RESERVE-1 | | 1083 | +-------------------->|QNR 1084 | | | | 1085 | | FLOW-2 | | 1086 |<===============================| 1087 | | |RESERVE-2 | 1088 | RESERVE-2 |<---------+QNI 1089 QNR|<--------------------+ | 1090 | | | | 1092 Figure 6: Bi-directional reservation for sender+sender scenario 1094 The scenario with a sender-initiated and a receiver-initiated 1095 reservation is shown on Figure 7. In this case, QNI B sends out two 1096 RESERVE messages, one for the sender-initiated and one for the 1097 receiver-initiated reservation. 1099 A QNE QNE B 1100 | | FLOW-1 | | 1101 |===============================>| 1102 | QUERY-1 | | | 1103 QNI+--------->| QUERY-1 | | 1104 | +-------------------->|QNR 1105 | | | | 1106 |RESERVE-1 | | | 1107 QNI+<---------|RESERVE-1 | | 1108 | +<--------------------|QNR 1109 | | | | 1110 | | FLOW-2 | | 1111 |<===============================| 1112 | | |RESERVE-2 | 1113 |RESERVE-2 | |<---------+QNI 1114 QNR|<--------------------+ | 1115 | | | | 1117 Figure 7: Bi-directional reservation for sender+receiver scenario 1119 4.5 Use of Local QoS Models 1121 In some cases it may be required to use a different QoS model along a 1122 particular segment of the signalling. In this case a node at the 1123 edge of this region needs to map between the two resource 1124 descriptions (and any auxiliary data). 1126 +-------- QoSM2 domain -------+ 1127 | | 1128 | | 1129 +----+ +----+ +----+ +----+ +----+ 1130 |QNI | |edge| |int.| |edge| |QNR | 1131 | |========>|QNE |========>|QNE |========>|QNE |========>| | 1132 +----+ RESERVE +----+ RESERVE +----+ RESERVE +----+ RESERVE +----+ 1133 QSPEC1 | QSPEC2 QSPEC2 | QSPEC1 1134 | {QSPEC1} {QSPEC1} | 1135 | | 1136 +-----------------------------+ 1138 Figure 8: Reservation with local QoS Models 1140 This initially proceeds as for the basic example, with peer-to-peer 1141 installation of reservations. However, within a region of the 1142 network a different QoSM (QoSM2) needs to be used. At the edge of 1143 this region the QNEs support both the end-to-end and local QoS 1144 models. When the RESERVE message reaches the QNE at the ingress, the 1145 initial processing of the RESERVE proceeds as normal. However, the 1146 QNE also determines the appropriate description using QoSM2. The 1147 RESERVE message to be sent out is constructed mostly as usual but 1148 with a second QSPEC object added on top, which becomes the 'current' 1149 one. 1151 When this RESERVE message is received at an node internal to the 1152 QoSM2 domain the QoS NSLP only uses the QSPEC at the top of the stack 1153 (i.e. the 'current' one), rather than the end-to-end QSPEC. 1154 Otherwise, processing proceeds as usual. The RESERVE message that it 1155 generates should include the complete stack of QSPECs from the 1156 message it received. 1158 At the QNE at the egress of the region the local QSPEC is removed 1159 from the message so that subsequent QNEs receive only the end-to-end 1160 QSPEC. 1162 QSPECs can be stacked in this way to an arbitrary depth. 1164 4.6 Aggregate Reservations 1166 In order to reduce signalling and per-flow state in the network, the 1167 reservations for a number of flows may be aggregated together. 1169 QNI QNE QNE/QNI' QNE' QNR'/QNE QNR 1170 aggregator deaggregator 1171 | | | | | | 1172 | RESERVE | | | | | 1173 +--------->| | | | | 1174 | | RESERVE | | | | 1175 | +--------->| | | | 1176 | | | RESERVE | | | 1177 | | +-------------------->| | 1178 | | | RESERVE' | | | 1179 | | +=========>| RESERVE' | | 1180 | | | +=========>| RESERVE | 1181 | | | | +--------->| 1182 | | | | RESPONSE'| | 1183 | | | RESPONSE'|<=========+ | 1184 | | |<=========+ | | 1185 | | | | | RESPONSE | 1186 | | | | RESPONSE |<---------+ 1187 | | |<--------------------+ | 1188 | | RESPONSE | | | | 1189 | |<---------+ | | | 1190 | RESPONSE | | | | | 1191 |<---------+ | | | | 1192 | | | | | | 1193 | | | | | | 1195 Figure 9: Sender Initiated Reservation with Aggregation 1197 An end-to-end per-flow reservation is initiated as normal (with 1198 messages shown in Figure 9 as "RESERVE"). 1200 At the aggregator a reservation for the aggregated flow is initiated 1201 (shown in Figure 9 as "RESERVE'"). This may use the same QoS model 1202 as the end-to-end reservation but has a flow identifier for the 1203 aggregated flow (e.g. tunnel) instead of for the individual flows. 1204 This document does not specify how the QSPEC of the aggregate session 1205 can be derived from the QSPECs of the end-to-end sessions. 1207 Markings are used so that intermediate routers do not need to inspect 1208 the individual flow reservations. The deaggregator then becomes the 1209 next hop QNE for the end-to-end per-flow reservation. 1211 Aggregator Deaggregator 1213 +---+ +---+ +---+ +---+ 1214 |QNI|-----|QNE|-----|QNE|-----|QNR| aggregate 1215 +---+ +---+ +---+ +---+ reservation 1217 +---+ +---+ ..... ..... +---+ +---+ 1218 |QNI|-----|QNE|-----. .-----. .-----|QNE|-----|QNR| end-to-end 1219 +---+ +---+ ..... ..... +---+ +---+ reservation 1221 The deaggregator acts as the QNR for the aggregate reservation. 1223 Information is carried in the reservations to enable the deaggregator 1224 to associate the end-to-end and aggregate reservations with one 1225 another. 1227 The key difference between this example, and previous ones is that 1228 the flow identifier for the aggregate is expected to be different to 1229 that for the end-to-end reservation. The aggregate reservation can 1230 be updated independently of the per-flow end-to-end reservations. 1232 4.7 Reduced State or Stateless Interior Nodes 1234 This example uses a different QoS model within a domain, in 1235 conjunction with GIMPS and NSLP functionality which allows the 1236 interior nodes to avoid storing GIMPS and QoS NSLP state. As a 1237 result the interior nodes only store the QSpec-related reservation 1238 state, or even no state at all. This allows the QoS model to use a 1239 form of "reduced-state" operation, where reservation states with a 1240 coarser granularity (e.g. per-class) are used, or a "stateless" 1241 operation where no QoS NSLP state is needed (or created). 1243 The key difference between this example and the use of different QoS 1244 models in Section 4.5 is that the transport characteristics for the 1245 'local' reservation can be different from that of the end-to-end 1246 reservation, i.e. GIMPS can be used in a different way for the 1247 edge-to-edge and hop-by-hop sessions. The reduced state reservation 1248 can be updated independently of the per-flow end-to-end reservations. 1250 QNE QNE QNE QNE 1251 ingress interior interior egress 1252 GIMPS stateful GIMPS stateless GIMPS stateless GIMPS stateful 1253 | | | | 1254 RESERVE | | | | 1255 -------->| RESERVE | | | 1256 +--------------------------------------------->| 1257 | RESERVE' | | | 1258 +-------------->| | | 1259 | | RESERVE' | | 1260 | +-------------->| | 1261 | | | RESERVE' | 1262 | | +------------->| 1263 | | | | RESERVE 1264 | | | +--------> 1265 | | | | RESPONSE 1266 | | | |<-------- 1267 | | | RESPONSE | 1268 |<---------------------------------------------+ 1269 RESPONSE| | | | 1270 <--------| | | | 1272 Figure 11: Sender-initiated reservation with Reduced State Interior 1273 Nodes 1275 The QNI performs the same processing as before to generate the 1276 initial RESERVE message, and it is forwarded by GIMPS as usual. At 1277 the QNEs at the edges of the stateless or reduced-state region the 1278 processing is different and the nodes support two QoS models. 1280 At the ingress the original RESERVE message is forwarded but ignored 1281 by the stateless or reduced-state nodes. The egress node is the next 1282 QoS NSLP hop for that session. After the initial discovery phase 1283 using unreliable GIMPS transfer mode, reliable GIMPS transfer mode 1284 between the ingress and egress can be used. At the egress node the 1285 RESERVE message is then forwarded normally. 1287 At the ingress a second RESERVE' message is also built. This makes 1288 use of a QoS model suitable for a reduced state or stateless form of 1289 operation (such as the RMD per hop reservation). Since the original 1290 RESERVE and the RESERVE' messages are addressed identically, RESERVE' 1291 visits the same nodes that were visited, including the egress QNE. 1293 When processed by interior (stateless) nodes the QoS NSLP processing 1294 excercises its options to not keep state wherever possible, so that 1295 no per flow QoS NSLP state is stored. Some state, e.g. per class, 1296 for the QSpec related data may be held at these interior nodes. The 1297 QoS NSLP also requests that GIMPS use different transport 1298 characteristics (i.e. sending of messages in unreliable GIMPS 1299 transfer mode). It also requests the local GIMPS processing not to 1300 retain messaging association state or reverse message routing state. 1302 Nodes, such as those in the interior of the stateless or 1303 reduced-state domain, that do not retain reservation state cannot 1304 send back RESPONSE messages (and so cannot use summary refreshes). 1306 At the egress node the RESERVE' message is interpreted in conjunction 1307 with the reservation state from the end-to-end RESERVE message (using 1308 information carried in the message to correlate the signalling 1309 flows). The RESERVE message is only forwarded further if the 1310 processing of the RESERVE' message was successful at all nodes in the 1311 local domain, otherwise the end-to-end reservation is regarded as 1312 having failed to be installed. 1314 Since GIMPS neighbour relations are not maintained in the 1315 reduced-state region, only sender initiated signalling can be 1316 supported. If a receiver-initiated reservation over a stateless or 1317 reduced state domain is required this can be implemented as shown 1318 below. 1320 QNE QNE QNE 1321 ingress interior egress 1322 GIMPS stateful GIMPS stateless GIMPS stateful 1323 | | | 1324 QUERY | | | 1325 -------->| QUERY | | 1326 +------------------------------>| 1327 | | | QUERY 1328 | | +--------> 1329 | | | RESERVE 1330 | | |<-------- 1331 | | RESERVE | 1332 |<------------------------------+ 1333 | RESERVE | RESERVE | 1334 |-------------->|-------------->| 1335 RESERVE | | | 1336 <--------| | | 1338 Figure 12: Receiver-initiated reservation with Reduced State Interior 1339 Nodes 1341 The RESERVE message that is received by the egress QNE of the 1342 stateless domain is sent transparantly to the ingress QNE (known as 1343 the source of the QUERY message). When the RESERVE message reaches 1344 the ingress, the ingress QNE knows it needs to send both a 1345 sender-initiated RESERVE over the stateless domain and send a RESERVE 1346 message further upstream. 1348 4.8 Re-routing scenario 1350 The QoS NSLP needs to adapt to route changes in the data path. This 1351 assumes the capability to detect rerouting events, perform QoS 1352 reservation on the new path and optionally tear down reservations on 1353 the old path. 1355 When the QoS NSLP is aware of the route change, it needs to set up 1356 the reservation on the new path. This is done by incrementing the 1357 RSN and sending a RESERVE message. On links that are common to the 1358 old and the new path, this RESERVE message is interpreted as a 1359 refreshing RESERVE. On new links, it creates the reservation. 1361 After the reservation on the new path is set up, the branching node 1362 or the merging node may want to tear down the reservation on the old 1363 path (faster than what would result from normal soft-state time-out). 