idnits 2.17.1 draft-ietf-anima-autonomic-control-plane-05.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 256: '... ACP1: The ACP SHOULD provide robust...' RFC 2119 keyword, line 261: '... ACP2: The ACP MUST have a separate ...' RFC 2119 keyword, line 265: '... ACP3: The ACP MUST use autonomicall...' RFC 2119 keyword, line 270: '... ACP4: The ACP MUST be generic. Usa...' RFC 2119 keyword, line 271: '...rastructure. It MUST NOT be tied to a...' (18 more instances...) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 11, 2017) is 2652 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 682, but not defined == Unused Reference: 'RFC4122' is defined on line 1443, but no explicit reference was found in the text == Unused Reference: 'RFC5082' is defined on line 1457, but no explicit reference was found in the text == Unused Reference: 'RFC6762' is defined on line 1484, but no explicit reference was found in the text == Unused Reference: 'RFC6763' is defined on line 1488, but no explicit reference was found in the text == Outdated reference: A later version (-03) exists of draft-carpenter-anima-ani-objectives-00 ** Downref: Normative reference to an Informational draft: draft-carpenter-anima-ani-objectives (ref. 'I-D.carpenter-anima-ani-objectives') == Outdated reference: A later version (-45) exists of draft-ietf-anima-bootstrapping-keyinfra-04 == Outdated reference: A later version (-15) exists of draft-ietf-anima-grasp-09 == 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') == Outdated reference: A later version (-10) exists of draft-ietf-anima-stable-connectivity-01 ** Downref: Normative reference to an Informational draft: draft-ietf-anima-stable-connectivity (ref. 'I-D.ietf-anima-stable-connectivity') ** Downref: Normative reference to an Informational draft: draft-richardson-anima-6join-discovery (ref. 'I-D.richardson-anima-6join-discovery') ** Obsolete normative reference: RFC 4941 (Obsoleted by RFC 8981) ** 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: 11 errors (**), 0 flaws (~~), 11 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 Cisco Systems 4 Intended status: Standards Track T. Eckert 5 Expires: July 15, 2017 6 S. Bjarnason 7 Arbor Networks 8 January 11, 2017 10 An Autonomic Control Plane 11 draft-ietf-anima-autonomic-control-plane-05 13 Abstract 15 Autonomic functions need a control plane to communicate, which 16 depends on some addressing and routing. This Autonomic Control Plane 17 should ideally be self-managing, and as independent as possible of 18 configuration. This document defines an "Autonomic Control Plane", 19 with the primary use as a control plane for autonomic functions. It 20 also serves as a "virtual out of band channel" for OAM communications 21 over a network that is not configured, or mis-configured. 23 Status of This Memo 25 This Internet-Draft is submitted in full conformance with the 26 provisions of BCP 78 and BCP 79. 28 Internet-Drafts are working documents of the Internet Engineering 29 Task Force (IETF). Note that other groups may also distribute 30 working documents as Internet-Drafts. The list of current Internet- 31 Drafts is at http://datatracker.ietf.org/drafts/current/. 33 Internet-Drafts are draft documents valid for a maximum of six months 34 and may be updated, replaced, or obsoleted by other documents at any 35 time. It is inappropriate to use Internet-Drafts as reference 36 material or to cite them other than as "work in progress." 38 This Internet-Draft will expire on July 15, 2017. 40 Copyright Notice 42 Copyright (c) 2017 IETF Trust and the persons identified as the 43 document authors. All rights reserved. 45 This document is subject to BCP 78 and the IETF Trust's Legal 46 Provisions Relating to IETF Documents 47 (http://trustee.ietf.org/license-info) in effect on the date of 48 publication of this document. Please review these documents 49 carefully, as they describe your rights and restrictions with respect 50 to this document. Code Components extracted from this document must 51 include Simplified BSD License text as described in Section 4.e of 52 the Trust Legal Provisions and are provided without warranty as 53 described in the Simplified BSD License. 55 Table of Contents 57 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 58 2. Use Cases for an Autonomic Control Plane . . . . . . . . . . 4 59 2.1. An Infrastructure for Autonomic Functions . . . . . . . . 4 60 2.2. Secure Bootstrap over an Unconfigured Network . . . . . . 5 61 2.3. Data Plane Independent Permanent Reachability . . . . . . 5 62 3. Requirements . . . . . . . . . . . . . . . . . . . . . . . . 6 63 4. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . 7 64 5. Self-Creation of an Autonomic Control Plane . . . . . . . . . 8 65 5.1. Preconditions . . . . . . . . . . . . . . . . . . . . . . 8 66 5.1.1. Domain Certificate with ANIMA information . . . . . . 8 67 5.1.2. AN Adjacency Table . . . . . . . . . . . . . . . . . 9 68 5.2. Neighbor discovery . . . . . . . . . . . . . . . . . . . 10 69 5.2.1. L2 topology considerations . . . . . . . . . . . . . 10 70 5.2.2. CDP/LLDP/mDNS considerations . . . . . . . . . . . . 11 71 5.2.3. Discovery with GRASP . . . . . . . . . . . . . . . . 11 72 5.2.4. Discovery and BRSKY . . . . . . . . . . . . . . . . . 12 73 5.3. Candidate ACP Neighbor Selection . . . . . . . . . . . . 12 74 5.4. Channel Selection . . . . . . . . . . . . . . . . . . . . 13 75 5.5. Security Association protocols . . . . . . . . . . . . . 14 76 5.5.1. ACP via IPsec . . . . . . . . . . . . . . . . . . . . 15 77 5.5.2. ACP via GRE/IPsec . . . . . . . . . . . . . . . . . . 15 78 5.5.3. ACP via dTLS . . . . . . . . . . . . . . . . . . . . 15 79 5.5.4. GRASP/TLS negotiation . . . . . . . . . . . . . . . . 15 80 5.5.5. ACP Security Profiles . . . . . . . . . . . . . . . . 16 81 5.6. GRASP instance details . . . . . . . . . . . . . . . . . 16 82 5.7. Context Separation . . . . . . . . . . . . . . . . . . . 16 83 5.8. Addressing inside the ACP . . . . . . . . . . . . . . . . 17 84 5.8.1. Fundamental Concepts of Autonomic Addressing . . . . 17 85 5.8.2. The ACP Addressing Base Scheme . . . . . . . . . . . 18 86 5.8.3. ACP Addressing Sub-Scheme . . . . . . . . . . . . . . 19 87 5.8.4. Usage of the Zone Field . . . . . . . . . . . . . . . 20 88 5.8.5. Other ACP Addressing Sub-Schemes . . . . . . . . . . 21 89 5.9. Routing in the ACP . . . . . . . . . . . . . . . . . . . 21 90 5.10. General ACP Considerations . . . . . . . . . . . . . . . 21 91 6. Workarounds for Non-Autonomic Nodes . . . . . . . . . . . . . 22 92 6.1. Connecting a Non-Autonomic Controller / NMS system . . . 22 93 6.2. ACP through Non-Autonomic L3 Clouds . . . . . . . . . . . 23 94 7. Self-Healing Properties . . . . . . . . . . . . . . . . . . . 23 95 8. Self-Protection Properties . . . . . . . . . . . . . . . . . 24 96 9. The Administrator View . . . . . . . . . . . . . . . . . . . 25 97 10. Security Considerations . . . . . . . . . . . . . . . . . . . 25 98 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26 99 12. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 26 100 13. Change log [RFC Editor: Please remove] . . . . . . . . . . . 27 101 13.1. Initial version . . . . . . . . . . . . . . . . . . . . 27 102 13.2. draft-behringer-anima-autonomic-control-plane-00 . . . . 27 103 13.3. draft-behringer-anima-autonomic-control-plane-01 . . . . 27 104 13.4. draft-behringer-anima-autonomic-control-plane-02 . . . . 27 105 13.5. draft-behringer-anima-autonomic-control-plane-03 . . . . 27 106 13.6. draft-ietf-anima-autonomic-control-plane-00 . . . . . . 28 107 13.7. draft-ietf-anima-autonomic-control-plane-01 . . . . . . 28 108 13.8. draft-ietf-anima-autonomic-control-plane-02 . . . . . . 29 109 13.9. draft-ietf-anima-autonomic-control-plane-03 . . . . . . 29 110 13.10. draft-ietf-anima-autonomic-control-plane-04 . . . . . . 29 111 13.11. draft-ietf-anima-autonomic-control-plane-05 . . . . . . 30 112 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 30 113 Appendix A. Background on the choice of routing protocol . . . . 32 114 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 34 116 1. Introduction 118 Autonomic Networking is a concept of self-management: Autonomic 119 functions self-configure, and negotiate parameters and settings 120 across the network. [RFC7575] defines the fundamental ideas and 121 design goals of Autonomic Networking. A gap analysis of Autonomic 122 Networking is given in [RFC7576]. The reference architecture for 123 Autonomic Networking in the IETF is currently being defined in the 124 document [I-D.ietf-anima-reference-model] 126 Autonomic functions need a stable and robust infrastructure to 127 communicate on. This infrastructure should be as robust as possible, 128 and it should be re-usable by all autonomic functions. [RFC7575] 129 calls it the "Autonomic Control Plane". This document defines the 130 Autonomic Control Plane. 