idnits 2.17.1 draft-ietf-anima-autonomic-control-plane-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack separate sections for Informative/Normative References. All references will be assumed normative when checking for downward references. ** The document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 251: '... 1. The ACP SHOULD provide robust c...' RFC 2119 keyword, line 256: '... 2. The ACP MUST have a separate ad...' RFC 2119 keyword, line 260: '... 3. The ACP MUST use autonomically ...' RFC 2119 keyword, line 265: '... 4. The ACP MUST be generic. Usabl...' RFC 2119 keyword, line 266: '...rastructure. It MUST NOT be tied to a...' (21 more instances...) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 8, 2016) is 2842 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) == Missing Reference: 'TBD' is mentioned on line 599, but not defined == Unused Reference: 'RFC5082' is defined on line 1355, but no explicit reference was found in the text ** Downref: Normative reference to an Informational draft: draft-eckert-anima-stable-connectivity (ref. 'I-D.eckert-anima-stable-connectivity') == Outdated reference: A later version (-45) exists of draft-ietf-anima-bootstrapping-keyinfra-03 == Outdated reference: A later version (-15) exists of draft-ietf-anima-grasp-06 == Outdated reference: A later version (-10) exists of draft-ietf-anima-reference-model-02 ** Downref: Normative reference to an Informational draft: draft-ietf-anima-reference-model (ref. 'I-D.ietf-anima-reference-model') ** Obsolete normative reference: RFC 6347 (Obsoleted by RFC 9147) ** Downref: Normative reference to an Informational RFC: RFC 7404 ** Downref: Normative reference to an Informational RFC: RFC 7575 ** Downref: Normative reference to an Informational RFC: RFC 7576 Summary: 8 errors (**), 0 flaws (~~), 6 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 ANIMA WG M. Behringer, Ed. 3 Internet-Draft T. Eckert 4 Intended status: Standards Track Cisco Systems 5 Expires: January 9, 2017 S. Bjarnason 6 Arbor Networks 7 July 8, 2016 9 An Autonomic Control Plane 10 draft-ietf-anima-autonomic-control-plane-03 12 Abstract 14 Autonomic functions need a control plane to communicate, which 15 depends on some addressing and routing. This Autonomic Control Plane 16 should ideally be self-managing, and as independent as possible of 17 configuration. This document defines an "Autonomic Control Plane", 18 with the primary use as a control plane for autonomic functions. It 19 also serves as a "virtual out of band channel" for OAM communications 20 over a network that is not configured, or mis-configured. 22 Status of This Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on January 9, 2017. 39 Copyright Notice 41 Copyright (c) 2016 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 57 2. Use Cases for an Autonomic Control Plane . . . . . . . . . . 4 58 2.1. An Infrastructure for Autonomic Functions . . . . . . . . 4 59 2.2. Secure Bootstrap over an Unconfigured Network . . . . . . 4 60 2.3. Data Plane Independent Permanent Reachability . . . . . . 5 61 3. Requirements . . . . . . . . . . . . . . . . . . . . . . . . 6 62 4. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . 6 63 5. Self-Creation of an Autonomic Control Plane . . . . . . . . . 8 64 5.1. Preconditions . . . . . . . . . . . . . . . . . . . . . . 8 65 5.1.1. Domain Certificate with ANIMA information . . . . . . 8 66 5.1.2. AN Adjacency Table . . . . . . . . . . . . . . . . . 9 67 5.2. Neighbor discovery via mDNS/DNS-SD . . . . . . . . . . . 9 68 5.3. Candidate ACP Neighbor Selection . . . . . . . . . . . . 10 69 5.4. Channel Selection . . . . . . . . . . . . . . . . . . . . 11 70 5.5. Security Association protocols . . . . . . . . . . . . . 12 71 5.5.1. ACP via IPsec . . . . . . . . . . . . . . . . . . . . 13 72 5.5.2. ACP via GRE/IPsec . . . . . . . . . . . . . . . . . . 13 73 5.5.3. ACP via dTLS . . . . . . . . . . . . . . . . . . . . 13 74 5.5.4. GRASP/TLS negotiation . . . . . . . . . . . . . . . . 13 75 5.5.5. ACP Security Profiles . . . . . . . . . . . . . . . . 14 76 5.6. GRASP instance details . . . . . . . . . . . . . . . . . 14 77 5.7. Context Separation . . . . . . . . . . . . . . . . . . . 14 78 5.8. Addressing inside the ACP . . . . . . . . . . . . . . . . 15 79 5.8.1. Fundamental Concepts of Autonomic Addressing . . . . 15 80 5.8.2. The ACP Addressing Base Scheme . . . . . . . . . . . 16 81 5.8.3. ACP Addressing Sub-Scheme . . . . . . . . . . . . . . 17 82 5.8.4. Usage of the Zone Field . . . . . . . . . . . . . . . 18 83 5.8.5. Other ACP Addressing Sub-Schemes . . . . . . . . . . 19 84 5.9. Routing in the ACP . . . . . . . . . . . . . . . . . . . 19 85 5.10. General ACP Considerations . . . . . . . . . . . . . . . 19 86 6. Workarounds for Non-Autonomic Nodes . . . . . . . . . . . . . 20 87 6.1. Connecting a Non-Autonomic Controller / NMS system . . . 20 88 6.2. ACP through Non-Autonomic L3 Clouds . . . . . . . . . . . 20 89 7. Self-Healing Properties . . . . . . . . . . . . . . . . . . . 21 90 8. Self-Protection Properties . . . . . . . . . . . . . . . . . 22 91 9. The Administrator View . . . . . . . . . . . . . . . . . . . 22 92 10. Explanations . . . . . . . . . . . . . . . . . . . . . . . . 23 93 10.1. Why GRASP to discover autonomic neighbors . . . . . . . 23 94 11. Security Considerations . . . . . . . . . . . . . . . . . . . 24 95 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 25 96 13. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 25 97 14. Change log [RFC Editor: Please remove] . . . . . . . . . . . 25 98 14.1. Initial version . . . . . . . . . . . . . . . . . . . . 26 99 14.2. draft-behringer-anima-autonomic-control-plane-00 . . . . 26 100 14.3. draft-behringer-anima-autonomic-control-plane-01 . . . . 26 101 14.4. draft-behringer-anima-autonomic-control-plane-02 . . . . 26 102 14.5. draft-behringer-anima-autonomic-control-plane-03 . . . . 26 103 14.6. draft-ietf-anima-autonomic-control-plane-00 . . . . . . 27 104 14.7. draft-ietf-anima-autonomic-control-plane-01 . . . . . . 27 105 14.8. draft-ietf-anima-autonomic-control-plane-02 . . . . . . 28 106 14.9. draft-ietf-anima-autonomic-control-plane-03 . . . . . . 28 107 15. References . . . . . . . . . . . . . . . . . . . . . . . . . 28 108 Appendix A. Background on the choice of routing protocol . . . . 30 109 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 32 111 1. Introduction 113 Autonomic Networking is a concept of self-management: Autonomic 114 functions self-configure, and negotiate parameters and settings 115 across the network. [RFC7575] defines the fundamental ideas and 116 design goals of Autonomic Networking. A gap analysis of Autonomic 117 Networking is given in [RFC7576]. The reference architecture for 118 Autonomic Networking in the IETF is currently being defined in the 119 document [I-D.ietf-anima-reference-model] 121 Autonomic functions need a stable and robust infrastructure to 122 communicate on. This infrastructure should be as robust as possible, 123 and it should be re-usable by all autonomic functions. [RFC7575] 124 calls it the "Autonomic Control Plane". This document defines the 125 Autonomic Control Plane. 127 Today, the management and control plane of networks typically runs in 128 the global routing table, which is dependent on correct configuration 129 and routing. Misconfigurations or routing problems can therefore 130 disrupt management and control channels. Traditionally, an out of 131 band network has been used to recover from such problems, or 132 personnel is sent on site to access devices through console ports. 133 However, both options are operationally expensive. 135 In increasingly automated networks either controllers or distributed 136 autonomic service agents in the network require a control plane which 137 is independent of the network they manage, to avoid impacting their 138 own operations. 140 This document describes options for a self-forming, self-managing and 141 self-protecting "Autonomic Control Plane" (ACP) which is inband on 142 the network, yet as independent as possible of configuration, 143 addressing and routing problems (for details how this achieved, see 144 Section 5). It therefore remains operational even in the presence of 145 configuration errors, addressing or routing issues, or where policy 146 could inadvertently affect control plane connectivity. The Autonomic 147 Control Plane serves several purposes at the same time: 149 o Autonomic functions communicate over the ACP. The ACP therefore 150 supports directly Autonomic Networking functions, as described in 151 [I-D.ietf-anima-reference-model]. For example, GRASP 152 [I-D.ietf-anima-grasp] can run inside the ACP. 154 o An operator can use it to log into remote devices, even if the 155 data plane is misconfigured or unconfigured. 