1364 This functionality is supported by keeping track of the old SII. 1365 This specification requests GIMPS design to provide support for an 1366 SII. The SII is opaque to the QoS NSLP, i.e. QoS NSLP does not make 1367 any assumptions on how this identifier is constructed. When passed 1368 over the API, it allows QoS NSLP to indicate that its messages should 1369 be sent to the QNE identified by that SII. 1371 In case of a receiver-initiated reservation, a QNE can detect a route 1372 change by receiving a RESERVE message with a different SII. In case 1373 of a sender-initiated reservation, the same information is learned 1374 from a RESPONSE message, or from a NOTIFY message sent by the 1375 downstream peer. A QNE that has detected the route change via the 1376 SII change sends a RESERVE message towards the QNR on the old path 1377 (using the old SII) with the TEAR flag set. Note that in case of 1378 receiver-initiated reservations, this involves A QNE that is notified 1379 of the route change in another way and wants to tear down the old 1380 branch needs to send the RESERVE on the new path with an RII object. 1381 When it receives the RESPONSE message back, it can check whether its 1382 peer has effectively changed and send a RESERVE with the TEAR flag 1383 set if it has. Otherwise, teardown is not needed. A QNE that is 1384 unable to support an RII or does not receive a RESPONSE needs to rely 1385 on soft-state timeout on the old branch. 1387 A QNI or a branch node may wish to keep the reservation on the old 1388 branch. This could for instance be the case when a mobile node has 1389 experienced a mobility event and wishes to keep reservation to its 1390 old attachment point in case it moves back there. In that case, it 1391 sets the REPLACE flag in the common header to zero. 1393 4.9 Authorization Model Examples 1395 Various authorization models can be used in conjunction with the QoS 1396 NSLP. 1398 4.9.1 Authorization for the two party approach 1400 The two party approach is conceptually the simplest authorization 1401 model. 1403 +-------------+ QoS request +--------------+ 1404 | Entity |----------------->| Entity | 1405 | requesting | | authorizing | 1406 | resource |granted / rejected| resource | 1407 | |<-----------------| request | 1408 +-------------+ +--------------+ 1409 ^ ^ 1410 +...........................+ 1411 financial establishment 1413 Figure 13: Two party approach 1415 In this example the authorization decision only involves the two 1416 entities, or makes use of previous authorisation using an out-of-band 1417 mechanism to avoid the need for active participation of an external 1418 entity during the NSIS protocol execution. 1420 This type of model may be applicable, for example, between two 1421 neighbouring networks (inter-domain signalling) where a long-term 1422 contract (or other out-of-band mechanisms) exists to manage charging 1423 and provides sufficient information to authorize individual requests. 1425 4.9.2 Token based three party approach 1427 An alternative approach makes use of authorization tokens, such as 1428 those described in RFC 3520 [RFC3520] and RFC 3521 [RFC3521] or used 1429 as part of the Open Settlement Protocol [OSP]. The former 1430 ('authorization tokens') are used to associate two different 1431 signalling protocols (i.e. SIP and NSIS) and their authorization 1432 with each other whereas the latter is a form of digital money. As an 1433 example, with the authorization token mechanism, some form of 1434 authorization is provided by the SIP proxy, which acts as the 1435 resource authorizing entity in Figure 14. If the request is 1436 authorized, then the SIP signalling returns an authorization token 1437 which can be included in the QoS signalling protocol messages to 1438 refer to the previous authorization decision. The tokens themselves 1439 may take a number of different forms, some of which may require the 1440 entity performing the QoS reservation to query external state. 1442 Authorization 1443 Token Request +--------------+ 1444 +-------------->| Entity C | financial settlement 1445 | | authorizing | <..................+ 1446 | | resource | . 1447 | +------+ request | . 1448 | | +--------------+ . 1449 | | . 1450 | |Authorization . 1451 | |Token . 1452 | | . 1453 | | . 1454 | | . 1455 | | QoS request . 1456 +-------------+ + Authz. Token +--------------+ . 1457 | Entity |----------------->| Entity B | . 1458 | requesting | | performing | . 1459 | resource |granted / rejected| QoS | <..+ 1460 | A |<-----------------| reservation | 1461 +-------------+ +--------------+ 1463 Figure 14: Token based three party approach 1465 For the digital money type of systems (e.g. OSP tokens), the token 1466 represents a limited amount of credit. So, new tokens must be sent 1467 with later refresh messages once the credit is exhausted. 1469 4.9.3 Generic three party approach 1471 Another method is for the node performing the QoS reservation to 1472 delegate the authorization decision to a third party, as illustrated 1473 in Figure 15. 1475 +--------------+ 1476 | Entity C | 1477 | authorizing | 1478 | resource | 1479 | request | 1480 +-----------+--+ 1481 ^ | 1482 | | 1483 QoS | | QoS 1484 authz| |authz 1485 req.| | res. 1486 | | 1487 QoS | v 1488 +-------------+ request +--+-----------+ 1489 | Entity |----------------->| Entity B | 1490 | requesting | | performing | 1491 | resource |granted / rejected| QoS | 1492 | A |<-----------------| reservation | 1493 +-------------+ +--------------+ 1495 Figure 15: Three party approach 1497 Authorization may be performed on a per-request basis, periodically, 1498 or on a per-session basis. The authorization request might make use 1499 of EAP authentication between entities A and C, and a subsequent 1500 protocol exchange between A and B to create a secure channel for 1501 further communications. Such a technique gives flexibility in terms 1502 of the authentication and key exchange protocols used. 1504 A further extension to this model is to allow Entity C to reference a 1505 AAA server in the user's home network when making the authorization 1506 decision. 1508 5. QoS NSLP Functional specification 1510 5.1 QoS NSLP Message and Object Formats 1512 A QoS NSLP message consists of a common header, followed by a body 1513 consisting of a variable number of variable-length, typed "objects". 1514 The common header and other objects are encapsulated together in a 1515 GIMPS NSLP-Data object. The following subsections define the formats 1516 of the common header and each of the QoS NSLP message types. In the 1517 message formats, the common header is denoted as COMMON_HEADER. 1519 For each QoS NSLP message type, there is a set of rules for the 1520 permissible choice of object types. These rules are specified using 1521 the Augmented Backus-Naur Form (ABNF) specified in RFC 2234 1522 [RFC2234]. The ABNF implies an order for the objects in a message. 1523 However, in many (but not all) cases, object order makes no logical 1524 difference. An implementation should create messages with the 1525 objects in the order shown here, but accept the objects in any order, 1526 except for QSPEC object(s) which always appear last in the message. 1528 5.1.1 Common header 1530 All GIMPS NSLP-Data objects for the QoS NSLP MUST contain this common 1531 header as the first 32 bits of the object (this is not the same as 1532 the GIMPS Common Header). 1534 0 1 2 3 1535 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1536 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1537 | Message Type | Flags | 1538 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1540 The fields in the common header are as follows: 1542 Msg Type: 16 bits 1543 1 = RESERVE 1544 2 = QUERY 1545 3 = RESPONSE 1546 4 = NOTIFY 1548 Flags: 16 bits 1549 The set of appropriate flags depends on the particular message 1550 being processed. Any bit not defined as a flag for a 1551 particular message MUST be set to zero on sending and MUST be 1552 ignored on receiving. 1554 5.1.2 Message formats 1556 5.1.2.1 RESERVE 1558 The format of a RESERVE message is as follows: 1560 RESERVE = COMMON_HEADER 1561 RSN [ RII ] [ REFRESH_PERIOD ] [ BOUND_SESSION_ID ] 1562 [ POLICY_DATA ] [ *QSPEC ] 1564 The RSN is the only mandatory object and MUST always be present. 1566 If any QSPEC objects are present, they MUST occur at the end of the 1567 message. There are no other requirements on transmission order, 1568 although the above order is recommended. 1570 Four flags are defined for use in the common header with the RESERVE 1571 message. These are: 1573 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1574 | Reserved |T|S|A|R| 1575 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1577 TEAR (T) - when set, indicates that reservation state and QoS NSLP 1578 operation state should be torn down. This is indicated to the 1579 RMF. 1581 SCOPING (S) - when set, indicates that the message is scoped and 1582 should not travel down the entire path but only as far as the next 1583 QNE (scope="next hop"). By default, this flag is not set (default 1584 scope="whole path"). 1586 ACKNOWLEDGE (A) - when set, indicates that an explicit 1587 confirmation of the state installation action is REQUIRED. This 1588 flag SHOULD be set on transmission by default. 1590 REPLACE (R) - when set, indicates that a RESERVE with different 1591 Message Routing Information (MRI) replaces an existing one, so the 1592 old one MAY be torn down immediately. This is the default 1593 situation. This flag may be unset to indicate a desire from an 1594 upstream node to keep an existing reservation on an old branch in 1595 place. 1597 If the REFRESH_PERIOD is not present, a default value of 30 seconds 1598 is assumed. 1600 If the session of this message is bound to another session, then the 1601 RESERVE message MUST include the SESSION_ID of that other session in 1602 a BOUND_SESSION_ID object. 1604 A "reservation collision" may occur if the sender believes that a 1605 sender-initiated reservation should be performed for a flow, whilst 1606 the other end believes that it should be starting a 1607 receiver-initiated reservation. If different session identifiers are 1608 used then this error condition is transparent to the QoS NSLP though 1609 it may result in an error from the RMF, otherwise the removal of the 1610 duplicate reservation is left to the QNIs/QNRs for the two sessions. 1612 If a reservation is already installed and a RESERVE message is 1613 received with the same session identifier from the other direction 1614 (i.e. going upstream where the reservation was installed by a 1615 downstream RESERVE messge, or vice versa) then an error indicating 1616 "RESERVE received from wrong direction" MUST be sent in a RESPONSE 1617 message to the signalling message source for this second RESERVE. 1619 5.1.2.2 QUERY 1621 The format of a QUERY message is as follows: 1623 QUERY = COMMON_HEADER 1624 [ RII ][ BOUND_SESSION_ID ] 1625 [ POLICY_DATA ] [ *QSPEC ] 1627 A QUERY message MUST contain an RII object to match an incoming 1628 RESPONSE to the QUERY, unless the QUERY is being used to initiate 1629 reverse-path state for a receiver-initiated reservation. 1631 A QUERY message MAY contain one or more QSPEC objects and a 1632 POLICY_DATA object. The QSPEC object describes what is being queried 1633 for and may contain objects that gather information along the data 1634 path. The POLICY_DATA object authorizes the requestor of the QUERY 1635 message. If any QSPEC objects are present, they MUST occur at the 1636 end of the message. There are no other requirements on transmission 1637 order, although the above order is recommended. 