132 Today, the management and control plane of networks typically runs in 133 the global routing table, which is dependent on correct configuration 134 and routing. Misconfigurations or routing problems can therefore 135 disrupt management and control channels. Traditionally, an out of 136 band network has been used to recover from such problems, or 137 personnel is sent on site to access devices through console ports 138 (craft ports). However, both options are operationally expensive. 140 In increasingly automated networks either controllers or distributed 141 autonomic service agents in the network require a control plane which 142 is independent of the network they manage, to avoid impacting their 143 own operations. 145 This document describes options for a self-forming, self-managing and 146 self-protecting "Autonomic Control Plane" (ACP) which is inband on 147 the network, yet as independent as possible of configuration, 148 addressing and routing problems (for details how this achieved, see 149 Section 5). It therefore remains operational even in the presence of 150 configuration errors, addressing or routing issues, or where policy 151 could inadvertently affect control plane connectivity. The Autonomic 152 Control Plane serves several purposes at the same time: 154 o Autonomic functions communicate over the ACP. The ACP therefore 155 supports directly Autonomic Networking functions, as described in 156 [I-D.ietf-anima-reference-model]. For example, GRASP 157 [I-D.ietf-anima-grasp] can run securely inside the ACP. 159 o An operator can use it to log into remote devices, even if the 160 data plane is misconfigured or unconfigured. 162 o A controller or network management system can use it to securely 163 bootstrap network devices in remote locations, even if the network 164 in between is not yet configured; no data-plane dependent 165 bootstrap configuration is required. An example of such a secure 166 bootstrap process is described in 167 [I-D.ietf-anima-bootstrapping-keyinfra] 169 This document describes some use cases for the ACP in Section 2, it 170 defines the requirements in Section 3, Section 4 gives an overview 171 how an Autonomic Control Plane is constructed, and in Section 5 the 172 detailed process is explained. Section 6 explains how non-autonomic 173 nodes and networks can be integrated, and Section 5.5 the first 174 channel types for the ACP. 176 The document "Autonomic Network Stable Connectivity" 177 [I-D.ietf-anima-stable-connectivity] describes how the ACP can be 178 used to provide stable connectivity for OAM applications. It also 179 explains on how existing management solutions can leverage the ACP in 180 parallel with traditional management models, when to use the ACP 181 versus the data plane, how to integrate IPv4 based management, etc. 183 2. Use Cases for an Autonomic Control Plane 185 2.1. An Infrastructure for Autonomic Functions 187 Autonomic Functions need a stable infrastructure to run on, and all 188 autonomic functions should use the same infrastructure to minimise 189 the complexity of the network. This way, there is only need for a 190 single discovery mechanism, a single security mechanism, and other 191 processes that distributed functions require. 193 2.2. Secure Bootstrap over an Unconfigured Network 195 Today, bootstrapping a new device typically requires all devices 196 between a controlling node (such as an SDN controller) and the new 197 device to be completely and correctly addressed, configured and 198 secured. Therefore, bootstrapping a network happens in layers around 199 the controller. Without console access (for example through an out 200 of band network) it is not possible today to make devices securely 201 reachable before having configured the entire network between. 203 With the ACP, secure bootstrap of new devices can happen without 204 requiring any configuration on the network. A new device can 205 automatically be bootstrapped in a secure fashion and be deployed 206 with a domain certificate. This does not require any configuration 207 on intermediate nodes, because they can communicate through the ACP. 209 2.3. Data Plane Independent Permanent Reachability 211 Today, most critical control plane protocols and network management 212 protocols are running in the data plane (global routing table) of the 213 network. This leads to undesirable dependencies between control and 214 management plane on one side and the data plane on the other: Only if 215 the data plane is operational, will the other planes work as 216 expected. 218 Data plane connectivity can be affected by errors and faults, for 219 example certain AAA misconfigurations can lock an administrator out 220 of a device; routing or addressing issues can make a device 221 unreachable; shutting down interfaces over which a current management 222 session is running can lock an admin irreversibly out of the device. 223 Traditionally only console access can help recover from such issues. 225 Data plane dependencies also affect NOC/SDN controller applications: 226 Certain network changes are today hard to operate, because the change 227 itself may affect reachability of the devices. Examples are address 228 or mask changes, routing changes, or security policies. Today such 229 changes require precise hop-by-hop planning. 231 The ACP provides reachability that is largely independent of the data 232 plane, which allows control plane and management plane to operate 233 more robustly: 235 o For management plane protocols, the ACP provides the functionality 236 of a "Virtual-out-of-band (VooB) channel", by providing 237 connectivity to all devices regardless of their configuration or 238 global routing table. 240 o For control plane protocols, the ACP allows their operation even 241 when the data plane is temporarily faulty, or during transitional 242 events, such as routing changes, which may affect the control 243 plane at least temporarily. This is specifically important for 244 autonomic service agents, which could affect data plane 245 connectivity. 247 The document "Autonomic Network Stable Connectivity" 248 [I-D.ietf-anima-stable-connectivity] explains the use cases for the 249 ACP in significantly more detail and explains how the ACP can be used 250 in practical network operations. 252 3. Requirements 254 The Autonomic Control Plane has the following requirements: 256 ACP1: The ACP SHOULD provide robust connectivity: As far as 257 possible, it should be independent of configured addressing, 258 configuration and routing. Requirements 2 and 3 build on this 259 requirement, but also have value on their own. 261 ACP2: The ACP MUST have a separate address space from the data 262 plane. Reason: traceability, debug-ability, separation from 263 data plane, security (can block easily at edge). 265 ACP3: The ACP MUST use autonomically managed address space. Reason: 266 easy bootstrap and setup ("autonomic"); robustness (admin 267 can't mess things up so easily). This document suggests to 268 use ULA addressing for this purpose. 270 ACP4: The ACP MUST be generic. Usable by all the functions and 271 protocols of the AN infrastructure. It MUST NOT be tied to a 272 particular protocol. 274 ACP5: The ACP MUST provide security: Messages coming through the ACP 275 MUST be authenticated to be from a trusted node, and SHOULD 276 (very strong SHOULD) be encrypted. 278 The default mode of operation of the ACP is hop-by-hop, because this 279 interaction can be built on IPv6 link local addressing, which is 280 autonomic, and has no dependency on configuration (requirement 1). 281 It may be necessary to have ACP connectivity over non-autonomic 282 nodes, for example to link autonomic nodes over the general Internet. 283 This is possible, but then has a dependency on routing over the non- 284 autonomic hops. 286 4. Overview 288 The Autonomic Control Plane is constructed in the following way (for 289 details, see Section 5): 291 1. An autonomic node creates a virtual routing and forwarding (VRF) 292 instance, or a similar virtual context. 294 2. It determines, following a policy, a candidate peer list. This 295 is the list of nodes to which it should establish an Autonomic 296 Control Plane. Default policy is: To all adjacent nodes in the 297 same domain. 299 3. For each node in the candidate peer list, it authenticates that 300 node and negotiates a mutually acceptable channel type. 302 4. It then establishes a secure tunnel of the negotiated channel 303 type. These tunnels are placed into the previously set up VRF. 304 This creates an overlay network with hop-by-hop tunnels. 306 5. Inside the ACP VRF, each node sets up a virtual interface with 307 its ULA IPv6 address. 