157 o A controller or network management system can use it to securely 158 bootstrap network devices in remote locations, even if the network 159 in between is not yet configured; no data-plane dependent 160 bootstrap configuration is required. An example of such a secure 161 bootstrap process is described in 162 [I-D.ietf-anima-bootstrapping-keyinfra] 164 This document describes some use cases for the ACP in Section 2, it 165 defines the requirements in Section 3, Section 4 gives an overview 166 how an Autonomic Control Plane is constructed, and in Section 5 the 167 detailed process is explained. Section 6 explains how non-autonomic 168 nodes and networks can be integrated, and Section 5.5 the first 169 channel types for the ACP. 171 The document "Autonomic Network Stable Connectivity" 172 [I-D.eckert-anima-stable-connectivity] describes how the ACP can be 173 used to provide stable connectivity for OAM applications. It also 174 explains on how existing management solutions can leverage the ACP in 175 parallel with traditional management models, when to use the ACP 176 versus the data plane, how to integrate IPv4 based management, etc. 178 2. Use Cases for an Autonomic Control Plane 180 2.1. An Infrastructure for Autonomic Functions 182 Autonomic Functions need a stable infrastructure to run on, and all 183 autonomic functions should use the same infrastructure to minimise 184 the complexity of the network. This way, there is only need for a 185 single discovery mechanism, a single security mechanism, and other 186 processes that distributed functions require. 188 2.2. Secure Bootstrap over an Unconfigured Network 190 Today, bootstrapping a new device typically requires all devices 191 between a controlling node (such as an SDN controller) and the new 192 device to be completely and correctly addressed, configured and 193 secured. Therefore, bootstrapping a network happens in layers around 194 the controller. Without console access (for example through an out 195 of band network) it is not possible today to make devices securely 196 reachable before having configured the entire network between. 198 With the ACP, secure bootstrap of new devices can happen without 199 requiring any configuration on the network. A new device can 200 automatically be bootstrapped in a secure fashion and be deployed 201 with a domain certificate. This does not require any configuration 202 on intermediate nodes, because they can communicate through the ACP. 204 2.3. Data Plane Independent Permanent Reachability 206 Today, most critical control plane protocols and network management 207 protocols are running in the data plane (global routing table) of the 208 network. This leads to undesirable dependencies between control and 209 management plane on one side and the data plane on the other: Only if 210 the data plane is operational, will the other planes work as 211 expected. 213 Data plane connectivity can be affected by errors and faults, for 214 example certain AAA misconfigurations can lock an administrator out 215 of a device; routing or addressing issues can make a device 216 unreachable; shutting down interfaces over which a current management 217 session is running can lock an admin irreversibly out of the device. 218 Traditionally only console access can help recover from such issues. 220 Data plane dependencies also affect NOC/SDN controller applications: 221 Certain network changes are today hard to operate, because the change 222 itself may affect reachability of the devices. Examples are address 223 or mask changes, routing changes, or security policies. Today such 224 changes require precise hop-by-hop planning. 226 The ACP provides reachability that is largely independent of the data 227 plane, which allows control plane and management plane to operate 228 more robustly: 230 o For management plane protocols, the ACP provides the functionality 231 of a "Virtual-out-of-band (VooB) channel", by providing 232 connectivity to all devices regardless of their configuration or 233 global routing table. 235 o For control plane protocols, the ACP allows their operation even 236 when the data plane is temporarily faulty, or during transitional 237 events, such as routing changes, which may affect the control 238 plane at least temporarily. This is specifically important for 239 autonomic service agents, which could affect data plane 240 connectivity. 242 The document "Autonomic Network Stable Connectivity" 243 [I-D.eckert-anima-stable-connectivity] explains the use cases for the 244 ACP in significantly more detail and explains how the ACP can be used 245 in practical network operations. 247 3. Requirements 249 The Autonomic Control Plane has the following requirements: 251 1. The ACP SHOULD provide robust connectivity: As far as possible, 252 it should be independent of configured addressing, configuration 253 and routing. Requirements 2 and 3 build on this requirement, but 254 also have value on their own. 256 2. The ACP MUST have a separate address space from the data plane. 257 Reason: traceability, debug-ability, separation from data plane, 258 security (can block easily at edge). 260 3. The ACP MUST use autonomically managed address space. Reason: 261 easy bootstrap and setup ("autonomic"); robustness (admin can't 262 mess things up so easily). This document suggests to use ULA 263 addressing for this purpose. 265 4. The ACP MUST be generic. Usable by all the functions and 266 protocols of the AN infrastructure. It MUST NOT be tied to a 267 particular protocol. 269 5. The ACP MUST provide security: Messages coming through the ACP 270 MUST be authenticated to be from a trusted node, and SHOULD (very 271 strong SHOULD) be encrypted. 273 The default mode of operation of the ACP is hop-by-hop, because this 274 interaction can be built on IPv6 link local addressing, which is 275 autonomic, and has no dependency on configuration (requirement 1). 276 It may be necessary to have end-to-end connectivity in some cases, 277 for example to provide an end-to-end security association for some 278 protocols. This is possible, but then has a dependency on routable 279 address space. 281 4. Overview 283 The Autonomic Control Plane is constructed in the following way (for 284 details, see Section 5): 286 o An autonomic node creates a virtual routing and forwarding (VRF) 287 instance, or a similar virtual context. 289 o It determines, following a policy, a candidate peer list. This is 290 the list of nodes to which it should establish an autonomic 291 control plane. Default policy is: To all adjacent nodes in the 292 same domain. Intent can override this default policy. 294 o For each node in the candidate peer list, it authenticates that 295 node and negotiates a mutually acceptable channel type. 297 o It then establishes a secure tunnel of the negotiated channel 298 type. These tunnels are placed into the previously set up VRF. 299 This creates an overlay network with hop-by-hop tunnels. 301 o Inside the ACP VRF, each node sets up a virtual interface with its 302 ULA IPv6 address. 304 o Each node runs a lightweight routing protocol, to announce 305 reachability of the virtual addresses inside the ACP. 307 o Non-autonomic NMS systems or controllers have to be manually 308 connected into the ACP. 310 o Connecting over non-autonomic Layer-3 clouds initially requires a 311 tunnel between autonomic nodes. 313 o None of the above operations (except manual ones) is reflected in 314 the configuration of the device. 316 The following figure illustrates the ACP. 318 autonomic node 1 autonomic node 2 319 ................... ................... 320 secure . . secure . . secure 321 tunnel : +-----------+ : tunnel : +-----------+ : tunnel 322 ..--------| ACP VRF |---------------------| ACP VRF |---------.. 323 : / \ / \ <--routing--> / \ / \ : 324 : \ / \ / \ / \ / : 325 ..--------| virtual |---------------------| virtual |---------.. 326 : | interface | : : | interface | : 327 : +-----------+ : : +-----------+ : 328 : : : : 329 : data plane :...............: data plane : 330 : : link : : 331 :.................: :.................: 333 Figure 1 335 The resulting overlay network is normally based exclusively on hop- 336 by-hop tunnels. This is because addressing used on links is IPv6 337 link local addressing, which does not require any prior set-up. This 338 way the ACP can be built even if there is no configuration on the 339 devices, or if the data plane has issues such as addressing or 340 routing problems. 342 5. Self-Creation of an Autonomic Control Plane 344 This section describes the steps to set up an Autonomic Control 345 Plane, and highlights the key properties which make it 346 "indestructible" against many inadvert changes to the data plane, for 347 example caused by misconfigurations. 