1639 One flag is defined for use in the common header with the QUERY 1640 message. This is: 1642 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1643 | Reserved |S| 1644 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1646 SCOPING - when set, indicates that the message is scoped an should 1647 not travel down the entire path but only as far as the next QNE 1648 (scope="next hop"). By default, this flag is not set (default 1649 scope="whole path"). 1651 If the session of this message is bound to another session, then the 1652 RESERVE message MUST include the SESSION_ID of that other session in 1653 a BOUND_SESSION_ID object. 1655 5.1.2.3 RESPONSE 1657 The format of a RESPONSE message is as follows: 1659 RESPONSE = COMMON_HEADER 1660 [ RII / RSN ] ERROR_SPEC 1661 [ *QSPEC ] 1663 A RESPONSE message MUST contain an ERROR_SPEC object which indicates 1664 the success of a reservation installation or an error condition. 1665 Depending on the value of the ERROR_SPEC, the RESPONSE MAY also 1666 contain a QSPEC object. 1668 If any QSPEC objects are present, they MUST occur at the end of the 1669 message. There are no other requirements on transmission order, 1670 although the above order is recommended. 1672 One flag is defined for use in the common header with the RESPONSE 1673 message. This is: 1675 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1676 | Reserved |S| 1677 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1679 SCOPING - when set, indicates that the message is scoped and 1680 should not travel down the entire path but only as far as the next 1681 QNE (scope="next hop"). By default, this flag is not set (default 1682 scope="whole path"). 1684 5.1.2.4 NOTIFY 1686 The format of a NOTIFY message is as follows: 1688 NOTIFY = COMMON_HEADER 1689 ERROR_SPEC [ QSPEC ] 1691 A NOTIFY message MUST contain an ERROR_SPEC object indicating the 1692 reason for the notification. Depending on the ERROR_SPEC value, it 1693 MAY contain a QSPEC providing additional information. 1695 No flags are defined for use with the NOTIFY message. 1697 5.1.3 Object Formats 1699 The QoS NSLP uses the Type-Length-Value (TLV) object format defined 1700 by GIMPS [I-D.ietf-nsis-ntlp]. Every object consists of one or more 1701 32-bit words with a one-word header. For convenience the standard 1702 object header is shown here: 1704 0 1 2 3 1705 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1706 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1707 |r|r|r|r| Type |r|r|r|r| Length | 1708 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1710 The value for the Type field comes from GIMPS object type space. The 1711 Length field is given in units of 32 bit words and and measures the 1712 length of the Value component of the TLV object (i.e. it does not 1713 include the standard header). 1715 The object diagrams here use '//' to indicate a variable sized field 1716 and ':' to indicate a field that is optionally present. 1718 A QoS NSLP implementation must recognize objects of the following 1719 types: RII, RSN, REFRESH_PERIOD, BOUND_SESSION_ID, ERROR_SPEC, QSPEC 1720 and POLICY_DATA. 1722 NB: This draft does not currently include the codepoints for the QoS 1723 NSLP related object types. 1725 The object header is followed by the Value field, which varies for 1726 different objects. The format of the Value field for currently 1727 defined objects is specified below. 1729 5.1.3.1 Request Identification Information (RII) 1731 Type: RII 1732 Length: Fixed - 1 32-bit word 1733 Value: An identifier which must be (probabilistically) unique within 1734 the context of a SESSION_ID, and SHOULD be different every time a 1735 RESPONSE is desired. Used by a QNE to match back a RESPONSE to a 1736 request in a RESERVE or QUERY message. 1738 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1739 | Response Identification Information (RII) | 1740 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1742 5.1.3.2 Reservation Sequence Number (RSN) 1744 Type: RSN 1745 Length: Fixed - 1 32-bit word 1746 Value: An incrementing sequence number that indicates the order in 1747 which state modifying actions are performed by a QNE. The RSN has 1748 local significance only, i.e. between a pair of neighbouring 1749 stateful QNEs. 1751 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1752 | Reservation Sequence Number (RSN) | 1753 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1755 5.1.3.3 REFRESH_PERIOD 1757 Type: REFRESH_PERIOD 1758 Length: Fixed - 1 32-bit word 1759 Value: The refresh timeout period R used to generate this message; in 1760 milliseconds. 1762 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1763 | Refresh Period (R) | 1764 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1766 5.1.3.4 BOUND_SESSION_ID 1768 Type: BOUND_SESSION_ID 1769 Length: Fixed - 4 32-bit words 1770 Value: Specifies the SESSION_ID (as specified in GIMPS 1771 [I-D.ietf-nsis-ntlp]) of the session that must be bound to the 1772 session associated with the message carrying this object. 1774 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1775 | | 1776 + + 1777 | | 1778 + Session ID + 1779 | | 1780 + + 1781 | | 1782 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1784 5.1.3.5 ERROR_SPEC 1786 The error object shares a common format with GIMPS and is specified 1787 in the GIMPS [I-D.ietf-nsis-ntlp] specification. 1789 Type: ERROR 1790 Length: Variable 1791 Value: Contains a 1 byte error class and 3 byte error code, an error 1792 source identifier and optionally variable length error-specific 1793 information. 1795 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1796 | Error Class | Error Code | 1797 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1798 | ESI-Length | Reserved | 1799 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1800 // Error Source Identifier // 1801 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1802 // Optional error-specific information // 1803 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1805 The first byte of the error code indicates the severity level. The 1806 currently defined severity levels are: 1807 o 0x01 - Informational 1808 o 0x02 - Success 1809 o 0x03 - Protocol Error 1810 o 0x04 - Transient Failure 1811 o 0x05 - Permanent Failure 1813 Within each severity class a number of error values are defined. 1814 o Informational: 1815 * 0x01000001 - Unknown BOUND_SESSION_ID: the message refers to an 1816 unknown SESSION_ID in its BOUND_SESSION_ID object. 1817 o Success: 1818 * 0x02000001 - State installation succeeded 1819 * 0x02000002 - Reservation created: reservation installed on 1820 complete path (sent by last node). 1822 * 0x02000003 - Reservation accepted: reservation installed at 1823 this QNE, but not yet installed on the rest of the path. 1824 * 0x02000004 - Reservation created but modified: reservation 1825 installed, but bandwidth reserved was not the maximum 1826 requested. 1827 o Protocol Error: 1828 * 0x03000001 - Illegal message type: the type given in the 1829 Message Type field of the common header is unknown. 1830 * 0x03000002 - Wrong message length: the length given for the 1831 message does not match the length of the message data. 1832 * 0x03000003 - Bad flags value: an undefined flag or combination 1833 of flags was set. 1834 * 0x03000004 - Mandatory object missing: an object required in a 1835 message of this type was missing. 1836 * 0x03000005 - Illegal object present: an object was present 1837 which must not be used in a message of this type. 1838 * 0x03000006 - Unknown object present: an object of an unknown 1839 type was present in the message. 1840 * 0x03000007 - Wrong object length: the length given for the 1841 object did not match the length of the object data present. 1842 * 0x03000008 - Unknown QSPEC type: the QoS Model ID refers to a 1843 QoS Model which is not known by this QNE. 1844 * 0x3000009 - RESERVE received from wrong direction. 1845 o Transient Failure: 1846 * 0x04000001 - Requested resources not available 1847 * 0x04000002 - Insufficient bandwidth available 1848 * 0x04000003 - Delay requirement cannot be met 1849 * 0x04000004 - Transient RMF-related error 1850 * 0x04000005 - Resources pre-empted 1851 * 0x04000006 - No GIMPS reverse-path forwarding state 1852 * 0x04000007 - NSLP soft-state expired 1853 o Permanent Failure: 1854 * 0x05000001 - Authentication failure 1855 * 0x05000002 - Unable to agree transport security with peer 1856 * 0x05000003 - Internal or system error 1857 * 0x05000004 - Resource request denied (authorization failed) 1858 * 0x05000005 - Permanent RMF-related error 1860 5.1.3.6 QSPEC 1862 Type: QSPEC 1863 Length: Variable 1864 Value: This object contains a 4 byte QoS Model ID and a variable 1865 length QSPEC (QoS specification) information, which is QoS Model 1866 dependent. Such a QoS Model can be a standardized one, a private 1867 one, or a well-known one. 1869 The contents and encoding rules for this object are specified in 1870 other documents, prepared by QSPEC template designers. 1872 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1873 | QoS Model Identifier (QoS Model ID) | 1874 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1875 | | 1876 // QSpec Data // 1877 | | 1878 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1880 5.1.3.7 POLICY_DATA 1882 POLICY_DATA objects may contain various items to authenticate the 1883 user and allow the reservation to be authorised. Some related issues 1884 are also discussed in Section 3.1.4. 1886 5.2 General Processing Rules 1888 5.2.1 State Manipulation 1890 The processing of a message and its component objects involves 1891 manipulating the QoS NSLP and reservation state of a QNE. 1893 For each flow, a QNE stores (RMF-related) reservation state which 1894 depends on the QoS model / QSpec used and QoS NSLP operation state 1895 which includes non-persistent state (e.g. the API parameters while a 1896 QNE is processing a message) and persistent state which is kept as 1897 long as the session is active. 1899 The persistent QoS NSLP state is conceptually organised in a table 1900 with the following structure. The primary key (index) for the table 1901 is the SESSION_ID: 1902 SESSION_ID 1904 A large identifier provided by GIMPS or set locally. 1906 The state information for a given key includes: 1907 Flow ID 1909 Copied from GIMPS. Several entries are possible in case of 1910 mobility events. 1912 QoS Model ID 1914 32 bit identification of the QoS Model. 1916 SII for each upstream and downstream peer 1918 The SII is a large identifier (minimum 128 bits) generated by the 1919 QoS NSLP and passed over the API. 1921 RSN from each upstream peer 1923 The RSN is a 32 bit counter. 1925 Current own RSN 1927 A 32 bit random number. 1929 List of RII for outstanding responses with processing information 1931 the RII is a 32 bit number. 1933 State lifetime 1935 The state lifetime indicates how long the state that is being 1936 signalled for remains valid. 1938 BOUND_SESSION_ID 1940 The BOUND_SESSION_ID is a 128 bit random number. 1942 Adding the state requirements of all these items gives an upper bound 1943 on the state to be kept by a QNE. The need to keep state depends on 1944 the desired functionality at the NSLP layer. 1946 5.2.2 Message Forwarding 1948 QoS NSLP messages are sent peer-to-peer along the path. The QoS NSLP 1949 does not have the concept of a message being sent along the entire 1950 path. Instead, messages are received by a QNE, which may then send 1951 another message (which may be identical to the received message, or 1952 contain some subset of objects from it) to continue in the same 1953 direction (i.e. towards QNI or QNR) as the message received. 1955 The decision on whether to generate a message to forward may be 1956 affected by the value of the SCOPING flag or by the presence of an 1957 RII object. 