309 6. Each node runs a lightweight routing protocol, to announce 310 reachability of the virtual addresses inside the ACP. 312 Note: 314 o Non-autonomic NMS systems or controllers have to be manually 315 connected into the ACP. 317 o Connecting over non-autonomic Layer-3 clouds initially requires a 318 tunnel between autonomic nodes. 320 o None of the above operations (except manual ones) is reflected in 321 the configuration of the device. 323 The following figure illustrates the ACP. 325 autonomic node 1 autonomic node 2 326 ................... ................... 327 secure . . secure . . secure 328 tunnel : +-----------+ : tunnel : +-----------+ : tunnel 329 ..--------| ACP VRF |---------------------| ACP VRF |---------.. 330 : / \ / \ <--routing--> / \ / \ : 331 : \ / \ / \ / \ / : 332 ..--------| virtual |---------------------| virtual |---------.. 333 : | interface | : : | interface | : 334 : +-----------+ : : +-----------+ : 335 : : : : 336 : data plane :...............: data plane : 337 : : link : : 338 :.................: :.................: 340 Figure 1 342 The resulting overlay network is normally based exclusively on hop- 343 by-hop tunnels. This is because addressing used on links is IPv6 344 link local addressing, which does not require any prior set-up. This 345 way the ACP can be built even if there is no configuration on the 346 devices, or if the data plane has issues such as addressing or 347 routing problems. 349 5. Self-Creation of an Autonomic Control Plane 351 This section describes the steps to set up an Autonomic Control 352 Plane, and highlights the key properties which make it 353 "indestructible" against many inadvert changes to the data plane, for 354 example caused by misconfigurations. 356 5.1. Preconditions 358 An autonomic node can be a router, switch, controller, NMS host, or 359 any other IP device. We assume an autonomic node has a globally 360 unique domain certificate (LDevID), as well as an adjacency table. 362 5.1.1. Domain Certificate with ANIMA information 364 To establish an ACP securely, an Autnomic Node MUST have a globally 365 unique domain certificate (LDevID), with which it can 366 cryptographically assert its membership of the domain. The document 367 [I-D.ietf-anima-bootstrapping-keyinfra] describes how a domain 368 certificate can be automatically and securely derived from a vendor 369 specific Unique Device Identifier (UDI) or IDevID certificate. 371 The domain certificate (LDevID) of an autonomic node MUST contain 372 ANIMA specific information, specifically the domain name, and its ACP 373 address with the zone-ID set to zero. This information MUST be 374 encoded in the LDevID in the subjectAltName / rfc822Name field in the 375 following way: 377 anima.acp+@ 379 An example: 381 anima.acp+FD99:B02D:8EC3:0:200:0:6400:1@example.com 383 The ACP address MUST be specified in its canonical form, as specified 384 in [RFC5952], to allow for easy textual comparisons. 386 The bootstrap process defined in 387 [I-D.ietf-anima-bootstrapping-keyinfra] MUST in an ANIMA network pass 388 on ACP address and domain to a new node, such that the new node can 389 add this to its enrolment request. 391 The Certificate Authority in an ANIMA network MUST honor these 392 parameters, and create the respective subjectAltName / rfc822Name in 393 the certificate. 395 ANIMA nodes can therefore find ACP address and domain using this 396 field in the domain certificate, both for themselves, as well as for 397 other nodes. 399 See section 4.2.1.6 of [RFC5280] for details on the subjectAltName 400 field. 402 5.1.2. AN Adjacency Table 404 To know to which nodes to establish an ACP channel, every autonomic 405 node maintains an adjacency table. The adjacency table contains 406 information about adjacent autonomic nodes, at a minimum: node-ID, IP 407 address, domain, certificate. An autonomic device MUST maintain this 408 adjacency table up to date. This table is used to determine to which 409 neighbor an ACP connection is established. 411 Where the next autonomic device is not directly adjacent, the 412 information in the adjacency table can be supplemented by 413 configuration. For example, the node-ID and IP address could be 414 configured. 416 The adjacency table MAY contain information about the validity and 417 trust of the adjacent autonomic node's certificate. However, 418 subsequent steps MUST always start with authenticating the peer. 420 The adjacency table contains information about adjacent autonomic 421 nodes in general, independently of their domain and trust status. 422 The next step determines to which of those autonomic nodes an ACP 423 connection should be established. 425 5.2. Neighbor discovery 427 5.2.1. L2 topology considerations 429 ANrtr1 ------ ANswitch1 --- ANswitch2 ------- ANrtr2 430 .../ \ \ ... 431 ANrtrM ------ \ ------- ANrtrN 432 ANswitchM ... 434 Figure 2 436 Consider a large L2 LAN with ANrtr1...ANrtrN connected via some 437 topology of L2 switches (eg: in a large enterprise campus or IoT 438 environment using large L2 LANs). If the discovery protocol used for 439 the ACP is operating at the subnet level, every AN router will see 440 all other AN routers on the LAN as neighbors and a full mesh of ACP 441 channels will be built. If some or all of the AN switches are 442 autonomic with the same discovery protocol, then the full mesh would 443 include those switches as well. 445 A full mesh of ACP connections like this can creates fundamental 446 challenges. The number of security associations of the secure 447 channel protocols will not scale arbitrarily, especially when they 448 leverage platform accelerated encryption/decryption. Likewise, any 449 other ACP operations needs to scale to the number of direct ACP 450 neigbors. An AN router with just 4 interfaces might be deployed into 451 a LAN with hundreds of neighbors connected via switches. Introducing 452 such a new unpredictable scaling factor requirement makes it harder 453 to support the ACP on arbitrary platforms and in arbitrary 454 deployments. 456 Predictable scaling requirements for ACP neighbors can most easily be 457 achieved if in topologies like these, AN capable L2 switches can 458 ensure that discovery messages terminate on them so that neighboring 459 AN routers and switches will only find the physcially connected AN L2 460 switches as their candidate ACP neighbors. With such a discovery 461 mechanism in place, the ACP and its security associations will only 462 need to scale to the number of physcial interfaces instead of a 463 potentially much larger number of "LAN-connected" neighbors. And the 464 ACP topology will follow directly the physical topology, something 465 which can then also be leveraged in management operations or by ASAs. 467 In the example above, consider ANswitch1 and ANswitchM are AN 468 capable, and ANswitch2 is not AN capable. The desired ACP topology 469 is therefore that ANrtr1 and ANrtrM only have an ACP connetion to 470 ANswitch1, and that ANswitch1, ANrtr2, ANrtrN have a full mesh of ACP 471 connection amongst each other. ANswitch1 also has an ACP connection 472 with ANswitchM and ANswitchM has ACP connections to anything else 473 behind it. 475 5.2.2. CDP/LLDP/mDNS considerations 477 LLDP (and Cisco's CDP) are example of L2 discovery protocols that 478 terminate their messages on L2 ports. If those protocols would be 479 chosen for ACP neighbor discovery, ACP neighbor discovery would 480 therefore also terminate on L2 ports. This would prevent ACP 481 construction over non-ANIMA switches. 483 mDNS operates at the subnet level, and is also used on L2 switches. 484 The authors of this document are not aware of mDNS implementation 485 that terminate their messages on L2 ports instead of the subnet 486 level. If mDNS was used as the ACP discovery mechanism on an ACP 487 capable L2 switch, then this would be necessary to implement. It is 488 likely that termination of mDNS messages could only be applied to all 489 mDNS messages from a port, which would then make it necessary to 490 software forward any non-ACP related mDNS messages to maintain prior 491 non-ACP mDNS functionality. With low performance of software 492 forwarding in many L2 switches, this could easily make the ACP 493 unsupportable on such L2 switches. 495 5.2.3. Discovery with GRASP 497 In conclusion for the above described considerations, the ACP uses 498 "insecure" instances of GRASP for discovery of ACP neighbors because 499 it can easily be set up to match the requiremetns without affecting 500 other uses of the protocol. 502 The name of the GRASP objective to announce/discover the capability 503 of a neighbor to run the ACP is "ACP". Section 3.5.2.2 of 504 [I-D.ietf-anima-grasp] describes the instance of GRASP to be used for 505 this purpose: "DULL" (Discovery Unsolicited Link Local). The precise 506 GRASP objective to be used is specified in Section 3 of 507 [I-D.carpenter-anima-ani-objectives]. 