349 5.1. Preconditions 351 An autonomic node can be a router, switch, controller, NMS host, or 352 any other IP device. We assume an autonomic node has a globally 353 unique domain certificate (LDevID), as well as an adjacency table. 355 5.1.1. Domain Certificate with ANIMA information 357 To establish an ACP securely, an Autnomic Node MUST have a globally 358 unique domain certificate (LDevID), with which it can 359 cryptographically assert its membership of the domain. The document 360 [I-D.ietf-anima-bootstrapping-keyinfra] describes how a domain 361 certificate can be automatically and securely derived from a vendor 362 specific Unique Device Identifier (UDI) or IDevID certificate. 363 (Note: the UDI used in this document is NOT the UUID specified in 364 [RFC4122].) 366 The domain certificate (LDevID) of an autonomic node MUST contain 367 ANIMA specific information, specifically the domain name, and its ACP 368 address with the zone-ID set to zero. This information MUST be 369 encoded in the LDevID in the subjectAltName / rfc822Name field in the 370 following way: 372 anima.acp+@ 374 An example: 376 anima.acp+FD99:B02D:8EC3:0:200:0:6400:1@example.com 378 The ACP address MUST be specified in its canonical form, as specified 379 in [RFC5952], to allow for easy textual comparisons. 381 The bootstrap process defined in 382 [I-D.ietf-anima-bootstrapping-keyinfra] MUST in an ANIMA network pass 383 on ACP address and domain to a new node, such that the new node can 384 add this to its enrolment request. 386 The Certificate Authority in an ANIMA network MUST honor these 387 parameters, and create the respective subjectAltName / rfc822Name in 388 the certificate. 390 ANIMA nodes can therefore find ACP address and domain using this 391 field in the domain certificate, both for themselves, as well as for 392 other nodes. 394 See section 4.2.1.6 of [RFC5280] for details on the subjectAltName 395 field. 397 5.1.2. AN Adjacency Table 399 To know to which nodes to establish an ACP channel, every autonomic 400 node maintains an adjacency table. The adjacency table contains 401 information about adjacent autonomic nodes, at a minimum: node-ID, IP 402 address, domain, certificate. An autonomic device MUST maintain this 403 adjacency table up to date. This table is used to determine to which 404 neighbor an ACP connection is established. 406 Where the next autonomic device is not directly adjacent, the 407 information in the adjacency table can be supplemented by 408 configuration. For example, the node-ID and IP address could be 409 configured. 411 The adjacency table MAY contain information about the validity and 412 trust of the adjacent autonomic node's certificate. However, 413 subsequent steps MUST always start with authenticating the peer. 415 The adjacency table contains information about adjacent autonomic 416 nodes in general, independently of their domain and trust status. 417 The next step determines to which of those autonomic nodes an ACP 418 connection should be established. 420 5.2. Neighbor discovery via mDNS/DNS-SD 422 Autonomic devices use DNS-SD/mDNS to discover the IPv6 link-local 423 address of autonomic neighbors across L2 adjacencies. The result is 424 stored in the AN Adjacency Table, see Section 5.1.2. 426 o If a node is not part an autonomic domain, it starts autonomic 427 enrollment with an adjacent node as the proxy using procedures 428 described in [I-D.ietf-anima-bootstrapping-keyinfra]. 430 o If a node is part of an autonomic domain (eg: enrolled with a 431 certificate) and has a working ACP towards one or more registrars 432 (including when it is a registrar herself), it MUST announce the 433 "_bootstrapks._tcp.local." service to indicate that it can act as 434 a proxy for bootstrap. This can be superceeded by Intent. 436 o If a node is part of an autonomic domain, it also announces the 437 "_anima_acp_ll._udp.local" service via DNS-based Service Discovery 438 [RFC6763] over Multicast DNS [RFC6762] and searches for that 439 services to discover neighbors. This can be superceeded by Intent 440 or future work (see below). 442 o TBD: Should the domain also be announced with the 443 "_anima_acp_ll._udp.local" service? Pro: Then a node "sees" which 444 domain a neighbor is and can make selection more efficient. Con: 445 A proxy exposes its domain by default. 447 o To prevent unaccceptable levels of network traffic the congestion 448 avoidance mechanisms specified in [RFC6762] section 7 MUST be 449 followed. Announcements for the "_anima_acp._udp.local" service 450 MUST only use the IPv6 link-local address of the announcing 451 interface. Other service parameters (SRV parameters, TXT records) 452 are ignored to avoid creating additional attack vectors from bogus 453 announcements by third parties. 455 o Note that two different service names are used between bootstrap 456 and ACP because these functions may not always go along with each 457 other. For example, an autonomic device connecting to the 458 Internet can still provide Bootstrap proxy functions via any of 459 the discovery mechanisms specified in the bootstrap document, 460 while building the ACP is only defined in this document to be 461 across IPv6 LL adjacencies. The reason for this difference is 462 that in case of non-L2-adjacent devices, it is appropriate to 463 perform bootstrap via a centralized bootstrap proxy, but the ACP 464 should be built towards the nearest possible ACP neighbors, and in 465 the absence of L2 adjacencies this is a more complex discovery 466 problem, left for future work - and different discovery 467 mechanisms. 469 o An autonomic node stores the results of the neighbor discovery in 470 the AN Adjacency table. 472 5.3. Candidate ACP Neighbor Selection 474 An autonomic node must determine to which other autonomic nodes in 475 the adjacency table it should build an ACP connection. This is based 476 on the information in the AN Adjacency table. 478 The ACP is by default established exclusively between nodes in the 479 same domain. 481 Intent can change this default behaviour. The precise format for 482 this Intent needs to be defined outside this document. Example 483 Intent policies are: 485 o The ACP should be built between all sub-domains for a given parent 486 domain. For example: For domain "example.com", nodes of 487 "example.com", "access.example.com", "core.example.com" and 488 "city.core.example.com" should all establish one single ACP. 490 o Two domains should build one single ACP between themselves, for 491 example "example1.com" should establish the ACP also with nodes 492 from "example2.com". For this case, the two domains must be able 493 to validate their trust, typically by cross-signing their 494 certificate infrastructure. 496 The result of the candidate ACP neighbor selection process is a list 497 of adjacent or configured autonomic neighbors to which an ACP channel 498 should be established. The next step begins that channel 499 establishment. 501 5.4. Channel Selection 503 To avoid attacks, initial discovery of candidate ACP peers can not 504 include any non-protected negotiation. To avoid re-inventing and 505 validating security association mechanisms, the next step after 506 discoving the address of a candidate neighbor can only be to try 507 first to establish a security association with that neighbor using a 508 well-known security association method. 510 At this time in the lifecycle of autonomic devices, it is unclear 511 whether it is feasible to even decide on a single MTI (mandatory to 512 implement) security association protocol across all autonomic 513 devices. 515 From the use-cases it is clear that not all type of autonomic devices 516 can or need to connect directly to each other or are able to support 517 or prefer all possible mechanisms. For example, code space limited 518 IoT devices may only support dTLS (because that code exists already 519 on them for end-to-end security use-cases), but low-end in-ceiling L2 520 switches may only want to support MacSec because that is also 521 supported in HW, and only a more flexible garteway device may need to 522 support both of these mechanisms and potentially more. 524 To support these requirements, and make ACP channel negotiation also 525 easily extensible, the secure channel selection between Alice and Bob 526 is a potentially two stage process. In the first stage, Alice and 527 Bob directly try to establish a secure channel using the security- 528 association and channel protocols they support. One or more of these 529 protocols may simply be protocols not directly resulting in an ACP 530 channel, but instead establishing a secure negotiation channel 531 through which the final secure channel protocol is decided. If both 532 Alice and Bob support such a negotiation step, then this secured 533 negotiation channel is the first step, and the secure channel 534 protocol is the second step. 536 If Alice supports multiple security association protocols in the 537 first step, it is a matter of Alices local policy to determine the 538 order in which she will try to build the connection to Bob. To 539 support multiple security association protocols, Alice must be able 540 to simultaneously act as a responder in parallel for all of them - so 541 that she can respond to any order in which Bob wants to prefer 542 building the security association. 