1959 5.2.3 Standard Message Processing Rules 1961 If a mandatory object is missing from a message then the receiving 1962 QNE MUST NOT propagate the message any further. It MUST construct an 1963 RESPONSE message indicating the error condition and send it back to 1964 the peer QNE that sent the message. 1966 If a message contains an object of an unrecognised type, then the 1967 behaviour depends on the object type value. 1969 5.3 Object Processing 1971 5.3.1 Reservation Sequence Number (RSN) 1973 A QNE's own RSN is a sequence number which applies to a particular 1974 NSIS signalling session (i.e. with a particular GIMPS SESSION_ID). 1975 It MUST be incremented for each new RESERVE message where the 1976 reservation for the session changes. Once the RSN has reached its 1977 maximum value, the next value it takes is zero. 1979 When receiving a RESERVE message a QNE uses the RSN given in the 1980 message to determine whether the state being requested is different 1981 to that already stored. If the RSN is the same as for the current 1982 reservation the current state MUST be refreshed. If the RSN is 1983 greater than the current stored value, the current reservation MUST 1984 be modified appropriately (provided that admission control and policy 1985 control succeed), and the stored RSN value updated to that for the 1986 new reservation. If the RSN is less than the current value, then it 1987 indicates an out-of-order message and the RESERVE message MUST be 1988 discarded. 1990 If the QNE does not store per-session state (and so does not keep any 1991 previous RSN values) then it MAY ignore the value of the RSN. It 1992 MUST also copy the same RSN into the RESERVE message (if any) it 1993 sends as a consequence of receiving this one. 1995 5.3.2 Request Identification Information (RII) 1997 A QNE sending some types of messages may require a response to be 1998 sent. It does so by including a Request Identification Information 1999 (RII) object. 2001 When creating an RII object the sender MUST select the value for the 2002 RII such that it is probabilistically unique within the given 2003 session. 2005 A number of choices are available when implementing this. 2006 Possibilities might include using a totally random value, or a node 2007 identifier together with a counter. If the value is selected by 2008 another QNE then RESPONSE messages may be incorrectly terminated, and 2009 not passed back to the node that requested them. 2011 When sending a message containing an RII object the sending node MUST 2012 remember the value used in the RII to match back any RESPONSE 2013 received. It SHOULD use a timer to identify situations where it has 2014 taken too long to receive the expected RESPONSE. If the timer 2015 expires without receiving a RESPONSE it MAY perform a retransmission. 2017 When receiving a message containing an RII object the node MUST send 2018 a RESPONSE if either 2019 o The SCOPING flag is set to one ('next hop' scope), or 2020 o This QNE is the last one on the path for the given session. 2021 and the QNE keeps per-session state for the given session. 2023 A message contains at most one RII object that is unique within a 2024 session and different for each message, in order to allow responses 2025 to be matched back to requests (without incorrectly matching at other 2026 nodes). Downstream nodes that desire responses may keep track of 2027 this RII to identify the RESPONSE when it passes back through them. 2029 5.3.3 BOUND_SESSION_ID 2031 As shown in the examples in Section 4, the QoS NSLP can relate 2032 multiple sessions together. It does this by including the SESSION_ID 2033 from one session in a BOUND_SESSION_ID object in messages in another 2034 session. 2036 When receiving a message with a BOUND_SESSION_ID object, a QNE MUST 2037 copy the BOUND_SESSION_ID object into all messages it sends for the 2038 same session. A QNE that stores per-session state SHOULD store the 2039 value of the BOUND_SESSION_ID. 2041 The BOUND_SESSION_ID is only indicative in nature. However, a QNE 2042 implementation MAY use BOUND_SESSION_ID information to optimize 2043 resource allocation, e.g. for bidirectional reservations. When 2044 receiving a tearing RESERVE for an aggregate reservation, it MAY use 2045 this information to initiate a tearing RESERVE for end-to-end 2046 sessions bound to the aggregate. 2048 5.3.4 REFRESH_PERIOD 2050 Refresh timer management values are carried by the REFRESH_PERIOD 2051 object which has local significance only. At the expiration of a 2052 "refresh timeout" period, each QNE independently examines its state 2053 and sends a refreshing RESERVE message to the next QNE peer where it 2054 is absorbed. This peer-to-peer refreshing (as opposed to the QNI 2055 initiating a refresh which travels all the way to the QNR) allows 2056 QNEs to choose refresh intervals as appropriate for their 2057 environment. For example, it is conceivable that refreshing 2058 intervals in the backbone, where reservations are relatively stable, 2059 are much larger than in an access network. The "refresh timeout" is 2060 calculated within the QNE and is not part of the protocol; however, 2061 it must be chosen to be compatible with the reservation lifetime as 2062 expressed by the REFRESH_PERIOD, and an assessment of the reliability 2063 of message delivery. 2065 The details of timer management and timer changes (slew handling and 2066 so on) are identical to the ones specified in Section 3.7 of RFC 2205 2067 [RFC2205]. 2069 There are two time parameters relevant to each QoS NSLP state in a 2070 node: the refresh period R between generation of successive refreshes 2071 for the state by the neighbor node, and the local state's lifetime L. 2072 Each RESERVE message may contain a REFRESH_PERIOD object specifying 2073 the R value that was used to generate this (refresh) message. This R 2074 value is then used to determine the value for L when the state is 2075 received and stored. The values for R and L may vary from peer to 2076 peer. This peer-to-peer refreshing (as opposed to the QNI initiating 2077 a refresh which travels all the way to the QNR) allows QNEs to choose 2078 refresh intervals as appropriate for their environment. For example, 2079 it is conceivable that refreshing intervals in the backbone, where 2080 reservations are relatively stable, are much larger than in an access 2081 network. 2083 In more detail (quoting directly from RFC2205): 2084 1. Floyd and Jacobson [_XREF_FJ94] have shown that periodic 2085 messages generated by independent network nodes can become 2086 synchronized. This can lead to disruption in network services as 2087 the periodic messages contend with other network traffic for link 2088 and forwarding resources. Since QoS NSLP sends periodic refresh 2089 messages, it must avoid message synchronization and ensure that 2090 any synchronization that may occur is not stable. For this 2091 reason, it is recommended that the the refresh timer should be 2092 randomly set to a value in the range [0.5R, 1.5R]. 2093 2. To avoid premature loss of state, L must satisfy L >= (K + 2094 0.5)*1.5*R, where K is a small integer. Then in the worst case, 2095 K-1 successive messages may be lost without state being deleted. 2096 To compute a lifetime L for a collection of state with different R 2097 values R0, R1, ..., replace R by max(Ri). 2098 Currently K = 3 is suggested as the default. However, it may be 2099 necessary to set a larger K value for hops with high loss rate. K 2100 may be set either by manual configuration per interface, or by 2101 some adaptive technique that has not yet been specified. 2103 3. Each RESERVE message carries a REFRESH_PERIOD object 2104 containing the refresh time R used to generate refreshes. The 2105 recipient node uses this R to determine the lifetime L of the 2106 stored state created or refreshed by the message. 2107 4. The refresh time R is chosen locally by each node. If the 2108 node does not implement local repair of reservations disrupted by 2109 route changes, a smaller R speeds up adaptation to routing 2110 changes, while increasing the QOS-NSLP overhead. With local 2111 repair, a router can be more relaxed about R since the periodic 2112 refresh becomes only a backstop robustness mechanism. A node may 2113 therefore adjust the effective R dynamically to control the amount 2114 of overhead due to refresh messages. 2115 The current suggested default for R is 30 seconds. However, the 2116 default value Rdef should be configurable per interface. 2117 5. When R is changed dynamically, there is a limit on how fast it 2118 may increase. Specifically, the ratio of two successive values 2119 R2/R1 must not exceed 1 + Slew.Max. 2120 Currently, Slew.Max is 0.30. With K = 3, one packet may be lost 2121 without state timeout while R is increasing 30 percent per refresh 2122 cycle. 2123 6. To improve robustness, a node may temporarily send refreshes 2124 more often than R after a state change (including initial state 2125 establishment). 2126 7. The values of Rdef, K, and Slew.Max used in an implementation 2127 should be easily modifiable per interface, as experience may lead 2128 to different values. The possibility of dynamically adapting K 2129 and/or Slew.Max in response to measured loss rates is for future 2130 study. 2132 5.3.5 ERROR_SPEC 2134 ERROR_SPEC processing rules are still to be defined in more detail. 2136 5.3.6 QSPEC 2138 The contents of the QSPEC depends on the QoS model being used. It 2139 may be that parts of the QSPEC are standardised across multiple QoS 2140 models. This topic is currently under further study. 2142 Upon reception, the complete QSPEC is passed to the Resource 2143 Management Function (RMF). 2145 A QNE that receives a QSPEC stack MUST only look at the top QSPEC in 2146 the stack. If this QSPEC is not understood by the RMF, the QNE MUST 2147 send an RESPONSE containing an ERROR_SPEC and MUST NOT attempt to 2148 recover by inspecting the rest of the stack. 2150 Parameters of the QoS Model that is being signalled for are carried 2151 in the QSPEC object. A domain may have local policies regarding QoS 2152 model implementation, i.e. it may map incoming traffic to its own 2153 locally defined Qos Models. The QoS NSLP supports this by allowing 2154 QSPEC objects to be stacked. 2156 When a domain wants to apply a certain QoS Model to an incoming 2157 per-flow reservation request, each edge of the domain is configured 2158 to map the incoming QSPEC object to a local QSPEC object and push 2159 that object onto the stack of QSPEC objects (typically immediately 2160 following the Common Control Information, i.e. the first QSPEC that 2161 is found in the message). 2163 A QNE that knows it is the last QNE to understand a local QSPEC 2164 object (e.g. by configuration of the egress QNEs of a domain) SHOULD 2165 remove the topmost QSPEC object from the stack. It SHOULD update the 2166 underlying QoS Model parameters if needed. 2168 A QNE that receives a message with a QSPEC object stack of which the 2169 topmost object is not understood MUST NOT forward the message and 2170 MUST send an error indication to its upstream neighbour. It MUST NOT 2171 attempt local recovery by inspecting the stack for a QSPEC object it 2172 understands. 