509 As explained above, in an ACP enabled L2 switch, each of these 510 instances would actually need to be per-L2-port. The result of the 511 discovery is the IPv6 link-local address of the neighbor. It is 512 stored in the AN Adjacency Table, see Section 5.1.2 which then drives 513 the further building of the ACP to that neighbor. 515 For example, ANswitch1 would run separate DULL GRASP instances on its 516 ports to ANrtr1, ANswitch2 and ANswitchI, even though all those three 517 ports may be in the data plane in the same (V)LAN. This is easily 518 achieved by extracting native GRASP multicast messages by their MAC 519 multicast destination address. None of the other type of GRASP 520 instances (eg: as used inside the ACP) use GRASP messages that would 521 be affected by such extraction, because all other GRASP messages have 522 encrypted encapsulations. 524 5.2.4. Discovery and BRSKY 526 Before a node has a domain certificate, it can not participate in the 527 ACP and therefore does also not try to discover an ACP neighbor. 528 Instead, it uses the discovery mechanism described in 529 [I-D.ietf-anima-grasp] to discover a bootstrap proxy. Currently, 530 that document describes mDNS as the protocol of choice for that 531 discovery. In the context of above topology example, ANrtr1 might 532 therefore discover and choose any ANrtr or ANswitch on the LAN that 533 is already part of the autonomic domain - instead of the closest one 534 which is ANswitch1. This choice of bootstrap proxy does not impact 535 in the later building of the ACP on ANrtr1 and is therefore not a 536 concern for the ACP. 538 Once a device has its domain certificate, it will start announcing 539 itself via GRASP as ACP capable. 541 When an autonomic device is a registrar, it will announce the 542 registrar function via GRASP in the ACP as the "6JOIN" objective. An 543 AN device that is a registrar or learns about one or more reachable 544 registrars via this GRASP/ACP announcements will announce itself as a 545 boostrap proxy via mDNS. See [I-D.richardson-anima-6join-discovery] 546 for more details. 548 5.3. Candidate ACP Neighbor Selection 550 An autonomic node must determine to which other autonomic nodes in 551 the adjacency table it should build an ACP connection. This is based 552 on the information in the AN Adjacency table. 554 The ACP is by default established exclusively between nodes in the 555 same domain. 557 Intent can change this default behaviour. Since Intent is 558 transported over the ACP, the first ACP connection a node establishes 559 is always following the default behaviour. The precise format for 560 this Intent needs to be defined outside this document. Example 561 Intent policies which need to be supported include: 563 o The ACP should be built between all sub-domains for a given parent 564 domain. For example: For domain "example.com", nodes of 565 "example.com", "access.example.com", "core.example.com" and 566 "city.core.example.com" should all establish one single ACP. 568 o Two domains should build one single ACP between themselves, for 569 example "example1.com" should establish the ACP also with nodes 570 from "example2.com". For this case, the two domains must be able 571 to validate their trust, typically by cross-signing their 572 certificate infrastructure. 574 The result of the candidate ACP neighbor selection process is a list 575 of adjacent or configured autonomic neighbors to which an ACP channel 576 should be established. The next step begins that channel 577 establishment. 579 5.4. Channel Selection 581 To avoid attacks, initial discovery of candidate ACP peers can not 582 include any non-protected negotiation. To avoid re-inventing and 583 validating security association mechanisms, the next step after 584 discoving the address of a candidate neighbor can only be to try 585 first to establish a security association with that neighbor using a 586 well-known security association method. 588 At this time in the lifecycle of autonomic devices, it is unclear 589 whether it is feasible to even decide on a single MTI (mandatory to 590 implement) security association protocol across all autonomic 591 devices. 593 From the use-cases it is clear that not all type of autonomic devices 594 can or need to connect directly to each other or are able to support 595 or prefer all possible mechanisms. For example, code space limited 596 IoT devices may only support dTLS (because that code exists already 597 on them for end-to-end security use-cases), but low-end in-ceiling L2 598 switches may only want to support MacSec because that is also 599 supported in HW, and only a more flexible garteway device may need to 600 support both of these mechanisms and potentially more. 602 To support these requirements, and make ACP channel negotiation also 603 easily extensible, the secure channel selection between Alice and Bob 604 is a potentially two stage process. In the first stage, Alice and 605 Bob directly try to establish a secure channel using the security- 606 association and channel protocols they support. One or more of these 607 protocols may simply be protocols not directly resulting in an ACP 608 channel, but instead establishing a secure negotiation channel 609 through which the final secure channel protocol is decided. If both 610 Alice and Bob support such a negotiation step, then this secured 611 negotiation channel is the first step, and the secure channel 612 protocol is the second step. 614 If Alice supports multiple security association protocols in the 615 first step, it is a matter of Alices local policy to determine the 616 order in which she will try to build the connection to Bob. To 617 support multiple security association protocols, Alice must be able 618 to simultaneously act as a responder in parallel for all of them - so 619 that she can respond to any order in which Bob wants to prefer 620 building the security association. 622 When ACP setup between Alice and Bob results in the first secure 623 association to be established, the peer with the higher Device-ID in 624 the certificate will stop building new security associations. The 625 peer with the lower certificate Device-ID is now responsible to 626 continue building its most preferred security association and to tear 627 down all but that most preferred one - unless the secure association 628 is one of a negotation protocols whose rules superceed this. 630 All this negotiation is in the context of an "L2 interface". Alice 631 and Bob will build ACP connections to each other on every "L2 632 interface" that they both connect to. 634 5.5. Security Association protocols 636 The following sections define the security association protocols that 637 we consider to be important and feasible to specify in this document. 638 In all cases, the mutual authentication is done via the autonomic 639 domain certificate of the peer as follows - unless superceeded by 640 Intent: 642 o The certificate is valid as proven by the security associations 643 protocol exchanges. 645 o If the certificate is included in a Certificate Revocation List 646 (CRL), the connection attempt is aborted and an error logged. 647 [EDNOTE: Do we want OCSP instead of CRL?] [EDNOTE: Distribution 648 of the CRL, and handling of CRL timeouts during network partition 649 needs to be discussed in more detail.] 651 o The peers certificate is signed by the same CA as the devices 652 domain certificate. 654 o The peers OU (Organizational Unit) field in the certificates 655 Subject is the same as in the devices certificate. 657 5.5.1. ACP via IPsec 659 To run ACP via IPsec transport mode, no further IANA assignments/ 660 definitions are required. All autonomic devices suppoting IPsec MUST 661 support IPsec security setup via IKEv2, transport mode encapsulation 662 via the device and peer link-local IPv6 addresses and AES256 663 encryption. 665 5.5.2. ACP via GRE/IPsec 667 In network devices it is often easier to provide virtual interfaces 668 on top of GRE encapsulation than natively on top of a simple IPsec 669 association. On those devices it may be necessary to run the ACP 670 secure channel on top of a GRE connection protected by the IPsec 671 association. The requirements for the IPsec association are the same 672 as described above, but instead of directly carrying the ACP IPv6 673 packets, the payload is an ACP IPv6 packet inside GRE/IPv6. 675 Note that without explicit negotiation (eg: via GRASP/TLS), this 676 method is incompatible to direct ACP via IPsec, so it must only be 677 used as an option during negotiation. 679 5.5.3. ACP via dTLS 681 To run ACP via UDP and dTLS v1.2 [RFC6347] an IANA assigned port 682 [TBD] is used. All autonomic devices supporting ACP via dTLS must 683 support AES256 encryption. 