544 When ACP setup between Alice and Bob results in the first secure 545 association to be established, the peer with the higher Device-ID in 546 the certificate will stop building new security associations. The 547 peer with the lower certificate Device-ID is now responsible to 548 continue building its most preferred security association and to tear 549 down all but that most preferred one - unless the secure association 550 is one of a negotation protocols whose rules superceed this. 552 All this negotiation is in the context of an "L2 interface". Alice 553 and Bob will build ACP connections to each other on every "L2 554 interface" that they both connect to. 556 5.5. Security Association protocols 558 The following sections define the security association protocols that 559 we consider to be important and feasible to specify in this document. 560 In all cases, the mutual authentication is done via the autonomic 561 domain certificate of the peer as follows - unless superceeded by 562 intent: 564 o The certificate is valid as proven by the security associations 565 protocol exchanges. 567 o The peers certificate is signed by the same CA as the devices 568 domain certificate. 570 o The peers OU (Organizational Unit) field in the certificates 571 Subject is the same as in the devices certificate. 573 5.5.1. ACP via IPsec 575 To run ACP via IPsec transport mode, no further IANA assignments/ 576 definitions are required. All autonomic devices suppoting IPsec MUST 577 support IPsec security setup via IKEv2, transport mode encapsulation 578 via the device and peer link-local IPv6 addresses and AES256 579 encryption. Further parameter options can be negotiated via IKEv2 or 580 via GRASP/TLS. 582 5.5.2. ACP via GRE/IPsec 584 In network devices it is often easier to provide virtual interfaces 585 on top of GRE encapsulation than natively on top of a simple IPsec 586 association. On those devices it may be necessary to run the ACP 587 secure channel on top of a GRE connection protected by the IPsec 588 association. The requirements for the IPsec association are the same 589 as described above, but instead of directly carrying the ACP IPv6 590 packets, the payload is an ACP IPv6 packet inside GRE/IPv6. 592 Note that without explicit negotiation (eg: via GRASP/TLS), this 593 method is incompatible to direct ACP via IPsec, so it must only be 594 used as an option during GRASP/TLS negotiation. 596 5.5.3. ACP via dTLS 598 To run ACP via UDP and dTLS v1.2 [RFC6347] an IANA assigned port 599 [TBD] is used. All autonomic devices supporting ACP via dTLS must 600 support AES256 encryption. 602 5.5.4. GRASP/TLS negotiation 604 To explicitly allow negotiation of the ACP channel protocol, GRASP 605 over a TLS connection using the GRASP_LISTEN_PORT and the devices and 606 peers link-local IPv6 address is used. When Alice and Bob support 607 GRASP negotiation, they do prefer it over any other non-explicitly 608 negotiated security association protocol and should wait trying any 609 non-negotiated ACP channel protocol until after it is clear that 610 GRASP/TLS will not work to the peer. 612 When Alice and Bob successfully establish the GRASP/TSL session, they 613 will initially negotiate the channel mechanism to use. Bob and Alice 614 each have a list of channel mehanisms they support, sorted by 615 preference. They negotiate the best mechansm supported by both of 616 them. In the absence of Intent, the mechanism choosen is the best 617 one for the device with the lower Device-ID. 619 After agreeing on a channel mechanism, Alice and Bob start the 620 selected Channel protocol. The GRASP/TLS connection can be kept 621 alive or timed out as long as the seelected channel protocol has a 622 secure association between Alice and Bob. When it terminates, it 623 needs to be re-negotiated via GRASP/TLS. 625 Negotiation of a channel type may require IANA assignments of code 626 points. See IANA Considerations (Section 12) for the formal 627 definition of those code points. 629 TBD: The exact negotiation steps in GRASP to achieve this outcome. 631 5.5.5. ACP Security Profiles 633 A baseline autonomic device MUST support IPsec and SHOULD support 634 GRASP/TLS and dTLS. A constrained autonomic device MUST support 635 dTLS. 637 Autonomic devices need to specify in documentation the set of secure 638 ACP mechanisms they suppport. 640 5.6. GRASP instance details 642 Received GRASP packets are assigned to an instance of GRASP by the 643 context they are received on: 645 o GRASP packets received on an ACP (virtual) interfaces are assigned 646 to the ACP instance of GRASP 648 o GRASP packets received inside a TLS connection established for 649 GRASP/TLS ACP negotiation are assigned to a separate instance of 650 GRASP for that negotiation. 652 TBD: The Details of the GRASP objective/packet formats still need to 653 be defined. Eg: Need to define an allocation for the objective of 654 "Autonomic Node". 656 5.7. Context Separation 658 The ACP is in a separate context from the normal data plane of the 659 device. This context includes the ACP channels IPv6 forwarding and 660 routing as well as any required higher layer ACP functions. 662 In classical network device platforms, a dedicated so called "Virtual 663 routing and forwarding instance" (VRF) is one logical implementation 664 option for the ACP. If possible by the platform SW architecture, 665 separation options that minimize shared components are preferred. 666 The context for the ACP needs to be established automatically during 667 bootstrap of a device. As much as possible it should be protected 668 from being modified unintentionally by data plane configuration. 670 Context separation improves security, because the ACP is not 671 reachable from the global routing table. Also, configuration errors 672 from the data plane setup do not affect the ACP. 674 5.8. Addressing inside the ACP 676 The channels explained above typically only establish communication 677 between two adjacent nodes. In order for communication to happen 678 across multiple hops, the autonomic control plane requires internal 679 network wide valid addresses and routing. Each autonomic node must 680 create a virtual interface with a network wide unique address inside 681 the ACP context mentioned in Section 5.7. This address may be used 682 also in other virtual contexts. 684 With the algorithm introduced here, all autonomic devices in the same 685 domain have the same /48 prefix. Conversely, global IDs from 686 different domains are unlikely to clash, such that two networks can 687 be merged, as long as the policy allows that merge. See also 688 Section 7 for a discussion on merging domains. 690 Links inside the ACP only use link-local IPv6 addressing, such that 691 each node only requires one routable virtual address. 693 5.8.1. Fundamental Concepts of Autonomic Addressing 695 o Usage: Autonomic addresses are exclusively used for self- 696 management functions inside a trusted domain. They are not used 697 for user traffic. Communications with entities outside the 698 trusted domain use another address space, for example normally 699 managed routable address space. 701 o Separation: Autonomic address space is used separately from user 702 address space and other address realms. This supports the 703 robustness requirement. 705 o Loopback-only: Only loopback interfaces of autonomic nodes carry a 706 routable address; all other interfaces exclusively use IPv6 link 707 local for autonomic functions. The usage of IPv6 link local 708 addressing is discussed in [RFC7404]. 710 o Use-ULA: For loopback interfaces of autonomic nodes, we use Unique 711 Local Addresses (ULA), as specified in [RFC4193]. An alternative 712 scheme was discussed, using assigned ULA addressing. The 713 consensus was to use standard ULA, because it was deemed to be 714 sufficient. 716 o No external connectivity: They do not provide access to the 717 Internet. If a node requires further reaching connectivity, it 718 should use another, traditionally managed address scheme in 719 parallel. 