2174 If the RMF indicates it cannot process the QSPEC, e.g. because the 2175 QoS Model is not supported the QNE sends a RESPONSE with the 2176 appropriate ERROR_SPEC. 2178 5.4 Message Processing Rules 2180 5.4.1 RESERVE Messages 2182 The RESERVE message is used to manipulate QoS reservation state in 2183 QNEs. A RESERVE message may create, refresh, modify or remove such 2184 state. The format of a RESERVE message is repeated here for 2185 convenience: 2187 RESERVE = COMMON_HEADER 2188 RSN [ RII ] [ REFRESH_PERIOD ] [ BOUND_SESSION_ID ] 2189 [ POLICY_DATA ] [ *QSPEC ] 2191 RESERVE messages MUST only be sent towards the QNR. 2193 A QNE that receives a RESERVE message checks the message format. In 2194 case of malformed messages, the QNE sends a RESPONSE message with the 2195 appropriate ERROR_SPEC. 2197 Before performing any state changing actions a QNE MUST determine 2198 whether the request is authorized. It SHOULD exercise its local 2199 policy in conjunction with the POLICY_DATA object to do this. 2201 When the RESERVE is authorized, a QNE checks the COMMON_HEADER flags. 2202 If the TEAR flag is set, the message is a tearing RESERVE which 2203 indicates complete QoS NSLP state removal (as opposed to a 2204 reservation of zero resources). On receiving such a RESERVE message 2205 the QNE MUST inform the RMF that the reservation is no longer 2206 required. The QNE SHOULD remove the QoS NSLP state. It MAY signal 2207 to GIMPS (over the API) that reverse path state for this reservation 2208 is no longer required. If the QNE has reservations which are bound 2209 to this session (they contained the SESSION_ID of this session in 2210 their BOUND_SESSION_ID object), it MUST send a NOTIFY message for 2211 each of these reservations with an appropriate ERROR_SPEC. The QNE 2212 MAY elect to send RESERVE messages with the TEAR flag set for these 2213 reservations. 2215 The default behaviour of a QNE that receives a RESERVE with a 2216 SESSION_ID for which it already has state installed but with a 2217 different flow ID is to replace the existing reservation (and tear 2218 down the reservation on the old branch if the RESERVE is received 2219 with a different SII). 2221 In some cases, this may not be the desired behaviour. In that case, 2222 the QNI or a QNE may set the REPLACE flag in the common header to 2223 zero to indicate that the new session does not replace the existing 2224 one. A QNE that receives a RESERVE with the REPLACE flag set to zero 2225 but with the same SII will update the flow ID and indicate REPLACE=0 2226 to the RMF (where it will be used for the resource handling). If the 2227 SII is different, this means that the QNE is a merge point. In that 2228 case, the REPLACE=0 also indicates that a tearing RESERVE SHOULD NOT 2229 be sent on the old branch. 2231 When a QNE receives a (refreshing) RESERVE message with an unknown 2232 SESSION_ID, it MAY send a NOTIFY message to its upstream peer, 2233 indicating the unknown SESSION_ID. This indicates a downstream route 2234 change to the upstream peer. The upstream peer SHOULD send a 2235 complete RESERVE on the new path (new SII). It identifies the old 2236 signalling association (old SII) and MAY start sending complete 2237 RESERVE messages for other SESSION_IDs linked to this association. 2239 At a QNE, resource handling is performed by the RMF. For sessions 2240 with the REPLACE flag set to zero, we assume that the QoS model 2241 includes directions to deal with resource sharing. This may include, 2242 adding the reservations, or taking the maximum of the two or more 2243 complex mathematical operations. 2245 This resource handling mechanism in the QoS Model is also applicable 2246 to sessions with different SESSION_ID but related through the 2247 BOUND_SESSION_ID object. Session replacement is not an issue here, 2248 but the QoS Model may specify whether to let the sessions that are 2249 bound together share resources on common links or not. 2251 Finally, it is possible that a RESERVE is received with no QSPEC at 2252 all. This is the case of a summary refresh. In this case, rather 2253 than sending a refreshing RESERVE with the full QSPEC, only the 2254 SESSION_ID and the SII are sent to refresh the reservation. Note 2255 that this mechanism just reduces the message size (and probably eases 2256 processing). One RESERVE per session is still needed. 2258 If the REPLACE flag is set, the QNE SHOULD update the reservation 2259 state according to the QSPEC contained in the message. It MUST 2260 update the lifetime of the reservation. If the REPLACE flag is not 2261 set, a QNE SHOULD NOT remove the old reservation state if the SII 2262 which is passed by GIMPS over the API is different than the SII that 2263 was stored for this reservation. The QNE MAY elect to keep sending 2264 refreshing RESERVE messages. 2266 If the ACKNOWLEDGE flag is set, the QNE MUST acknowledge its state 2267 installation action. It does so by sending a RESPONSE with an 2268 ERROR_SPEC value of 0x02000003, indicating that the reservation is 2269 installed at the QNE. 2271 If the SCOPING flag is set, or if the QNE is the last QNE on the path 2272 to the destination, the QNE MUST send a RESPONSE message. 2274 When a QNE receives a RESERVE message, its processing may involve 2275 sending out another RESERVE message. When sending a RESERVE message, 2276 the QNE may insert or remove 'local' QSPEC objects from the top of 2277 the stack. If there are one or more QSPECs in the received RESERVE 2278 message, the last QSPEC MUST NOT be removed when sending on the 2279 RESERVE message. 2281 Upon transmission, a QNE SHOULD set the ACKNOWLEDGE flag. It MUST do 2282 so if it wishes to use the reduced overhead refresh mechanism 2283 described in Section 3.2.3. It MUST NOT send a reduced overhead 2284 refresh message (i.e. a RESERVE with a non-incremented RSN and no 2285 QSPEC) unless it has received a RESPONSE message for that RESERVE 2286 message. 2288 If the session of this message is bound to another session, then the 2289 RESERVE message MUST include the SESSION_ID of that other session in 2290 a BOUND_SESSION_ID object. 2292 In case of receiver-initiated reservations, the RESERVE message must 2293 follow the same path that has been followed by the QUERY message. 2294 Therefore, GIMPS is informed, over the QoS NSLP/GIMPS API, to pass 2295 the message upstream, i.e., by setting GIMPS "D" flag, see GIMPS 2296 [I-D.ietf-nsis-ntlp]. 2298 5.4.2 QUERY Messages 2300 A QUERY message is used to request information about the data path 2301 without making a reservation. This functionality can be used to 2302 'probe' the network for path characteristics or for support of 2303 certain QoS models. 2305 The format of a QUERY message is as follows: 2307 QUERY = COMMON_HEADER 2308 [ RII ] [ BOUND_SESSION_ID ] 2309 [ POLICY_DATA ] [ *QSPEC ] 2311 When a QNE receives a QUERY message the QSpec is passed to the RMF 2312 for processing. The RMF may return a modified QSpec which is used in 2313 any QUERY or RESPONSE message sent out as a result of the QUERY 2314 processing. 2316 When processing a QUERY message, a QNE checks whether an RII object 2317 is present. If not, the QUERY is an empty QUERY which is used to 2318 install reverse path state. In this case, if the QNE is not the QNR, 2319 it creates a new QUERY message to send downstream. If the QUERY 2320 contained a QSPEC, this MUST be passed to the RMF where it MAY be 2321 modified by QoS Model specific QUERY processing. If the QNE is the 2322 QNR, the QNE creates a RESERVE message, which contains a QSPEC 2323 received from the RMF and which MAY be based on the received QSPEC. 2324 If this node was not expecting to perform a receiver-initiated 2325 reservation then an error MUST be sent back along the path. 2327 If an RII object is present, and if the QNE is the QNR or the SCOPING 2328 flag is set, the QNE MUST generate a RESPONSE message and pass it 2329 back along the reverse of the path used by the QUERY. 2331 In other cases, the QNE MUST generate a QUERY message which is then 2332 forwarded further along the path using the same MRI, Session ID and 2333 Direction as provided when the QUERY was received over the GIMPS API. 2334 The QSpec to be used is that provided by the RMF as described 2335 previously. When generating a QUERY to send out to pass the query 2336 further along the path, the QNE MUST copy the RII object (if present) 2337 into the new QUERY message unchanged. A QNE that is also interested 2338 in the response to the query keeps track of the RII to identify the 2339 RESPONSE when it passes through it. 2341 5.4.3 RESPONSE Messages 2343 The RESPONSE message is used to provide information about the result 2344 of a previous QoS NSLP message, e.g. confirmation of a reservation 2345 or information resulting from a query. The RESPONSE message is 2346 impotent, it does not cause any state to be installed or modified. 2348 The format of a RESPONSE message is repeated here for convenience: 2350 RESPONSE = COMMON_HEADER 2351 [ RII / RSN ] ERROR_SPEC 2352 [ *QSPEC ] 2354 A RESPONSE message MUST be sent where the QNE is the last node to 2355 process a RESERVE or QUERY message containing an RII object (based on 2356 scoping of the RESERVE or QUERY, or because this is the last node on 2357 the path). In this case, the RESPONSE MUST copy the RII object from 2358 the RESERVE or QUERY. 2360 In addition, a RESPONSE message MUST be sent when the ACKNOWLEDGE 2361 flag is set or when an error occurs while processing a received 2362 message. If the received message contains an RII object, this object 2363 MUST be put in the RESPONSE, as described above. If the RESPONSE is 2364 sent as a result of the receipt of a RESERVE message without an RII 2365 object, then the RSN of the received RESERVE message MUST be copied 2366 into the RESPONSE message. 2368 On receipt of a RESPONSE message containing an RII object, the QNE 2369 MUST attempt to match it to the outstanding response requests for 2370 that signalling session. If the match succeeds, then the RESPONSE 2371 MUST NOT be forwarded further along the path. If the match fails, 2372 then the QNE MUST attempt to forward the RESPONSE to the next peer 2373 QNE. 2375 On receipt of a RESPONSE message containing an RSN object, the QNE 2376 MUST compare the RSN to that of the appropriate signalling session. 2377 If the match succeeds then the ERROR_SPEC MUST be processed. The 2378 RESPONSE message MUST NOT be forwarded further along the path whether 2379 or not the match succeeds. 2381 5.4.4 NOTIFY Messages 2383 NOTIFY messages are used to convey information to a QNE 2384 asynchronously. The format of a NOTIFY message is as follows: 2386 NOTIFY = COMMON_HEADER 2387 ERROR_SPEC [ QSPEC ] 2389 NOTIFY messages are impotent. They do not cause any state to be 2390 installed or modified and they do do not directly cause other 2391 messages to be sent. NOTIFY messages are sent asynchronously, rather 2392 than in response to other messages. They may be sent in either 2393 direction (upstream or downstream). 2395 6. IANA considerations 2397 This section provides guidance to the Internet Assigned Numbers 2398 Authority (IANA) regarding registration of values related to the QoS 2399 NSLP, in accordance with BCP 26 RFC 2434 [RFC2434]. 2401 The QoS NSLP requires IANA to create two new registries. One for QoS 2402 NSLP Message Types, the other for QoS Signaling Policy Identifiers. 2404 The QoS NSLP Message Type is a 16 bit value. The allocation of 2405 values for new message types requires standards action. This 2406 specification defines four QoS NSLP message types, which form the 2407 initial contents of this registry: RESERVE, QUERY, RESPONSE and 2408 NOTIFY. 2410 QoS NSLP Messages have a messages-specific 16 bit flags/reserved 2411 field in their header. The allocation policy for new flags is TBD. 2413 The QoS Model Identifier (QoS Model ID) is a 32 bit value carried in 2414 a QSPEC object. The allocation policy for new QoS Model IDs is TBD. 2416 This specification defines a NSLP for use with GIMPS. Consequently, 2417 a new identifier must be assigned for it from GIMPS NSLP Identifier 2418 registry. 