685 5.5.4. GRASP/TLS negotiation 687 To explicitly allow negotiation of the ACP channel protocol, GRASP 688 over a TLS connection using the GRASP_LISTEN_PORT and the devices and 689 peers link-local IPv6 address is used. When Alice and Bob support 690 GRASP negotiation, they do prefer it over any other non-explicitly 691 negotiated security association protocol and should wait trying any 692 non-negotiated ACP channel protocol until after it is clear that 693 GRASP/TLS will not work to the peer. 695 When Alice and Bob successfully establish the GRASP/TSL session, they 696 will initially negotiate the channel mechanism to use. Bob and Alice 697 each have a list of channel mehanisms they support, sorted by 698 preference. They negotiate the best mechansm supported by both of 699 them. In the absence of Intent, the mechanism choosen is the best 700 one for the device with the lower Device-ID. 702 After agreeing on a channel mechanism, Alice and Bob start the 703 selected Channel protocol. The GRASP/TLS connection can be kept 704 alive or timed out as long as the seelected channel protocol has a 705 secure association between Alice and Bob. When it terminates, it 706 needs to be re-negotiated via GRASP/TLS. 708 Negotiation of a channel type may require IANA assignments of code 709 points. See IANA Considerations (Section 11) for the formal 710 definition of those code points. 712 TBD: The exact negotiation steps in GRASP to achieve this outcome. 714 5.5.5. ACP Security Profiles 716 A baseline autonomic device MUST support IPsec and SHOULD support 717 GRASP/TLS and dTLS. A constrained autonomic device MUST support 718 dTLS. 720 Autonomic devices need to specify in documentation the set of secure 721 ACP mechanisms they suppport. 723 5.6. GRASP instance details 725 Received GRASP packets are assigned to an instance of GRASP by the 726 context they are received on: 728 o GRASP packets received on an ACP (virtual) interfaces are assigned 729 to the ACP instance of GRASP 731 o GRASP/UDP packets received on L2 interfaces/ports where the device 732 is willing to run ACP are assigned to a DULL instance of GRASP for 733 that interface/port. 735 o GRASP packets received inside a TLS connection established for 736 GRASP/TLS ACP negotiation are assigned to a separate instance of 737 GRASP for that negotiation. 739 5.7. Context Separation 741 The ACP is in a separate context from the normal data plane of the 742 device. This context includes the ACP channels IPv6 forwarding and 743 routing as well as any required higher layer ACP functions. 745 In classical network device platforms, a dedicated so called "Virtual 746 routing and forwarding instance" (VRF) is one logical implementation 747 option for the ACP. If possible by the platform SW architecture, 748 separation options that minimize shared components are preferred, 749 such as a logical container or virtual machine instance. The context 750 for the ACP needs to be established automatically during bootstrap of 751 a device. As much as possible it should be protected from being 752 modified unintentionally by data plane configuration. 754 Context separation improves security, because the ACP is not 755 reachable from the global routing table. Also, configuration errors 756 from the data plane setup do not affect the ACP. 758 5.8. Addressing inside the ACP 760 The channels explained above typically only establish communication 761 between two adjacent nodes. In order for communication to happen 762 across multiple hops, the autonomic control plane requires internal 763 network wide valid addresses and routing. Each autonomic node must 764 create a virtual interface with a network wide unique address inside 765 the ACP context mentioned in Section 5.7. This address may be used 766 also in other virtual contexts. 768 With the algorithm introduced here, all autonomic devices in the same 769 domain have the same /48 prefix. Conversely, global IDs from 770 different domains are unlikely to clash, such that two networks can 771 be merged, as long as the policy allows that merge. See also 772 Section 7 for a discussion on merging domains. 774 Links inside the ACP only use link-local IPv6 addressing, such that 775 each node only requires one routable virtual address. 777 5.8.1. Fundamental Concepts of Autonomic Addressing 779 o Usage: Autonomic addresses are exclusively used for self- 780 management functions inside a trusted domain. They are not used 781 for user traffic. Communications with entities outside the 782 trusted domain use another address space, for example normally 783 managed routable address space. 785 o Separation: Autonomic address space is used separately from user 786 address space and other address realms. This supports the 787 robustness requirement. 789 o Loopback-only: Only loopback interfaces of autonomic nodes carry a 790 routable address; all other interfaces exclusively use IPv6 link 791 local for autonomic functions. The usage of IPv6 link local 792 addressing is discussed in [RFC7404]. 794 o Use-ULA: For loopback interfaces of autonomic nodes, we use Unique 795 Local Addresses (ULA), as specified in [RFC4193]. An alternative 796 scheme was discussed, using assigned ULA addressing. The 797 consensus was to use ULA-random [[RFC4193] with L=1], because it 798 was deemed to be sufficient. 800 o No external connectivity: They do not provide access to the 801 Internet. If a node requires further reaching connectivity, it 802 should use another, traditionally managed address scheme in 803 parallel. 805 o Addresses in the ACP are permanent, and do not support temporary 806 addresses as defined in [RFC4941]. 808 The ACP is based exclusively on IPv6 addressing, for a variety of 809 reasons: 811 o Simplicity, reliability and scale: If other network layer 812 protocols were supported, each would have to have its own set of 813 security associations, routing table and process, etc. 815 o Autonomic functions do not require IPv4: Autonomic functions and 816 autonomic service agents are new concepts. They can be 817 exclusively built on IPv6 from day one. There is no need for 818 backward compatibility. 820 o OAM protocols no not require IPv4: The ACP may carry OAM 821 protocols. All relevant protocols (SNMP, TFTP, SSH, SCP, Radius, 822 Diameter, ...) are available in IPv6. 824 5.8.2. The ACP Addressing Base Scheme 826 The Base ULA addressing scheme for autonomic nodes has the following 827 format: 829 8 40 3 77 830 +--+--------------+------+------------------------------------------+ 831 |FD| hash(domain) | Type | (sub-scheme) | 832 +--+--------------+------+------------------------------------------+ 834 Figure 3: ACP Addressing Base Scheme 836 The first 48 bits follow the ULA scheme, as defined in [RFC4193], to 837 which a type field is added: 839 o "FD" identifies a locally defined ULA address. 841 o The ULA "global ID" is set here to be a hash of the domain name, 842 which results in a pseudo-random 40 bit value. It is calculated 843 as the first 40 bits of the SHA256 hash of the domain name, in the 844 example "example.com". 846 o Type: This field allows different address sub-schemes in the 847 future. The goal is to start with a single sub-schemes, but to 848 allow for extensions later if and when required. This addresses 849 the "upgradability" requirement. Assignment of types for this 850 field should be maintained by IANA. 852 5.8.3. ACP Addressing Sub-Scheme 854 The sub-scheme defined here is defined by the Type value 0 (zero) in 855 the base scheme. 857 51 13 63 1 858 +------------------------+---------+----------------------------+---+ 859 | (base scheme) | Zone ID | Device ID | V | 860 +------------------------+---------+----------------------------+---+ 862 Figure 4: ACP Addressing Sub-Scheme 864 The fields are defined as follows: [Editor's note: The lengths of the 865 fields is for discussion.] 867 o Zone ID: If set to all zero bits: The device ID bits are used as 868 an identifier (as opposed to a locator). This results in a non- 869 hierarchical, flat addressing scheme. Any other value indicates a 870 zone. See section Section 5.8.4 on how this field is used in 871 detail. 873 o Device ID: A unique value for each device. 875 o V: Virtualization bit: 0: autonomic node base system; 1: a virtual 876 context on an autonomic node. 878 The device ID is derived as follows: In an Autonomic Network, a 879 registrar is enrolling new devices. As part of the enrolment process 880 the registrar assigns a number to the device, which is unique for 881 this registrar, but not necessarily unique in the domain. The 64 bit 882 device ID is then composed as: 884 o 48 bit: Registrar ID, a number unique inside the domain that 885 identifies the registrar which assigned the name to the device. A 886 MAC address of the registrar can be used for this purpose. 888 o 15 bit: Device number, a number which is unique for a given 889 registrar, to identify the device. This can be a sequentially 890 assigned number. 