721 The ACP is based exclusively on IPv6 addressing, for a variety of 722 reasons: 724 o Simplicity, reliability and scale: If other network layer 725 protocols were supported, each would have to have its own set of 726 security associations, routing table and process, etc. 728 o Autonomic functions do not require IPv4: Autonomic functions and 729 autonomic service agents are new concepts. They can be 730 exclusively built on IPv6 from day one. There is no need for 731 backward compatibility. 733 o OAM protocols no not require IPv4: The ACP may carry OAM 734 protocols. All relevant protocols (SNMP, TFTP, SSH, SCP, Radius, 735 Diameter, ...) are available in IPv6. 737 5.8.2. The ACP Addressing Base Scheme 739 The Base ULA addressing scheme for autonomic nodes has the following 740 format: 742 8 40 3 77 743 +--+--------------+------+------------------------------------------+ 744 |FD| hash(domain) | Type | (sub-scheme) | 745 +--+--------------+------+------------------------------------------+ 747 Figure 2: ACP Addressing Base Scheme 749 The first 48 bits follow the ULA scheme, as defined in [RFC4193], to 750 which a type field is added: 752 o "FD" identifies a locally defined ULA address. 754 o The ULA "global ID" is set here to be a hash of the domain name, 755 which results in a pseudo-random 40 bit value. It is calculated 756 as the first 40 bits of the MD5 hash of the domain name, in the 757 example "example.com". 759 o Type: This field allows different address sub-schemes in the 760 future. The goal is to start with a single sub-schemes, but to 761 allow for extensions later if and when required. This addresses 762 the "upgradability" requirement. Assignment of types for this 763 field should be maintained by IANA. 765 5.8.3. ACP Addressing Sub-Scheme 767 The sub-scheme defined here is defined by the Type value 0 (zero) in 768 the base scheme. 770 51 13 63 1 771 +------------------------+---------+----------------------------+---+ 772 | (base scheme) | Zone ID | Device ID | V | 773 +------------------------+---------+----------------------------+---+ 775 Figure 3: ACP Addressing Sub-Scheme 777 The fields are defined as follows: [Editor's note: The lengths of the 778 fields is for discussion.] 780 o Zone ID: If set to all zero bits: The device ID bits are used as 781 an identifier (as opposed to a locator). This results in a non- 782 hierarchical, flat addressing scheme. Any other value indicates a 783 zone. See section Section 5.8.4 on how this field is used in 784 detail. 786 o Device ID: A unique value for each device. 788 o V: Virtualization bit: 0: autonomic node base system; 1: a virtual 789 context on an autonomic node. 791 The device ID is derived as follows: In an Autonomic Network, a 792 registrar is enrolling new devices. As part of the enrolment process 793 the registrar assigns a number to the device, which is unique for 794 this registrar, but not necessarily unique in the domain. The 64 bit 795 device ID is then composed as: 797 o 48 bit: Registrar ID, a number unique inside the domain that 798 identifies the registrar which assigned the name to the device. A 799 MAC address of the registrar can be used for this purpose. 801 o 15 bit: Device number, a number which is unique for a given 802 registrar, to identify the device. This can be a sequentially 803 assigned number. 805 The "device ID" itself is unique in a domain (i.e., the Zone-ID is 806 not required for uniqueness). Therefore, a device can be addressed 807 either as part of a flat hierarchy (zone ID = 0), or with an 808 aggregation scheme (any other zone ID). A address with zone-ID = 0 809 is an identifier, with another zone-ID as a locator. See 810 Section 5.8.4 for a description of the zone bits. 812 This addressing sub-scheme allows the direct addressing of specific 813 virtual containers / VMs on an autonomic node. An increasing number 814 of hardware platforms have a distributed architecture, with a base OS 815 for the node itself, and the support for hardware blades with 816 potentially different OSs. The VMs on the blades could be considered 817 as separate autonomic nodes, in which case it would make sense to be 818 able to address them directly. Autonomic Service Agents (ASAs) could 819 be instantiated in either the base OS, or one of the VMs on a blade. 820 This addressing scheme allows for the easy separation of the hardware 821 context. 823 The location of the V bit(s) at the end of the address allows to 824 announce a single prefix for each autonomic node, while having 825 separate virtual contexts addressable directly. 827 5.8.4. Usage of the Zone Field 829 The "zone ID" allows for the introduction of structure in the 830 addressing scheme. 832 Zone = zero is the default addressing scheme in an autonomic domain. 833 Every autonomic node MUST respond to its ACP address with zone=0. 834 Used on its own this leads to a non-hierarchical address scheme, 835 which is suitable for networks up to a certain size. In this case, 836 the addresses primarily act as identifiers for the nodes, and 837 aggregation is not possible. 839 If aggregation is required, the 13 bit value allows for up to 8191 840 zones. The allocation of zone numbers may either happen 841 automatically through a to-be-defined algorithm; or it could be 842 configured and maintained manually. [We could divide the zone space 843 into manual and automatic space - to be discussed.] 845 If a device learns through an autonomic method or through 846 configuration that it is part of a zone, it MUST also respond to its 847 ACP address with that zone number. In this case the ACP loopback is 848 configured with two ACP addresses: One for zone 0 and one for the 849 assigned zone. This method allows for a smooth transition between a 850 flat addressing scheme and an hierarchical one. 852 (Theoretically, the 13 bits for the zone ID would allow also for two 853 levels of zones, introducing a sub-hierarchy. We do not think this 854 is required at this point, but a new type could be used in the future 855 to support such a scheme.) 857 Note: Another way to introduce hierarchy is to use sub-domains in the 858 naming scheme. The node names "node17.subdomainA.example.com" and 859 "node4.subdomainB.example.com" would automatically lead to different 860 ULA prefixes, which can be used to introduce a routing hierarchy in 861 the network, assuming that the subdomains are aligned with routing 862 areas. 864 5.8.5. Other ACP Addressing Sub-Schemes 866 Other ACP addressing sub-schemes can be defined if and when required. 867 IANA will assign a new "type" for each new addressing sub-scheme. 869 5.9. Routing in the ACP 871 Once ULA address are set up all autonomic entities should run a 872 routing protocol within the autonomic control plane context. This 873 routing protocol distributes the ULA created in the previous section 874 for reachability. The use of the autonomic control plane specific 875 context eliminates the probable clash with the global routing table 876 and also secures the ACP from interference from the configuration 877 mismatch or incorrect routing updates. 879 The establishment of the routing plane and its parameters are 880 automatic and strictly within the confines of the autonomic control 881 plane. Therefore, no manual configuration is required. 883 All routing updates are automatically secured in transit as the 884 channels of the autonomic control plane are by default secured. 886 The routing protocol inside the ACP should be light weight and highly 887 scalable to ensure that the ACP does not become a limiting factor in 888 network scalability. We suggest the use of RPL [RFC6550] as one such 889 protocol which is light weight and scales well for the control plane 890 traffic. See Appendix A for more details on the choice of RPL. 892 5.10. General ACP Considerations 894 In order to be independent of configured link addresses, channels 895 SHOULD use IPv6 link local addresses between adjacent neighbors 896 wherever possible. This way, the ACP tunnels are independent of 897 correct network wide routing. 899 Since channels are by default established between adjacent neighbors, 900 the resulting overlay network does hop by hop encryption. Each node 901 decrypts incoming traffic from the ACP, and encrypts outgoing traffic 902 to its neighbors in the ACP. Routing is discussed in Section 5.9. 904 If two nodes are connected via several links, the ACP SHOULD be 905 established on every link, but it is possible to establish the ACP 906 only on a sub-set of links. Having an ACP channel on every link has 907 a number of advantages, for example it allows for a faster failover 908 in case of link failure, and it reflects the physical topology more 909 closely. Using a subset of links (for example, a single link), 910 reduces resource consumption on the devices, because state needs to 911 be kept per ACP channel. 