2420 This document also defines six new objects for the QoS NSLP: RII, 2421 RSN, REFRESH_PERIOD, BOUND_SESSION_ID, QSPEC and POLICY_DATA. Values 2422 are to be assigned for them from GIMPS Object Type registry. 2424 In addition it defines a number of Error Codes for the QoS NSLP. 2425 These can be found in section Section 5.1.3 and are to be assigned 2426 values from GIMPS Error Code registry. 2428 7. QoS use of GIMPS service interface 2430 This section describes the use of GIMPS service interface to 2431 implement QoS NSLP requirements on GIMPS. 2433 7.1 Example sender-initiated reservation 2435 We first describe the use of the service interface in a very basic 2436 scenario: message reception and transmission for a RESERVE message in 2437 a sender-initiated reservation. 2439 A QNE that wishes to initiate a sender-initiated reservation 2440 constructs a new RESERVE message to send downstream. The use of 2441 GIMPS service interface in this case is explained on Figure 35. Note 2442 that we assume the SII handling in GIMPS [I-D.ietf-nsis-ntlp] is 2443 extended to distinguish between own and peer SII. 2445 GIMPS QoS NSLP 2446 | | 2447 |<=====================================| 2448 | SendMessage{ | 2449 | NSLP-Data=RESERVE, | 2450 | Retain-State=TRUE, | 2451 | Size=X bytes, | 2452 | Message-Handle=NULL, | 2453 | NSLP-ID=QoS, | 2454 | Session-ID=SID_X, | 2455 | MRI=MRI, | 2456 | Direction=downstream, | 2457 | Own-SII-Handle=Own_SII_X, | 2458 | Peer-SII-Handle=empty | 2459 | Transfer-attributes=default, | 2460 | Timeout=default, | 2461 | IP-TTL=default} | 2462 | | 2464 Figure 35: GIMPS service interface usage for sending a 2465 sender-initiated reservation 2467 Note that an explicit preference for a particular type of transport, 2468 such as reliable/unreliable, may change the values of some service 2469 interface parameters (e.g. Transfer-attributes=unreliable). 2471 The message is received by the peer QNE. The use of GIMPS service 2472 interface when receiving a RESERVE message for a sender-initiated 2473 reservation is explained on Figure 36. 2475 GIMPS QoS NSLP 2476 | | 2477 |=====================================>| 2478 | RecvMessage{ | 2479 | NSLP-Data=RESERVE, | 2480 | Size=X bytes, | 2481 | Message-Handle=GIMPS_X, | 2482 | NSLP-ID=QoS, | 2483 | Session-ID=SID_X, | 2484 | MRI=MRI, | 2485 | Direction=downstream, | 2486 | Peer-SII-Handle=UP_SII_X, | 2487 | Transfer-attributes=default, | 2488 | IP-TTL=TTL_X, | 2489 | Original-TTL=TTL_Y} | 2490 | | 2491 |<=====================================| 2492 | MessageReceived{ | 2493 | Message-Handle=GIMPS_X, | 2494 | Retain-State=TRUE | 2495 | | 2497 Figure 36: GIMPS service interface usage for message reception of 2498 sender-initiated reservation 2500 7.2 Session identification 2502 The QoS NSLP keeps message and reservation state per session. A 2503 session is identified by a Session Identifier (SESSION_ID). The 2504 SESSION_ID is the primary index for stored NSLP state and needs to be 2505 constant and unique (with a sufficiently high probability) along a 2506 path through the network. On Figure 35, QoS NSLP picks a value SID_X 2507 for Session-ID. This value is subsequently used by GIMPS and QoS 2508 NSLP to refer to this session. 2510 7.3 Support for bypassing intermediate nodes 2512 The QoS NSLP may want to restrict the handling of its messages to 2513 specific nodes. This functionality is needed to support layering 2514 (explained in Section 3.2.8), when only the edge QNEs of a domain 2515 process the message. This requires a mechanism at GIMPS level (which 2516 can be invoked by the QoS NSLP) to bypass intermediates nodes between 2517 the edges of the domain. 2519 As a suggestion, we identified two ways for bypassing intermediate 2520 nodes. One solution is for the end-to-end session to carry a 2521 different protocol ID (QoS NSLP-E2E-IGNORE protocol ID, similar to 2522 the RSVP-E2E-IGNORE that is used for RSVP aggregation (RFC 3175 2523 [RFC3175]). Another solution is based on the use of multiple levels 2524 of the router alert option. In that case, internal routers are 2525 configured to handle only certain levels of router alerts. The 2526 choice between both approaches or another approach that fulfills the 2527 requirement is left to GIMPS design. 2529 7.4 Support for peer change identification 2531 There are several circumstances where it is necessary for a QNE to 2532 identify the adjacent QNE peer, which is the source of a signalling 2533 application message; for example, it may be to apply the policy that 2534 "state can only be modified by messages from the node that created 2535 it" or it might be that keeping track of peer identity is used as a 2536 (fallback) mechanism for rerouting detection at the NSLP layer. 2538 This functionality is implemented in GIMPS service interface with 2539 SII-handle. As shown in the above example, we assume the 2540 SII-handling will support both own SII and peer SII. 2542 Keeping track of the SII of a certain reservation also provides a 2543 means for the QoS NSLP to detect route changes. When a QNE receives 2544 a RESERVE referring to existing state but with a different SII, it 2545 knows that its upstream peer has changed. It can then use the old 2546 SII to initiate a teardown along the old section of the path. This 2547 functionality is supported in GIMPS service interface when the peer's 2548 SII which is stored on message reception is passed to GIMPS upon 2549 message transmission. 2551 7.5 Support for stateless operation 2553 Stateless or reduced state QoS NSLP operation makes the most sense 2554 when some nodes are able to operate in a stateless way at GIMPS level 2555 as well. Such nodes should not worry about keeping reverse state, 2556 message fragmentation and reassembly (at GIMPS), congestion control 2557 or security associations. A stateless or reduced state QNE will be 2558 able to inform the underlying GIMPS of this situation. GIMPS service 2559 interface supports this functionality with the Retain-State attribute 2560 in the MessageReceived primitive. 2562 7.6 Last node detection 2564 There are situations in which a QNE needs to determine whether it is 2565 the last QNE on the data path (QNR), e.g. to construct and send a 2566 RESPONSE message. 2568 A number of conditions may result in a QNE determining that it is the 2569 QNR: 2571 o the QNE may be the flow destination 2572 o the QNE have some other prior knowledge that it should act as the 2573 QNR 2574 o the QNE may be the last NSIS-capable node on the path 2575 o the QNE may be the last NSIS-capable node on the path supporting 2576 the QoS NSLP 2578 Of these four conditions, the last two can only be detected by GIMPS. 2579 We rely on GIMPS to inform the QoS NSLP about these cases by 2580 providing a trigger to the QoS NSLP when it determines that it is the 2581 last NE on the path, which supports the QoS NSLP. GIMPS supports 2582 this by the MessageDeliverError primitive. The error type 'no next 2583 node found' which is given as an example can be used. It is expected 2584 that additional error codes need to be defined. 2586 7.7 Re-routing detection 2588 Route changes may be detected at GIMPS layer or the information may 2589 be obtained by GIMPS through local interaction with or notification 2590 from routing protocols or modules. GIMPS allows to pass such 2591 information over the service interface using the NetworkNotification 2592 primitive with the appropriate 'downstream route change' or 'upstream 2593 route change' notification. 2595 7.8 Priority of signalling messages 2597 The QoS NSLP will generate messages with a range of performance 2598 requirements for GIMPS. These requirements may result from a 2599 prioritization at the QoS NSLP (Section 3.2.8) or from the 2600 responsiveness expected by certain applications supported by the QoS 2601 NSLP. 2603 GIMPS design should be able to ensure that performance for one class 2604 of messages was not degraded by aggregation with other classes of 2605 messages. GIMPS service interface supports this with the 'priority' 2606 transfer attribute. 2608 7.9 Knowledge of intermediate QoS NSLP unaware nodes 2610 In some cases it is useful to know that a reservation has not been 2611 installed at every router along the path. It is not possible to 2612 determine this using only NSLP functionality. 2614 GIMPS should be able to provide information to the NSLP about whether 2615 the message has passed through nodes that did not provide support for 2616 this NSLP. 2618 GIMPS service interface supports this by keeping track of IP-TTL and 2619 Original-TTL in the RecvMessage primitive. A difference between the 2620 two indiactes the number of QoS NSLP unaware nodes. 2622 7.10 NSLP Data Size 2624 When GIMPS passes the QoS NSLP data to the NSLP for processing, it 2625 must also indicate the size of that data. This is supported by the 2626 NSLP-Data-Size attribute. 2628 7.11 Notification of GIMPS 'D' flag value 2630 When GIMPS passes the QoS NSLP data to the NSLP for processing, it 2631 must also indicate the value of the 'D' (Direction) flag for that 2632 message. This is done in the Direction attribute of the SendMessage 2633 and RecvMessage primitives. 2635 7.12 NAT Traversal 2637 The QoS NSLP relies on GIMPS for NAT traversal. 2639 8. Assumptions on the QoS Model 2641 8.1 Resource sharing 2643 This specification assumes that resource sharing is possible between 2644 flows with the same SESSION_ID that originate from the same QNI or 2645 between flows with a different SESSION_ID that are related through 2646 the BOUND_SESSION_ID object. For flows with the same SESSION_ID, 2647 resource sharing is only applicable when the existing reservation is 2648 not just replaced (which is indicated by the REPLACE flag in the 2649 common header. 2651 The Resource Management Function (RMF) reserves resources for each 2652 flow. We assume that the QoS model supports resource sharing between 2653 flows. A QoS Model may elect to implement a more general behaviour 2654 of supporting relative operations on existing reservations, such as 2655 ADDING or SUBTRACTING a certain amount of resources from the current 2656 reservation. A QoS Model may also elect to allow resource sharing 2657 more generally, e.g. between all flows with the same DSCP. 2659 8.2 Reserve/commit support 2661 Reserve/commit behaviour means that the time at which the reservation 2662 is made may be different from the time when the reserved resources 2663 are actually set aside for the requesting session. This 2664 specification acknowledges the usefulness of such a mechanism but 2665 assumes that its implementation is opaque to QoS NSLP and is fully 2666 handled by the QoS Model. 2668 9. Open issues 2670 9.1 Peering agreements on interdomain links 2672 This specification proposes ways to carry AAA information that may be 2673 used at the edges of a domain to check whether the requestor is 2674 allowed to use the requested resources. It is less likely that the 2675 AAA information will be used inside a domain. In practice, there may 2676 be peering relations between domains that allow for a certain amount 2677 of traffic to be sent on an interdomain link without the need to 2678 check the authorization of each individual session (effectively 2679 making the peering domain the requestor of the resources). The 2680 per-session authorization check may be avoided by setting up an 2681 aggregate reservation on the inter-domain link for a specified amount 2682 of resources and relating the end-to-end sessions to it using the 2683 BOUND_SESSION_ID. In this way, the aggregate session is authorized 2684 once (and infrequently updated). An alternative is for the edge node 2685 of a domain to insert a token that authorizes the flow for the next 2686 domain. 