892 The "device ID" itself is unique in a domain (i.e., the Zone-ID is 893 not required for uniqueness). Therefore, a device can be addressed 894 either as part of a flat hierarchy (zone ID = 0), or with an 895 aggregation scheme (any other zone ID). A address with zone-ID = 0 896 is an identifier, with another zone-ID as a locator. See 897 Section 5.8.4 for a description of the zone bits. 899 This addressing sub-scheme allows the direct addressing of specific 900 virtual containers / VMs on an autonomic node. An increasing number 901 of hardware platforms have a distributed architecture, with a base OS 902 for the node itself, and the support for hardware blades with 903 potentially different OSs. The VMs on the blades could be considered 904 as separate autonomic nodes, in which case it would make sense to be 905 able to address them directly. Autonomic Service Agents (ASAs) could 906 be instantiated in either the base OS, or one of the VMs on a blade. 907 This addressing scheme allows for the easy separation of the hardware 908 context. 910 The location of the V bit(s) at the end of the address allows to 911 announce a single prefix for each autonomic node, while having 912 separate virtual contexts addressable directly. 914 [EDNOTE: various suggestions from mcr in his mail from 30 Nov 2016 to 915 be considered (https://mailarchive.ietf.org/arch/msg/anima/ 916 nZpEphrTqDCBdzsKMpaIn2gsIzI).] 918 5.8.4. Usage of the Zone Field 920 The "zone ID" allows for the introduction of structure in the 921 addressing scheme. 923 Zone = zero is the default addressing scheme in an autonomic domain. 924 Every autonomic node MUST respond to its ACP address with zone=0. 925 Used on its own this leads to a non-hierarchical address scheme, 926 which is suitable for networks up to a certain size. In this case, 927 the addresses primarily act as identifiers for the nodes, and 928 aggregation is not possible. 930 If aggregation is required, the 13 bit value allows for up to 8191 931 zones. The allocation of zone numbers may either happen 932 automatically through a to-be-defined algorithm; or it could be 933 configured and maintained manually. 935 If a device learns through an autonomic method or through 936 configuration that it is part of a zone, it MUST also respond to its 937 ACP address with that zone number. In this case the ACP loopback is 938 configured with two ACP addresses: One for zone 0 and one for the 939 assigned zone. This method allows for a smooth transition between a 940 flat addressing scheme and an hierarchical one. 942 (Theoretically, the 13 bits for the zone ID would allow also for two 943 levels of zones, introducing a sub-hierarchy. We do not think this 944 is required at this point, but a new type could be used in the future 945 to support such a scheme.) 947 Note: Another way to introduce hierarchy is to use sub-domains in the 948 naming scheme. The node names "node17.subdomainA.example.com" and 949 "node4.subdomainB.example.com" would automatically lead to different 950 ULA prefixes, which can be used to introduce a routing hierarchy in 951 the network, assuming that the subdomains are aligned with routing 952 areas. 954 5.8.5. Other ACP Addressing Sub-Schemes 956 Other ACP addressing sub-schemes can be defined if and when required. 957 IANA will assign a new "type" for each new addressing sub-scheme. 959 5.9. Routing in the ACP 961 Once ULA address are set up all autonomic entities should run a 962 routing protocol within the autonomic control plane context. This 963 routing protocol distributes the ULA created in the previous section 964 for reachability. The use of the autonomic control plane specific 965 context eliminates the probable clash with the global routing table 966 and also secures the ACP from interference from the configuration 967 mismatch or incorrect routing updates. 969 The establishment of the routing plane and its parameters are 970 automatic and strictly within the confines of the autonomic control 971 plane. Therefore, no manual configuration is required. 973 All routing updates are automatically secured in transit as the 974 channels of the autonomic control plane are by default secured, and 975 this routing runs only inside the ACP. 977 The routing protocol inside the ACP is RPL [RFC6550]. See Appendix A 978 for more details on the choice of RPL. 980 [EDNOTE: Need to decide: storing / non-storing mode; mcr suggests 981 storing mode. Need to define more parameters in detail.] 983 5.10. General ACP Considerations 985 In order to be independent of configured link addresses, channels 986 SHOULD use IPv6 link local addresses between adjacent neighbors 987 wherever possible. This way, the ACP tunnels are independent of 988 correct network wide routing. 990 Since channels are by default established between adjacent neighbors, 991 the resulting overlay network does hop by hop encryption. Each node 992 decrypts incoming traffic from the ACP, and encrypts outgoing traffic 993 to its neighbors in the ACP. Routing is discussed in Section 5.9. 995 If two nodes are connected via several links, the ACP SHOULD be 996 established on every link, but it is possible to establish the ACP 997 only on a sub-set of links. Having an ACP channel on every link has 998 a number of advantages, for example it allows for a faster failover 999 in case of link failure, and it reflects the physical topology more 1000 closely. Using a subset of links (for example, a single link), 1001 reduces resource consumption on the devices, because state needs to 1002 be kept per ACP channel. 1004 6. Workarounds for Non-Autonomic Nodes 1006 6.1. Connecting a Non-Autonomic Controller / NMS system 1008 The Autonomic Control Plane can be used by management systems, such 1009 as controllers or network management system (NMS) hosts (henceforth 1010 called simply "NMS hosts"), to connect to devices through it. For 1011 this, an NMS host must have access to the ACP. The ACP is a self- 1012 protecting overlay network, which allows by default access only to 1013 trusted, autonomic systems. Therefore, a traditional, non-autonomic 1014 NMS system does not have access to the ACP by default, just like any 1015 other external device. 1017 If the NMS host is not autonomic, i.e., it does not support autonomic 1018 negotiation of the ACP, then it can be brought into the ACP by 1019 explicit configuration. On an adjacent autonomic node with ACP, the 1020 interface with the NMS host can be configured as "ACP Connect". In 1021 this case, all devices on this port, including the NMS host, is 1022 entirely and exclusively inside the ACP. It would likely require a 1023 second interface outside the ACP for connections between the NMS host 1024 and administrators, or Internet based services. This mode of 1025 connecting an NMS host has security consequences: All systems and 1026 processes connected to this implicitly trusted "ACP Connect" 1027 interface have access to all autonomic nodes on the entire ACP, 1028 without further authentication. Thus, this connection must be 1029 physically controlled. 1031 The non-autonomic NMS host must be routed in the ACP. This involves 1032 two parts: 1) the NMS host must point default to the AN device for 1033 the ULA prefix used inside the ACP, and 2) the prefix used between AN 1034 node and NMS host must be announced into the ACP, and distributed 1035 there. 1037 The document "Autonomic Network Stable Connectivity" 1038 [I-D.ietf-anima-stable-connectivity] explains in more detail how the 1039 ACP can be integrated in a mixed NOC environment. 1041 If an NMS host is autonomic itself, it negotiates access to the ACP 1042 with its neighbor, like any other autonomic node. 1044 6.2. ACP through Non-Autonomic L3 Clouds 1046 Not all devices in a network may be autonomic. If non-autonomic 1047 Layer-2 devices are between autonomic nodes, the communications 1048 described in this document should work, since it is IP based. 1049 However, non-autonomic Layer-3 devices do not forward link local 1050 autonomic messages, and thus break the Autonomic Control Plane. 1052 One workaround is to manually configure IP tunnels between autonomic 1053 nodes across a non-autonomic Layer-3 cloud. The tunnels are 1054 represented on each autonomic node as virtual interfaces, and all 1055 autonomic transactions work across such tunnels. 1057 Such manually configured tunnels are less "indestructible" than an 1058 automatically created ACP based on link local addressing, since they 1059 depend on correct data plane operations, such as routing and 1060 addressing. 1062 Future work should envisage an option where the edge device of the L3 1063 cloud is configured to automatically forward ACP discovery messages 1064 to the right exit point. This optimisation is not considered in this 1065 document. 