913 6. Workarounds for Non-Autonomic Nodes 915 6.1. Connecting a Non-Autonomic Controller / NMS system 917 The Autonomic Control Plane can be used by management systems, such 918 as controllers or network management system (NMS) hosts (henceforth 919 called simply "NMS hosts"), to connect to devices through it. For 920 this, an NMS host must have access to the ACP. By default, the ACP 921 is a self-protecting overlay network, which only allows access to 922 trusted systems. Therefore, a traditional, non-autonomic NMS system 923 does not have access to the ACP by default, just like any other 924 external device. 926 If the NMS host is not autonomic, i.e., it does not support autonomic 927 negotiation of the ACP, then it can be brought into the ACP by 928 explicit configuration. On an adjacent autonomic node with ACP, the 929 interface with the NMS host can be configured to be part of the ACP. 930 In this case, the NMS host is with this interface entirely and 931 exclusively inside the ACP. It would likely require a second 932 interface for connections between the NMS host and administrators, or 933 Internet based services. This mode of connecting an NMS host has 934 security consequences: All systems and processes connected to this 935 implicitly trusted interface have access to all autonomic nodes on 936 the entire ACP, without further authentication. Thus, this 937 connection must be physically controlled. 939 The non-autonomic NMS host must be routed in the ACP. This involves 940 two parts: 1) the NMS host must point default to the AN device for 941 the ULA prefix used inside the ACP, and 2) the prefix used between AN 942 node and NMS host must be announced into the ACP, and distributed 943 there. 945 The document "Autonomic Network Stable Connectivity" 946 [I-D.eckert-anima-stable-connectivity] explains in more detail how 947 the ACP can be integrated in a mixed NOC environment. 949 6.2. ACP through Non-Autonomic L3 Clouds 951 Not all devices in a network may be autonomic. If non-autonomic 952 Layer-2 devices are between autonomic nodes, the communications 953 described in this document should work, since it is IP based. 954 However, non-autonomic Layer-3 devices do not forward link local 955 autonomic messages, and thus break the Autonomic Control Plane. 957 One workaround is to manually configure IP tunnels between autonomic 958 nodes across a non-autonomic Layer-3 cloud. The tunnels are 959 represented on each autonomic node as virtual interfaces, and all 960 autonomic transactions work across such tunnels. 962 Such manually configured tunnels are less "indestructible" than an 963 automatically created ACP based on link local addressing, since they 964 depend on correct data plane operations, such as routing and 965 addressing. 967 7. Self-Healing Properties 969 The ACP is self-healing: 971 o New neighbors will automatically join the ACP after successful 972 validation and will become reachable using their unique ULA 973 address across the ACP. 975 o When any changes happen in the topology, the routing protocol used 976 in the ACP will automatically adapt to the changes and will 977 continue to provide reachability to all devices. 979 o If an existing device gets revoked, it will automatically be 980 denied access to the ACP as its domain certificate will be 981 validated against a Certificate Revocation List during 982 authentication. Since the revocation check is only done at the 983 establishment of a new security association, existing ones are not 984 automatically torn down. If an immediate disconnect is required, 985 existing sessions to a freshly revoked device can be re-set. 987 The ACP can also sustain network partitions and mergers. Practically 988 all ACP operations are link local, where a network partition has no 989 impact. Devices authenticate each other using the domain 990 certificates to establish the ACP locally. Addressing inside the ACP 991 remains unchanged, and the routing protocol inside both parts of the 992 ACP will lead to two working (although partitioned) ACPs. 994 There are few central dependencies: A certificate revocation list 995 (CRL) may not be available during a network partition; a suitable 996 policy to not immediately disconnect neighbors when no CRL is 997 available can address this issue. Also, a registrar or Certificate 998 Authority might not be available during a partition. This may delay 999 renewal of certificates that are to expire in the future, and it may 1000 prevent the enrolment of new devices during the partition. 1002 After a network partition, a re-merge will just establish the 1003 previous status, certificates can be renewed, the CRL is available, 1004 and new devices can be enrolled everywhere. Since all devices use 1005 the same trust anchor, a re-merge will be smooth. 1007 Merging two networks with different trust anchors requires the trust 1008 anchors to mutually trust each other (for example, by cross-signing). 1009 As long as the domain names are different, the addressing will not 1010 overlap (see Section 5.8). 1012 8. Self-Protection Properties 1014 As explained in Section 5, the ACP is based on channels being built 1015 between devices which have been previously authenticated based on 1016 their domain certificates. The channels themselves are protected 1017 using standard encryption technologies like DTLS or IPsec which 1018 provide additional authentication during channel establishment, data 1019 integrity and data confidentiality protection of data inside the ACP 1020 and in addition, provide replay protection. 1022 An attacker will therefore not be able to join the ACP unless having 1023 a valid domain certificate, also packet injection and sniffing 1024 traffic will not be possible due to the security provided by the 1025 encryption protocol. 1027 The remaining attack vector would be to attack the underlying AN 1028 protocols themselves, either via directed attacks or by denial-of- 1029 service attacks. However, as the ACP is built using link-local IPv6 1030 address, remote attacks are impossible. The ULA addresses are only 1031 reachable inside the ACP context, therefore unreachable from the data 1032 plane. Also, the ACP protocols should be implemented to be attack 1033 resistant and not consume unnecessary resources even while under 1034 attack. 1036 9. The Administrator View 1038 An ACP is self-forming, self-managing and self-protecting, therefore 1039 has minimal dependencies on the administrator of the network. 1040 Specifically, since it is independent of configuration, there is no 1041 scope for configuration errors on the ACP itself. The administrator 1042 may have the option to enable or disable the entire approach, but 1043 detailed configuration is not possible. This means that the ACP must 1044 not be reflected in the running configuration of devices, except a 1045 possible on/off switch. 1047 While configuration is not possible, an administrator must have full 1048 visibility of the ACP and all its parameters, to be able to do 1049 trouble-shooting. Therefore, an ACP must support all show and debug 1050 options, as for any other network function. Specifically, a network 1051 management system or controller must be able to discover the ACP, and 1052 monitor its health. This visibility of ACP operations must clearly 1053 be separated from visibility of data plane so automated systems will 1054 never have to deal with ACP aspect unless they explicitly desire to 1055 do so. 1057 Since an ACP is self-protecting, a device not supporting the ACP, or 1058 without a valid domain certificate cannot connect to it. This means 1059 that by default a traditional controller or network management system 1060 cannot connect to an ACP. See Section 6.1 for more details on how to 1061 connect an NMS host into the ACP. 1063 10. Explanations 1065 This section is non-normative and intended to provide further 1066 explanations for the choices made in this document. 1068 10.1. Why GRASP to discover autonomic neighbors 1070 None of the considered mechanisms to establish security associations 1071 (eg: IPsec or dTLS) include mechanisms to discover candidate 1072 neighbors, so these security mechanisms themselves are insufficient 1073 for the discovery. 1075 Existing L2 mechanisms such as LLDP (or vendor speccific alternatives 1076 like Ciscos CDP) are L2 link-local. If an autonomic device connects 1077 via an LLDP capable, but non-autonomic capable L2 switch to another 1078 autonomic device, then the non-autonomic L2 switch would not 1079 propagate the LLDP messages, so discovery would not work as desired. 