2688 9.2 Protocol Operating Environment Assumptions 2690 The NSIS protocol is not used alone. Rather, it is used in 2691 conjunction with a variety of applications. For receiver initiated 2692 and bidirectional reservations the question arises of what the 2693 interactions are between the NSIS protocols and the end-to-end 2694 applications. An assumption needs to be made about what information 2695 should be determined outside the NSIS protocols, and what should be 2696 carried end-to-end in NSLP messages in order to initiate signalling. 2698 For a receiver initiated reservation, the we have the questions: How 2699 do the sender and receiver determine that a receiver initiated 2700 reservation is to be performed? And, how does information needed by 2701 the receiver to perform the reservation, but only available at the 2702 sender, be made transferred to the receiver so that the RESERVE 2703 message can be sent? 2705 In the bi-directional reservation case, we can either perform this as 2706 a pair of two sender-initiated reservations or as a combination of 2707 sender-initiated and receiver-initiated reservations. The latter 2708 case has the same issues as for the general receiver initiated 2709 reservation problem. The former raises similar questions: How does 2710 the remote end know that a reservation is needed? And, how does it 2711 know what resources to request? 2713 Is it reasonable to assume that the decision that an end should 2714 initiate a reservation is made totally outside the QoS NSLP itself 2715 (e.g. through prior configuration, or application end-to-end 2716 signalling such as SIP) or, should the QoS NSLP messages include some 2717 method to trigger the other end to perform a reservation (whether 2718 that be a receiver initiated reservation, or a sender initiated 2719 reservation for the first bidirectional reservation case)? 2721 In addition, should the QoS NSLP messages be able to carry extra data 2722 (e.g. a QSPEC object for the reverse direction) end-to-end that is 2723 needed by the remote end to perform its reservation? (And, should 2724 this be in the QoS NSLP, or through individual QoS models?) The 2725 alternative to providing support in the QoS NSLP for this is to leave 2726 it to application signalling to transfer any required information. 2728 9.3 Authorization components in QoS NSLP 2730 It is unclear whether the Authorization Model Examples (in 2731 Section 4.9) really belong in this document. They do not have a 2732 direct impact on the NSLP itself, but rather discuss other 2733 interactions that may occur at a QNE due to QoS NSLP processing. 2734 They may also be generic across multiple NSLPs, and so fit better in 2735 an external document for that reason. 2737 10. Security Considerations 2739 10.1 Introduction and Threat Overview 2741 The security requirement for the QoS NSLP is to protect the 2742 signalling exchange for establishing QoS reservations against 2743 identified security threats. For the signalling problem as a whole, 2744 these threats have been outlined in NSIS threats 2745 [I-D.ietf-nsis-threats]; the NSIS framework [I-D.ietf-nsis-fw] 2746 assigns a subset of the responsibility to GIMPS and the remaining 2747 threats need to be addressed by NSLPs. The main issues to be handled 2748 can be summarised as: 2749 Authorization: 2751 The QoS NSLP must assure that the network is protected against 2752 theft-of-service by offering mechanisms to authorize the QoS 2753 reservation requestor. A user requesting a QoS reservation might 2754 want proper resource accounting and protection against spoofing 2755 and other security vulnerabilities which lead to denial of service 2756 and financial loss. In many cases authorization is based on the 2757 authenticated identity. The authorization model must provide 2758 guarantees that replay attacks are either not possible or limited 2759 to a certain extent. Authorization can also be based on traits 2760 which enables the user to remain anonymous. Support for user 2761 identity confidentiality can be accomplished. 2763 Message Protection: 2765 Signalling message content should be protected against 2766 modification, replay, injection and eavesdropping while in 2767 transit. Authorization information, such as authorization tokens, 2768 need protection. This type of protection at the NSLP layer is 2769 neccessary to protect messages between NSLP nodes which includes 2770 end-to-middle, middle-to-middle and even end-to-end protection. 2772 In addition to the above-raised issues we see the following 2773 functionality provided at the NSLP layer: 2774 Prevention of Denial of Service Attacks: 2776 GIMPS and QoS NSLP nodes have finite resources (state storage, 2777 processing power, bandwidth). The protocol mechanisms suggested 2778 in this document should try to minimise exhaustion attacks against 2779 these resources when performing authentication and authorization 2780 for QoS resources. 2782 To some extent the QoS NSLP relies on the security mechanisms 2783 provided by GIMPS which by itself relies on existing authentication 2784 and key exchange protocols. Some signalling messages cannot be 2785 protected by GIMPS and hence should be used with care by the QoS 2786 NSLP. An API must ensure that the QoS NSLP implementation is aware 2787 of the underlying security mechanisms and must be able to indicate 2788 which degree of security is provided between two GIMPS peers. If a 2789 level of security protection for QoS NSLP messages is required which 2790 goes beyond the security offered by GIMPS or underlying security 2791 mechanisms, additional security mechanisms described in this document 2792 must be used. The different usage environments and the different 2793 scenarios where NSIS is used make it very difficult to make general 2794 statements without reducing its flexibility. 2796 10.2 Trust Model 2798 For this version of the document we will rely on a model which 2799 requires trust between neighboring NSLP nodes to establish a 2800 chain-of-trust along the QoS signalling path. This model is simple 2801 to deploy, was used in previous QoS authorization environments (such 2802 as RSVP) and seems to provide sufficiently strong security 2803 properties. We refer to this model as the 'New Jersey Turnpike' 2804 model. 2806 On the New Jersey Turnpike, motorists pick up a ticket at a toll 2807 booth when entering the highway. At the highway exit the ticket is 2808 presented and payment is made at the toll booth for the distance 2809 driven. For QoS signalling in the Internet this procedure is roughly 2810 similar. In most cases the data sender is charged for transmitted 2811 data traffic where charging is provided only between neighboring 2812 entities. 2814 +------------------+ +------------------+ +------------------+ 2815 | Network | | Network | | Network | 2816 | X | | Y | | Z | 2817 | | | | | | 2818 | -----------> -----------> | 2819 | | | | | | 2820 | | | | | | 2821 +--------^---------+ +------------------+ +-------+----------+ 2822 | . 2823 | . 2824 | v 2825 +--+---+ Data Data +--+---+ 2826 | Node | ==============================> | Node | 2827 | A | Sender Receiver | B | 2828 +------+ +------+ 2830 Legend: 2832 ----> Peering relationship which allows neighboring 2833 networks/entities to charge each other for the 2834 QoS reservation and data traffic 2836 ====> Data flow 2838 ..... Communication to the end host 2840 Figure 37: New Jersey Turnpike Model 2842 The model shown in Figure 37 uses peer-to-peer relationships between 2843 different administrative domains as a basis for accounting and 2844 charging. As mentioned above, based on the peering relationship a 2845 chain-of-trust is established. There are several issues which come 2846 to mind when considering this type of model: 2847 o This model allows authorization on a request basis or on a 2848 per-session basis. Authorization mechanisms will be elaborated in 2849 Section 4.9. The duration for which the QoS authorization is 2850 valid needs to be controlled. Combining the interval with the 2851 soft-state interval is possible. Notifications from the networks 2852 also seem to be viable approach. 2853 o The price for a QoS reservation needs to be determined somehow and 2854 communicated to the charged entity and to the network where the 2855 charged entity is attached. Price distribution protocols are not 2856 covered in this version of the document. This model assumes, per 2857 default, that the data sender is authorizing the QoS reservation. 2858 Please note that this is only a simplification and further 2859 extensions are possible and left for a future version of this 2860 document. 2862 o This architecture seems to be simple enough to allow a scalable 2863 solution (ignoring reverse charging, multicast issues and price 2864 distribution). 2866 Charging the data sender as performed in this model simplifies 2867 security handling by demanding only peer-to-peer security protection. 2868 Node A would perform authentication and key establishment. The 2869 established security association (together with the session key) 2870 would allow the user to protect QoS signalling messages. The 2871 identity used during the authentication and key establishment phase 2872 would be used by Network X (see Figure 37) to perform the so-called 2873 policy-based admission control procedure. In our context this user 2874 identifier would be used to establish the necessary infrastructure to 2875 provide authorization and charging. Signalling messages later 2876 exchanged between the different networks are then also subject to 2877 authentication and authorization. The authenticated entity thereby 2878 is, however, the neighboring network and not the end host. 2880 The New Jersey Turnpike model is attractive because of its 2881 simplicity. S. Schenker et. al. [shenker-pricing] discuss various 2882 accounting implications and introduced the edge pricing model. The 2883 edge pricing model shows similarity to the model described in this 2884 section with the exception that mobility and the security 2885 implications itself are not addressed. 2887 10.3 Computing the authorization decision 2889 Whenever an authorization decision has to be made then there is the 2890 question which information serves as an input to the authorizing 2891 entity. The following information items have been mentioned in the 2892 past for computing the authorization decision (in addition to the 2893 authenticated identity): 2894 Price 2895 QoS objects 2896 Policy rules 2898 Policy rules include attributes like time of day, subscription to 2899 certain services, membership, etc. into consideration when computing 2900 an authorization decision. 2902 A detailed description of the authorization handling will be left for 2903 a future version of this document. The authors assume that the QoS 2904 NSLP needs to provide a number of attributes to support the large 2905 range of scenarios. 2907 11. Acknowledgements 2909 The authors would like to thank Eleanor Hepworth, Ruediger Geib, 2910 Roland Bless and Nemeth Krisztian for their useful comments. 2912 Bob Braden provided helpful comments and guidance which were 2913 gratefully received. 2915 12. Contributors 2917 This draft combines work from three individual drafts. The following 2918 authors from these drafts also contributed to this document: Robert 2919 Hancock (Siemens/Roke Manor Research), Hannes Tschofenig and Cornelia 2920 Kappler (Siemens AG), Lars Westberg and Attila Bader (Ericsson) and 2921 Maarten Buechli (Dante) and Eric Waegeman (Alcatel). 