1067 7. Self-Healing Properties 1069 The ACP is self-healing: 1071 o New neighbors will automatically join the ACP after successful 1072 validation and will become reachable using their unique ULA 1073 address across the ACP. 1075 o When any changes happen in the topology, the routing protocol used 1076 in the ACP will automatically adapt to the changes and will 1077 continue to provide reachability to all devices. 1079 o If an existing device gets revoked, it will automatically be 1080 denied access to the ACP as its domain certificate will be 1081 validated against a Certificate Revocation List during 1082 authentication. Since the revocation check is only done at the 1083 establishment of a new security association, existing ones are not 1084 automatically torn down. If an immediate disconnect is required, 1085 existing sessions to a freshly revoked device can be re-set. 1087 The ACP can also sustain network partitions and mergers. Practically 1088 all ACP operations are link local, where a network partition has no 1089 impact. Devices authenticate each other using the domain 1090 certificates to establish the ACP locally. Addressing inside the ACP 1091 remains unchanged, and the routing protocol inside both parts of the 1092 ACP will lead to two working (although partitioned) ACPs. 1094 There are few central dependencies: A certificate revocation list 1095 (CRL) may not be available during a network partition; a suitable 1096 policy to not immediately disconnect neighbors when no CRL is 1097 available can address this issue. Also, a registrar or Certificate 1098 Authority might not be available during a partition. This may delay 1099 renewal of certificates that are to expire in the future, and it may 1100 prevent the enrolment of new devices during the partition. 1102 After a network partition, a re-merge will just establish the 1103 previous status, certificates can be renewed, the CRL is available, 1104 and new devices can be enrolled everywhere. Since all devices use 1105 the same trust anchor, a re-merge will be smooth. 1107 Merging two networks with different trust anchors requires the trust 1108 anchors to mutually trust each other (for example, by cross-signing). 1109 As long as the domain names are different, the addressing will not 1110 overlap (see Section 5.8). 1112 8. Self-Protection Properties 1114 As explained in Section 5, the ACP is based on secure channels built 1115 between devices that have mutually authenticated each other with 1116 their domain certificates. The channels themselves are protected 1117 using standard encryption technologies like DTLS or IPsec which 1118 provide additional authentication during channel establishment, data 1119 integrity and data confidentiality protection of data inside the ACP 1120 and in addition, provide replay protection. 1122 An attacker will therefore not be able to join the ACP unless having 1123 a valid domain certificate, also packet injection and sniffing 1124 traffic will not be possible due to the security provided by the 1125 encryption protocol. 1127 The remaining attack vector would be to attack the underlying AN 1128 protocols themselves, either via directed attacks or by denial-of- 1129 service attacks. However, as the ACP is built using link-local IPv6 1130 address, remote attacks are impossible. The ULA addresses are only 1131 reachable inside the ACP context, therefore unreachable from the data 1132 plane. Also, the ACP protocols should be implemented to be attack 1133 resistant and not consume unnecessary resources even while under 1134 attack. 1136 9. The Administrator View 1138 An ACP is self-forming, self-managing and self-protecting, therefore 1139 has minimal dependencies on the administrator of the network. 1140 Specifically, since it is independent of configuration, there is no 1141 scope for configuration errors on the ACP itself. The administrator 1142 may have the option to enable or disable the entire approach, but 1143 detailed configuration is not possible. This means that the ACP must 1144 not be reflected in the running configuration of devices, except a 1145 possible on/off switch. 1147 While configuration is not possible, an administrator must have full 1148 visibility of the ACP and all its parameters, to be able to do 1149 trouble-shooting. Therefore, an ACP must support all show and debug 1150 options, as for any other network function. Specifically, a network 1151 management system or controller must be able to discover the ACP, and 1152 monitor its health. This visibility of ACP operations must clearly 1153 be separated from visibility of data plane so automated systems will 1154 never have to deal with ACP aspect unless they explicitly desire to 1155 do so. 1157 Since an ACP is self-protecting, a device not supporting the ACP, or 1158 without a valid domain certificate cannot connect to it. This means 1159 that by default a traditional controller or network management system 1160 cannot connect to an ACP. See Section 6.1 for more details on how to 1161 connect an NMS host into the ACP. 1163 10. Security Considerations 1165 An ACP is self-protecting and there is no need to apply configuration 1166 to make it secure. Its security therefore does not depend on 1167 configuration. 1169 However, the security of the ACP depends on a number of other 1170 factors: 1172 o The usage of domain certificates depends on a valid supporting PKI 1173 infrastructure. If the chain of trust of this PKI infrastructure 1174 is compromised, the security of the ACP is also compromised. This 1175 is typically under the control of the network administrator. 1177 o Security can be compromised by implementation errors (bugs), as in 1178 all products. 1180 There is no prevention of source-address spoofing inside the ACP. 1181 This implies that if an attacker gains access to the ACP, (s)he can 1182 spoof all addresses inside the ACP and fake messages from any other 1183 device. 1185 Fundamentally, security depends on correct operation, implementation 1186 and architecture. Autonomic approaches such as the ACP largely 1187 eliminate the dependency on correct operation; implementation and 1188 architectural mistakes are still possible, as in all networking 1189 technologies. 1191 11. IANA Considerations 1193 Section 5.5.3 describes ACP over dTLS, which requires a well-known 1194 UDP port. We request IANA to assign this UDP port for 'ACP over 1195 dTLS'. 1197 Section 5.5.4 describes an option for the channel negotiation, the 1198 'ACP channel type'. We request IANA to create a registry for 'ACP 1199 channel type'. 1201 The ACP channel type is a 8-bit unsigned integer. This document only 1202 assigns the first value. 1204 Number | Channel Type | RFC 1205 ---------+-----------------------------------+------------ 1206 0 | GRE tunnel protected with | this document 1207 | IPsec transport mode | 1208 1-255 | reserved for future channel types | 1210 Section 5.8.2 describes a 'type' field in the base addressing scheme. 1211 We request IANA to create a registry for the 'ACP addressing scheme 1212 type', with the following initial values: 1214 Number | Address Type (sub-scheme) | RFC 1215 ---------+-----------------------------------+------------ 1216 0 | Default address sub-scheme | this document 1217 7 | Reserved for private use | 1218 | sub-scheme | 1220 12. Acknowledgements 1222 This work originated from an Autonomic Networking project at Cisco 1223 Systems, which started in early 2010. Many people contributed to 1224 this project and the idea of the Autonomic Control Plane, amongst 1225 which (in alphabetical order): Ignas Bagdonas, Parag Bhide, Balaji 1226 BL, Alex Clemm, Yves Hertoghs, Bruno Klauser, Max Pritikin, Ravi 1227 Kumar Vadapalli. 1229 Further input and suggestions were received from: Rene Struik, Brian 1230 Carpenter, Benoit Claise. 1232 13. Change log [RFC Editor: Please remove] 1234 13.1. Initial version 1236 First version of this document: draft-behringer-autonomic-control- 1237 plane 1239 13.2. draft-behringer-anima-autonomic-control-plane-00 1241 Initial version of the anima document; only minor edits. 1243 13.3. draft-behringer-anima-autonomic-control-plane-01 1245 o Clarified that the ACP should be based on, and support only IPv6. 1247 o Clarified in intro that ACP is for both, between devices, as well 1248 as for access from a central entity, such as an NMS. 1250 o Added a section on how to connect an NMS system. 1252 o Clarified the hop-by-hop crypto nature of the ACP. 1254 o Added several references to GDNP as a candidate protocol. 1256 o Added a discussion on network split and merge. Although, this 1257 should probably go into the certificate management story longer 1258 term. 1260 13.4. draft-behringer-anima-autonomic-control-plane-02 1262 Addresses (numerous) comments from Brian Carpenter. See mailing list 1263 for details. The most important changes are: 1265 o Introduced a new section "overview", to ease the understanding of 1266 the approach. 