1081 Existing L3/L4 link local discovery mechanisms such as mDNS or Web- 1082 Services Discovery (http://specs.xmlsoap.org/ws/2005/04/discovery/ws- 1083 discovery.pdf) are capable to support the simple discovery required 1084 by autonomic devices but have the following downsides compared to 1085 GRASP. 1087 There is no clear single ubiquitoous protocol that would apply 1088 equally well to all market segments in which autonomic routers are 1089 intended to be deployed. Making a choice is therefore difficult. 1091 In some of these protocols, the fact that they operate L3 link local 1092 is often seen as a limitation rather than as a necessity for the 1093 application. 1095 Various mechanisms are used or considered in these protocols to 1096 expand the scope of discovery beyond a single L3 subnet. If 1097 autonomic devices would use such a protocol, then autonomic discovery 1098 messages could more likely leak into remote networks and give more 1099 undesired (insecured) visibility into the presence of autonomic 1100 devices and potentially leading to more attempts to establish 1101 autonomic associations with those discovered devices. To achieve the 1102 maximum resilience with the minimum number of ACP channels, those 1103 channels need to follow as closely the physcial hops in the topology 1104 as possible. 1106 Visibility of discovery protocols in other domains may be 1107 undesirable: Visibility of mDNS messages for example could extend all 1108 the way into end user application level service browsers. It is 1109 undesirable to see desvices announcing themselves as automic there. 1111 Existing protocols can be more complex compared to GRASP as they have 1112 been designed for different purposes, for example to be more flexible 1113 and generic. In mDNS, if DNS-SD was used, it would require at least 1114 four RRs to be exchanged for a single service: a PTR, a SRV, a TXT 1115 and a AAAA RR. Minimizing the number of protocol exchanges by 1116 coalescing these RRs is possible but requires additional software 1117 design considerations. 1119 GRASP is already required inside the ACP and a highly desirable 1120 option for secure ACP channel negotiation (GRASP/TLS). Using it for 1121 discovery allows to reuse that already necessary code basis. If any 1122 other protocol was used for discovery, then autonomic discovery might 1123 be the only purpose for which the protocol code exists in the device. 1125 None of the above arguments individually are strong reasons not to 1126 use one of these GRASP alternatives, but together they make it 1127 reasonable to first define GRASP as the MTI (Mandatory To Implement) 1128 for the discovery step. 1130 11. Security Considerations 1132 An ACP is self-protecting and there is no need to apply configuration 1133 to make it secure. Its security therefore does not depend on 1134 configuration. 1136 However, the security of the ACP depends on a number of other 1137 factors: 1139 o The usage of domain certificates depends on a valid supporting PKI 1140 infrastructure. If the chain of trust of this PKI infrastructure 1141 is compromised, the security of the ACP is also compromised. This 1142 is typically under the control of the network administrator. 1144 o Security can be compromised by implementation errors (bugs), as in 1145 all products. 1147 Fundamentally, security depends on correct operation, implementation 1148 and architecture. Autonomic approaches such as the ACP largely 1149 eliminate the dependency on correct operation; implementation and 1150 architectural mistakes are still possible, as in all networking 1151 technologies. 1153 12. IANA Considerations 1155 Section 5.5.3 describes ACP over dTLS, which requires a well-known 1156 UDP port. We request IANA to assign this UDP port for 'ACP over 1157 dTLS'. 1159 Section 5.5.4 describes an option for the channel negotiation, the 1160 'ACP channel type'. We request IANA to create a registry for 'ACP 1161 channel type'. 1163 The ACP channel type is a 8-bit unsigned integer. This document only 1164 assigns the first value. 1166 Number | Channel Type | RFC 1167 ---------+-----------------------------------+------------ 1168 0 | GRE tunnel protected with | this document 1169 | IPsec transport mode | 1170 1-255 | reserved for future channel types | 1172 Section 5.8.2 describes a 'type' field in the base addressing scheme. 1173 We request IANA to create a registry for the 'ACP addressing scheme 1174 type'. The initial value and definition will be determined in a 1175 later version of this document. 1177 13. Acknowledgements 1179 This work originated from an Autonomic Networking project at Cisco 1180 Systems, which started in early 2010. Many people contributed to 1181 this project and the idea of the Autonomic Control Plane, amongst 1182 which (in alphabetical order): Ignas Bagdonas, Parag Bhide, Balaji 1183 BL, Alex Clemm, Yves Hertoghs, Bruno Klauser, Max Pritikin, Ravi 1184 Kumar Vadapalli. 1186 Further input and suggestions were received from: Rene Struik, Brian 1187 Carpenter, Benoit Claise. 1189 14. Change log [RFC Editor: Please remove] 1190 14.1. Initial version 1192 First version of this document: draft-behringer-autonomic-control- 1193 plane 1195 14.2. draft-behringer-anima-autonomic-control-plane-00 1197 Initial version of the anima document; only minor edits. 1199 14.3. draft-behringer-anima-autonomic-control-plane-01 1201 o Clarified that the ACP should be based on, and support only IPv6. 1203 o Clarified in intro that ACP is for both, between devices, as well 1204 as for access from a central entity, such as an NMS. 1206 o Added a section on how to connect an NMS system. 1208 o Clarified the hop-by-hop crypto nature of the ACP. 1210 o Added several references to GDNP as a candidate protocol. 1212 o Added a discussion on network split and merge. Although, this 1213 should probably go into the certificate management story longer 1214 term. 1216 14.4. draft-behringer-anima-autonomic-control-plane-02 1218 Addresses (numerous) comments from Brian Carpenter. See mailing list 1219 for details. The most important changes are: 1221 o Introduced a new section "overview", to ease the understanding of 1222 the approach. 1224 o Merged the previous "problem statement" and "use case" sections 1225 into a mostly re-written "use cases" section, since they were 1226 overlapping. 1228 o Clarified the relationship with draft-eckert-anima-stable- 1229 connectivity 1231 14.5. draft-behringer-anima-autonomic-control-plane-03 1233 o Took out requirement for IPv6 --> that's in the reference doc. 1235 o Added requirement section. 1237 o Changed focus: more focus on autonomic functions, not only virtual 1238 out of band. This goes a bit throughout the document, starting 1239 with a changed abstract and intro. 1241 14.6. draft-ietf-anima-autonomic-control-plane-00 1243 No changes; re-submitted as WG document. 1245 14.7. draft-ietf-anima-autonomic-control-plane-01 1247 o Added some paragraphs in addressing section on "why IPv6 only", to 1248 reflect the discussion on the list. 1250 o Moved the data-plane ACP out of the main document, into an 1251 appendix. The focus is now the virtually separated ACP, since it 1252 has significant advantages, and isn't much harder to do. 1254 o Changed the self-creation algorithm: Part of the initial steps go 1255 into the reference document. This document now assumes an 1256 adjacency table, and domain certificate. How those get onto the 1257 device is outside scope for this document. 1259 o Created a new section 6 "workarounds for non-autonomic nodes", and 1260 put the previous controller section (5.9) into this new section. 1261 Now, section 5 is "autonomic only", and section 6 explains what to 1262 do with non-autonomic stuff. Much cleaner now. 1264 o Added an appendix explaining the choice of RPL as a routing 1265 protocol. 1267 o Formalised the creation process a bit more. Now, we create a 1268 "candidate peer list" from the adjacency table, and form the ACP 1269 with those candidates. Also it explains now better that policy 1270 (Intent) can influence the peer selection. (section 4 and 5) 1272 o Introduce a section for the capability negotiation protocol 1273 (section 7). This needs to be worked out in more detail. This 1274 will likely be based on GRASP. 1276 o Introduce a new parameter: ACP tunnel type. And defines it in the 1277 IANA considerations section. Suggest GRE protected with IPSec 1278 transport mode as the default tunnel type. 1280 o Updated links, lots of small edits. 1282 14.8. draft-ietf-anima-autonomic-control-plane-02 1284 o Added explicitly text for the ACP channel negotiation. 1286 o Merged draft-behringer-anima-autonomic-addressing-02 into this 1287 document, as suggested by WG chairs. 