2923 Yacine El Mghazli (Alcatel) contributed text on AAA. 2925 13. References 2927 13.1 Normative References 2929 [I-D.ietf-nsis-ntlp] 2930 Schulzrinne, H., "GIMPS: General Internet Messaging 2931 Protocol for Signaling", 2932 Internet-Draft draft-ietf-nsis-ntlp-04, October 2004. 2934 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2935 Requirement Levels", BCP 14, RFC 2119, March 1997. 2937 [RFC2234] Crocker, D., Ed. and P. Overell, "Augmented BNF for Syntax 2938 Specifications: ABNF", RFC 2234, November 1997. 2940 13.2 Informative References 2942 [I-D.ash-nsis-y1541-qsp] 2943 Ash, J., "NSIS QoS Signaling Policy for Networks Using 2944 Y.1541 QoS Classes", 2945 Internet-Draft draft-ash-nsis-y1541-qsp-00, December 2004. 2947 [I-D.ietf-nsis-fw] 2948 Hancock, R., "Next Steps in Signaling: Framework", 2949 Internet-Draft draft-ietf-nsis-fw-07, December 2004. 2951 [I-D.ietf-nsis-qspec] 2952 Ash, J., "QoS-NSLP QSpec Template", 2953 Internet-Draft draft-ietf-nsis-qspec-03, February 2005. 2955 [I-D.ietf-nsis-rmd] 2956 Bader, A., "RMD-QOSM - The Resource Management in Diffserv 2957 QoS model", Internet-Draft draft-ietf-nsis-rmd-01, 2958 February 2005. 2960 [I-D.ietf-nsis-threats] 2961 Tschofenig, H. and D. Kroeselberg, "Security Threats for 2962 NSIS", Internet-Draft draft-ietf-nsis-threats-06, October 2963 2004. 2965 [I-D.kappler-nsis-qosmodel-controlledload] 2966 Kappler, C., "A QoS Model for Signaling IntServ 2967 Controlled-Load Service with NSIS", 2968 Internet-Draft draft-kappler-nsis-qosmodel-controlledload-00 2969 , February 2004. 2971 [I-D.manner-lrsvp] 2972 Manner, J., "Localized RSVP", 2973 Internet-Draft draft-manner-lrsvp-04, September 2004. 2975 [I-D.tschofenig-nsis-aaa-issues] 2976 Tschofenig, H., "NSIS Authentication, Authorization and 2977 Accounting Issues", 2978 Internet-Draft draft-tschofenig-nsis-aaa-issues-01, March 2979 2003. 2981 [I-D.tschofenig-nsis-qos-authz-issues] 2982 Tschofenig, H., "QoS NSLP Authorization Issues", 2983 Internet-Draft draft-tschofenig-nsis-qos-authz-issues-00, 2984 June 2003. 2986 [I-D.westberg-rmd-framework] 2987 Westberg, L., "Resource Management In Diffserv: An NSIS 2988 QoS Signalling Model for Diffserv Networks", 2989 draft-westberg-rmd-framework-04.txt, work in progress, 2990 September 2003. 2992 [MEF.EthernetServicesModel] 2993 Metro Ethernet Forum, "Ethernet Services Model", letter 2994 ballot document , August 2003. 2996 [OSP] ETSI, "Telecommunications and internet protocol 2997 harmonization over networks (tiphon); open settlement 2998 protocol (osp) for inter- domain pricing, authorization, 2999 and usage exchange", Technical Specification 101 321, 3000 version 2.1.0. 3002 [RFC1633] Braden, B., Clark, D. and S. Shenker, "Integrated Services 3003 in the Internet Architecture: an Overview", RFC 1633, June 3004 1994. 3006 [RFC2205] Braden, B., Zhang, L., Berson, S., Herzog, S. and S. 3007 Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1 3008 Functional Specification", RFC 2205, September 1997. 3010 [RFC2210] Wroclawski, J., "The Use of RSVP with IETF Integrated 3011 Services", RFC 2210, September 1997. 3013 [RFC2434] Narten, T. and H. Alvestrand, "Guidelines for Writing an 3014 IANA Considerations Section in RFCs", BCP 26, RFC 2434, 3015 October 1998. 3017 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z. 3018 and W. Weiss, "An Architecture for Differentiated 3019 Services", RFC 2475, December 1998. 3021 [RFC2961] Berger, L., Gan, D., Swallow, G., Pan, P., Tommasi, F. and 3022 S. Molendini, "RSVP Refresh Overhead Reduction 3023 Extensions", RFC 2961, April 2001. 3025 [RFC3175] Baker, F., Iturralde, C., Le Faucheur, F. and B. Davie, 3026 "Aggregation of RSVP for IPv4 and IPv6 Reservations", 3027 RFC 3175, September 2001. 3029 [RFC3520] Hamer, L-N., Gage, B., Kosinski, B. and H. Shieh, "Session 3030 Authorization Policy Element", RFC 3520, April 2003. 3032 [RFC3521] Hamer, L-N., Gage, B. and H. Shieh, "Framework for Session 3033 Set-up with Media Authorization", RFC 3521, April 2003. 3035 [RFC3583] Chaskar, H., "Requirements of a Quality of Service (QoS) 3036 Solution for Mobile IP", RFC 3583, September 2003. 3038 [RFC3726] Brunner, M., "Requirements for Signaling Protocols", 3039 RFC 3726, April 2004. 3041 [_XREF_FJ94] 3042 Jacobson, V., "Synchronization of Periodic Routing 3043 Messages", IEEE/ACM Transactions on Networking , Vol. 2 , 3044 No. 2 , April 1994. 3046 [_XREF_OPWA95] 3047 Breslau, L., "Two Issues in Reservation Establishment", 3048 Proc. ACM SIGCOMM '95 , Cambridge , MA , August 1995. 3050 [shenker-pricing] 3051 Shenker, S., Clark, D., Estrin, D. and S. Herzog, "Pricing 3052 in computer networks: Reshaping the research agenda", 3053 Proc. of TPRC 1995, 1995. 3055 Authors' Addresses 3057 Sven Van den Bosch 3058 Alcatel 3059 Francis Wellesplein 1 3060 Antwerpen B-2018 3061 Belgium 3063 Email: sven.van_den_bosch@alcatel.be 3064 Georgios Karagiannis 3065 University of Twente/Ericsson 3066 P.O. Box 217 3067 Enschede 7500 AE 3068 The Netherlands 3070 Email: karagian@cs.utwente.nl 3072 Andrew McDonald 3073 Siemens/Roke Manor Research 3074 Roke Manor Research Ltd. 3075 Romsey, Hants SO51 0ZN 3076 UK 3078 Email: andrew.mcdonald@roke.co.uk 3080 Appendix A. Glossary 3082 AAA: Authentication, Authorization and Accounting 3083 EAP: Extensible Authentication Protocol 3084 MRI: Message Routing Information (see [I-D.ietf-nsis-ntlp]) 3085 NAT: Network Address Translator 3086 NSLP: NSIS Signaling Layer Protocol (see [I-D.ietf-nsis-fw]) 3087 NTLP: NSIS Transport Layer Protocol (see [I-D.ietf-nsis-fw]) 3088 OPWA: One Pass With Advertising 3089 OSP: Open Settlement Protocol 3090 PIN: Policy Ignorant Node 3091 QNE: an NSIS Entity (NE), which supports the QoS-NSLP (see Section 2) 3092 QNI: the first node in the sequence of QNEs that issues a reservation 3093 request for a session (see Section 2) 3094 QNR: the last node in the sequence of QNEs that receives a 3095 reservation request for a session (see Section 2) 3096 QSpec: Quality of Service Specification 3097 SII: Source Identification Information 3098 SIP: Session Initiation Protocol 3099 RII: Request Identification Information 3100 RMD: Resource Management for DiffServ 3101 RMF: Resource Management Function 3102 RSN: Reservation Sequence Number 3103 RSVP: Resource reSerVation Protocol (see [RFC2205]) 3105 Appendix B. Change History 3107 Note to RFC Editor: This section is to be removed before publication. 3109 Changes from -00 3110 * Additional explanation of RSN versus Session ID differences. 3111 (Session IDs still need to be present and aren't replaced by 3112 RSNs. Explain how QoS NSLP could react once it notes that it 3113 maintains stale state.) 3114 * Additional explanation of message types - why we don't just 3115 have RESERVE and RESPONSE. 3116 * Clarified that figure 1 is not an implementation restriction. 3117 Changes from -01 3118 * Significant restructuring. 3119 * Added more concrete details of message formats and processing. 3120 * Added description of layering/aggregation concepts. 3121 * Added details of authentication/authorisation aspects. 3122 Changes from -02 3123 * Addressed comments from early review. 3124 * Added text on receiver-initiated and bi-directional 3125 reservations. 3126 * Extended description of session binding. Added support for 3127 fate sharing. 3128 * Restructured message formats and processing section. 3129 * Clarified refresh reduction mechanism. 3130 * Added assumptions on QSM. 3131 * Added assumptions on operating environment. 3132 Changes from -03 3133 * Removed overlaps between sections. 3134 * Clarified document does not specify how to aggregate individual 3135 end-to-end flow from a resource point of view but rather how 3136 such an aggregate can be signalled for. 3137 * Made session binding purely informational. 3138 * Clarified QSPEC stacking. 3139 * Added object format for ERROR_SPEC object. 3140 * Made RII a separate object from RESPONSE_REQUEST and outside of 3141 the SCOPING object. Then removed RESPONSE_REQUEST and made 3142 SCOPING a flag rather than an object. 3143 * Closed open issue of "PATH" message functionality. An empty 3144 QUERY is used to install reverse state along the path. 3145 * Made all flag names positive. Removed NO_FATE_SHARING flag: 3146 fate sharing is not supported by the signalling. 3147 * Removed the open issue on one-sided bidirectional reservation. 3148 Clarified how it can be done, even for stateless or reduced 3149 state domains in an example. 3150 * Removed open issue on priority. Message priority will be 3151 handled over GIMPS API, reservation priority is an issue for 3152 the RMF. 3154 Changes from -04 3155 * Resolved a number of outstanding comments on clarifications 3156 (likelihood of transport type, bidirectional reservations, 3157 handle of RESERVE messages inside a domain in case of 3158 aggregation or reduced state operation) from the mailing list. 3159 * Introduced a default value for REFRESH_PERIOD. 3160 * Introduced explicit feedback mechanism in case of route 3161 changes. 3162 * State acknowledgment is now supported by means of an 3163 ACKNOWLEDGE flag. This is made the default case. 3164 * Changed section 7 to reflect the use of GIMPS service 3165 interface. 3166 Changes from -05 3167 * Modified definitions of QoS Model and NSLP/QSpec relationships. 3168 Removed concepts of QoS Signalling Model (QSM) and QoS 3169 Signalling Policy (QSP). 3170 * Made changes to the policy control and authentication concepts. 3171 Removed old appendix on original POLICY_CONTROL object. 3172 * Added a glossary. 3173 * Added text on reservation collision handling. 3174 * Moved this list of changes to the last appendix to make it 3175 easier to remove at publication time. 3177 Intellectual Property Statement 3179 The IETF takes no position regarding the validity or scope of any 3180 Intellectual Property Rights or other rights that might be claimed to 3181 pertain to the implementation or use of the technology described in 3182 this document or the extent to which any license under such rights 3183 might or might not be available; nor does it represent that it has 3184 made any independent effort to identify any such rights. Information 3185 on the procedures with respect to rights in RFC documents can be 3186 found in BCP 78 and BCP 79. 3188 Copies of IPR disclosures made to the IETF Secretariat and any 3189 assurances of licenses to be made available, or the result of an 3190 attempt made to obtain a general license or permission for the use of 3191 such proprietary rights by implementers or users of this 3192 specification can be obtained from the IETF on-line IPR repository at 3193 http://www.ietf.org/ipr. 3195 The IETF invites any interested party to bring to its attention any 3196 copyrights, patents or patent applications, or other proprietary 3197 rights that may cover technology that may be required to implement 3198 this standard. Please address the information to the IETF at 3199 ietf-ipr@ietf.org. 3201 Disclaimer of Validity 3203 This document and the information contained herein are provided on an 3204 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 3205 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET 3206 ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, 3207 INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE 3208 INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 3209 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 3211 Copyright Statement 3213 Copyright (C) The Internet Society (2005). This document is subject 3214 to the rights, licenses and restrictions contained in BCP 78, and 3215 except as set forth therein, the authors retain all their rights. 3217 Acknowledgment 3219 Funding for the RFC Editor function is currently provided by the 3220 Internet Society.