1289 14.9. draft-ietf-anima-autonomic-control-plane-03 1291 o Changed Neighbor discovery protocol from GRASP to mDNS. Bootstrap 1292 protocol team decided to go with mDNS to discover bootstrap proxy, 1293 and ACP should be consistent with this. Reasons to go with mDNS 1294 in bootstrap were a) Bootstrap should be reuseable also outside of 1295 full anima solutions and introduce as few as possible new 1296 elements. mDNS was considered well-known and very-likely even pre- 1297 existing in low-end devices (IoT). b) Using GRASP both for the 1298 insecure neighbor discovery and secure ACP operatations raises the 1299 risk of introducing security issues through implementation issues/ 1300 non-isolation between those two instances of GRASP. 1302 o Shortened the section on GRASP instances, because with mDNS being 1303 used for discovery, there is no insecure GRASP session any longer, 1304 simplifying the GRASP considerations. 1306 o Added certificate requirements for ANIMA in section 5.1.1, 1307 specifically how the ANIMA information is encoded in 1308 subjectAltName. 1310 o Deleted the appendix on "ACP without separation", as originally 1311 planned, and the paragraph in the main text referring to it. 1313 o Deleted one sub-addressing scheme, focusing on a single scheme 1314 now. 1316 o Included information on how ANIMA information must be encoded in 1317 the domain certificate in Section 5.1. 1319 o Editorial changes, updated draft references, etc. 1321 15. References 1323 [I-D.eckert-anima-stable-connectivity] 1324 Eckert, T. and M. Behringer, "Using Autonomic Control 1325 Plane for Stable Connectivity of Network OAM", draft- 1326 eckert-anima-stable-connectivity-02 (work in progress), 1327 October 2015. 1329 [I-D.ietf-anima-bootstrapping-keyinfra] 1330 Pritikin, M., Richardson, M., Behringer, M., and S. 1331 Bjarnason, "Bootstrapping Remote Secure Key 1332 Infrastructures (BRSKI)", draft-ietf-anima-bootstrapping- 1333 keyinfra-03 (work in progress), June 2016. 1335 [I-D.ietf-anima-grasp] 1336 Bormann, D., Carpenter, B., and B. Liu, "A Generic 1337 Autonomic Signaling Protocol (GRASP)", draft-ietf-anima- 1338 grasp-06 (work in progress), June 2016. 1340 [I-D.ietf-anima-reference-model] 1341 Behringer, M., Carpenter, B., Eckert, T., Ciavaglia, L., 1342 Pierre, P., Liu, B., Nobre, J., and J. Strassner, "A 1343 Reference Model for Autonomic Networking", draft-ietf- 1344 anima-reference-model-02 (work in progress), July 2016. 1346 [RFC4122] Leach, P., Mealling, M., and R. Salz, "A Universally 1347 Unique IDentifier (UUID) URN Namespace", RFC 4122, 1348 DOI 10.17487/RFC4122, July 2005, 1349 . 1351 [RFC4193] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 1352 Addresses", RFC 4193, DOI 10.17487/RFC4193, October 2005, 1353 . 1355 [RFC5082] Gill, V., Heasley, J., Meyer, D., Savola, P., Ed., and C. 1356 Pignataro, "The Generalized TTL Security Mechanism 1357 (GTSM)", RFC 5082, DOI 10.17487/RFC5082, October 2007, 1358 . 1360 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 1361 Housley, R., and W. Polk, "Internet X.509 Public Key 1362 Infrastructure Certificate and Certificate Revocation List 1363 (CRL) Profile", RFC 5280, DOI 10.17487/RFC5280, May 2008, 1364 . 1366 [RFC5952] Kawamura, S. and M. Kawashima, "A Recommendation for IPv6 1367 Address Text Representation", RFC 5952, 1368 DOI 10.17487/RFC5952, August 2010, 1369 . 1371 [RFC6347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1372 Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347, 1373 January 2012, . 1375 [RFC6550] Winter, T., Ed., Thubert, P., Ed., Brandt, A., Hui, J., 1376 Kelsey, R., Levis, P., Pister, K., Struik, R., Vasseur, 1377 JP., and R. Alexander, "RPL: IPv6 Routing Protocol for 1378 Low-Power and Lossy Networks", RFC 6550, 1379 DOI 10.17487/RFC6550, March 2012, 1380 . 1382 [RFC6762] Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762, 1383 DOI 10.17487/RFC6762, February 2013, 1384 . 1386 [RFC6763] Cheshire, S. and M. Krochmal, "DNS-Based Service 1387 Discovery", RFC 6763, DOI 10.17487/RFC6763, February 2013, 1388 . 1390 [RFC7404] Behringer, M. and E. Vyncke, "Using Only Link-Local 1391 Addressing inside an IPv6 Network", RFC 7404, 1392 DOI 10.17487/RFC7404, November 2014, 1393 . 1395 [RFC7575] Behringer, M., Pritikin, M., Bjarnason, S., Clemm, A., 1396 Carpenter, B., Jiang, S., and L. Ciavaglia, "Autonomic 1397 Networking: Definitions and Design Goals", RFC 7575, 1398 DOI 10.17487/RFC7575, June 2015, 1399 . 1401 [RFC7576] Jiang, S., Carpenter, B., and M. Behringer, "General Gap 1402 Analysis for Autonomic Networking", RFC 7576, 1403 DOI 10.17487/RFC7576, June 2015, 1404 . 1406 Appendix A. Background on the choice of routing protocol 1408 In a pre-standard implementation, the "IPv6 Routing Protocol for Low- 1409 Power and Lossy Networks (RPL, [RFC6550] was chosen. This 1410 Appendix explains the reasoning behind that decision. 1412 Requirements for routing in the ACP are: 1414 o Self-management: The ACP must build automatically, without human 1415 intervention. Therefore routing protocol must also work 1416 completely automatically. RPL is a simple, self-managing 1417 protocol, which does not require zones or areas; it is also self- 1418 configuring, since configuration is carried as part of the 1419 protocol (see Section 6.7.6 of [RFC6550]). 1421 o Scale: The ACP builds over an entire domain, which could be a 1422 large enterprise or service provider network. The routing 1423 protocol must therefore support domains of 100,000 nodes or more, 1424 ideally without the need for zoning or separation into areas. RPL 1425 has this scale property. This is based on extensive use of 1426 default routing. RPL also has other scalability improvements, 1427 such as selecting only a subset of peers instead of all possible 1428 ones, and trickle support for information synchronisation. 1430 o Low resource consumption: The ACP supports traditional network 1431 infrastructure, thus runs in addition to traditional protocols. 1432 The ACP, and specifically the routing protocol must have low 1433 resource consumption both in terms of memory and CPU requirements. 1434 Specifically, at edge nodes, where memory and CPU are scarce, 1435 consumption should be minimal. RPL builds a destination-oriented 1436 directed acyclic graph (DODAG), where the main resource 1437 consumption is at the root of the DODAG. The closer to the edge 1438 of the network, the less state needs to be maintained. This 1439 adapts nicely to the typical network design. Also, all changes 1440 below a common parent node are kept below that parent node. 1442 o Support for unstructured address space: In the Autonomic 1443 Networking Infrastructure, node addresses are identifiers, and may 1444 not be assigned in a topological way. Also, nodes may move 1445 topologically, without changing their address. Therefore, the 1446 routing protocol must support completely unstructured address 1447 space. RPL is specifically made for mobile ad-hoc networks, with 1448 no assumptions on topologically aligned addressing. 1450 o Modularity: To keep the initial implementation small, yet allow 1451 later for more complex methods, it is highly desirable that the 1452 routing protocol has a simple base functionality, but can import 1453 new functional modules if needed. RPL has this property with the 1454 concept of "objective function", which is a plugin to modify 1455 routing behaviour. 1457 o Extensibility: Since the Autonomic Networking Infrastructure is a 1458 new concept, it is likely that changes in the way of operation 1459 will happen over time. RPL allows for new objective functions to 1460 be introduced later, which allow changes to the way the routing 1461 protocol creates the DAGs. 1463 o Multi-topology support: It may become necessary in the future to 1464 support more than one DODAG for different purposes, using 1465 different objective functions. RPL allow for the creation of 1466 several parallel DODAGs, should this be required. This could be 1467 used to create different topologies to reach different roots. 1469 o No need for path optimisation: RPL does not necessarily compute 1470 the optimal path between any two nodes. However, the ACP does not 1471 require this today, since it carries mainly non-delay-sensitive 1472 feedback loops. It is possible that different optimisation 1473 schemes become necessary in the future, but RPL can be expanded 1474 (see point "Extensibility" above). 1476 Authors' Addresses 1478 Michael H. Behringer (editor) 1479 Cisco Systems 1480 Building D, 45 Allee des Ormes 1481 Mougins 06250 1482 France 1484 Email: mbehring@cisco.com 1486 Toerless Eckert 1487 Cisco Systems 1489 Email: eckert@cisco.com 1491 Steinthor Bjarnason 1492 Arbor Networks 1493 2727 South State Street, Suite 200 1494 Ann Arbor MI 48104 1495 United States 1497 Email: sbjarnason@arbor.net