1268 o Merged the previous "problem statement" and "use case" sections 1269 into a mostly re-written "use cases" section, since they were 1270 overlapping. 1272 o Clarified the relationship with draft-ietf-anima-stable- 1273 connectivity 1275 13.5. draft-behringer-anima-autonomic-control-plane-03 1277 o Took out requirement for IPv6 --> that's in the reference doc. 1279 o Added requirement section. 1281 o Changed focus: more focus on autonomic functions, not only virtual 1282 out of band. This goes a bit throughout the document, starting 1283 with a changed abstract and intro. 1285 13.6. draft-ietf-anima-autonomic-control-plane-00 1287 No changes; re-submitted as WG document. 1289 13.7. draft-ietf-anima-autonomic-control-plane-01 1291 o Added some paragraphs in addressing section on "why IPv6 only", to 1292 reflect the discussion on the list. 1294 o Moved the data-plane ACP out of the main document, into an 1295 appendix. The focus is now the virtually separated ACP, since it 1296 has significant advantages, and isn't much harder to do. 1298 o Changed the self-creation algorithm: Part of the initial steps go 1299 into the reference document. This document now assumes an 1300 adjacency table, and domain certificate. How those get onto the 1301 device is outside scope for this document. 1303 o Created a new section 6 "workarounds for non-autonomic nodes", and 1304 put the previous controller section (5.9) into this new section. 1305 Now, section 5 is "autonomic only", and section 6 explains what to 1306 do with non-autonomic stuff. Much cleaner now. 1308 o Added an appendix explaining the choice of RPL as a routing 1309 protocol. 1311 o Formalised the creation process a bit more. Now, we create a 1312 "candidate peer list" from the adjacency table, and form the ACP 1313 with those candidates. Also it explains now better that policy 1314 (Intent) can influence the peer selection. (section 4 and 5) 1316 o Introduce a section for the capability negotiation protocol 1317 (section 7). This needs to be worked out in more detail. This 1318 will likely be based on GRASP. 1320 o Introduce a new parameter: ACP tunnel type. And defines it in the 1321 IANA considerations section. Suggest GRE protected with IPSec 1322 transport mode as the default tunnel type. 1324 o Updated links, lots of small edits. 1326 13.8. draft-ietf-anima-autonomic-control-plane-02 1328 o Added explicitly text for the ACP channel negotiation. 1330 o Merged draft-behringer-anima-autonomic-addressing-02 into this 1331 document, as suggested by WG chairs. 1333 13.9. draft-ietf-anima-autonomic-control-plane-03 1335 o Changed Neighbor discovery protocol from GRASP to mDNS. Bootstrap 1336 protocol team decided to go with mDNS to discover bootstrap proxy, 1337 and ACP should be consistent with this. Reasons to go with mDNS 1338 in bootstrap were a) Bootstrap should be reuseable also outside of 1339 full anima solutions and introduce as few as possible new 1340 elements. mDNS was considered well-known and very-likely even pre- 1341 existing in low-end devices (IoT). b) Using GRASP both for the 1342 insecure neighbor discovery and secure ACP operatations raises the 1343 risk of introducing security issues through implementation issues/ 1344 non-isolation between those two instances of GRASP. 1346 o Shortened the section on GRASP instances, because with mDNS being 1347 used for discovery, there is no insecure GRASP session any longer, 1348 simplifying the GRASP considerations. 1350 o Added certificate requirements for ANIMA in section 5.1.1, 1351 specifically how the ANIMA information is encoded in 1352 subjectAltName. 1354 o Deleted the appendix on "ACP without separation", as originally 1355 planned, and the paragraph in the main text referring to it. 1357 o Deleted one sub-addressing scheme, focusing on a single scheme 1358 now. 1360 o Included information on how ANIMA information must be encoded in 1361 the domain certificate in Section 5.1. 1363 o Editorial changes, updated draft references, etc. 1365 13.10. draft-ietf-anima-autonomic-control-plane-04 1367 Changed discovery of ACP neighbor back from mDNS to GRASP after 1368 revisiting the L2 problem. Described problem in discovery section 1369 itself to justify. Added text to explain how ACP discovery relates 1370 to BRSKY (bootstrap) discovery and pointed to Michael Richardsons 1371 draft detailing it. Removed appendix section that contained the 1372 original explanations why GRASP would be useful (current text is 1373 meant to be better). 1375 13.11. draft-ietf-anima-autonomic-control-plane-05 1377 o Section 5.3 (candidate ACP neighbor selection): Add that Intent 1378 can override only AFTER an initial default ACP establishment. 1380 o Section 5.8.1 (addressing): State that addresses in the ACP are 1381 permanent, and do not support temporary addresses as defined in 1382 RFC4941. 1384 o Modified Section 5.2.3 to point to the GRASP objective defined in 1385 [I-D.carpenter-anima-ani-objectives]. (and added that reference) 1387 o Section 5.8.2: changed from MD5 for calculating the first 40 bits 1388 to SHA256; reason is MD5 should not be used any more. 1390 o Added address sub-scheme to the IANA section. 1392 o Made the routing section more prescriptive. 1394 o Clarified in Section 6.1 the ACP Connect port, and defined that 1395 term "ACP Connect". 1397 o Section 6.2: Added some thoughts (from mcr) on how traversing a L3 1398 cloud could be automated. 1400 o Added a CRL check in Section 5.5. 1402 o Added a note on the possibility of source-address spoofing into 1403 the security considerations section. 1405 o Other editoral changes, including those proposed by Michael 1406 Richardson on 30 Nov 2016 (see ANIMA list). 1408 14. References 1410 [I-D.carpenter-anima-ani-objectives] 1411 Carpenter, B. and B. Liu, "Technical Objectives for the 1412 Autonomic Network Infrastructure", draft-carpenter-anima- 1413 ani-objectives-00 (work in progress), November 2016. 1415 [I-D.ietf-anima-bootstrapping-keyinfra] 1416 Pritikin, M., Richardson, M., Behringer, M., Bjarnason, 1417 S., and K. Watsen, "Bootstrapping Remote Secure Key 1418 Infrastructures (BRSKI)", draft-ietf-anima-bootstrapping- 1419 keyinfra-04 (work in progress), October 2016. 1421 [I-D.ietf-anima-grasp] 1422 Bormann, C., Carpenter, B., and B. Liu, "A Generic 1423 Autonomic Signaling Protocol (GRASP)", draft-ietf-anima- 1424 grasp-09 (work in progress), December 2016. 1426 [I-D.ietf-anima-reference-model] 1427 Behringer, M., Carpenter, B., Eckert, T., Ciavaglia, L., 1428 Pierre, P., Liu, B., Nobre, J., and J. Strassner, "A 1429 Reference Model for Autonomic Networking", draft-ietf- 1430 anima-reference-model-02 (work in progress), July 2016. 1432 [I-D.ietf-anima-stable-connectivity] 1433 Eckert, T. and M. Behringer, "Using Autonomic Control 1434 Plane for Stable Connectivity of Network OAM", draft-ietf- 1435 anima-stable-connectivity-01 (work in progress), July 1436 2016. 1438 [I-D.richardson-anima-6join-discovery] 1439 Richardson, M., "GRASP discovery of Registrar by Join 1440 Assistant", draft-richardson-anima-6join-discovery-00 1441 (work in progress), October 2016. 1443 [RFC4122] Leach, P., Mealling, M., and R. Salz, "A Universally 1444 Unique IDentifier (UUID) URN Namespace", RFC 4122, 1445 DOI 10.17487/RFC4122, July 2005, 1446 . 1448 [RFC4193] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 1449 Addresses", RFC 4193, DOI 10.17487/RFC4193, October 2005, 1450 . 1452 [RFC4941] Narten, T., Draves, R., and S. Krishnan, "Privacy 1453 Extensions for Stateless Address Autoconfiguration in 1454 IPv6", RFC 4941, DOI 10.17487/RFC4941, September 2007, 1455 . 1457 [RFC5082] Gill, V., Heasley, J., Meyer, D., Savola, P., Ed., and C. 1458 Pignataro, "The Generalized TTL Security Mechanism 1459 (GTSM)", RFC 5082, DOI 10.17487/RFC5082, October 2007, 1460 . 1462 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 1463 Housley, R., and W. Polk, "Internet X.509 Public Key 1464 Infrastructure Certificate and Certificate Revocation List 1465 (CRL) Profile", RFC 5280, DOI 10.17487/RFC5280, May 2008, 1466 . 1468 [RFC5952] Kawamura, S. and M. Kawashima, "A Recommendation for IPv6 1469 Address Text Representation", RFC 5952, 1470 DOI 10.17487/RFC5952, August 2010, 1471 . 1473 [RFC6347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1474 Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347, 1475 January 2012, . 1477 [RFC6550] Winter, T., Ed., Thubert, P., Ed., Brandt, A., Hui, J., 1478 Kelsey, R., Levis, P., Pister, K., Struik, R., Vasseur, 1479 JP., and R. Alexander, "RPL: IPv6 Routing Protocol for 1480 Low-Power and Lossy Networks", RFC 6550, 1481 DOI 10.17487/RFC6550, March 2012, 1482 . 1484 [RFC6762] Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762, 1485 DOI 10.17487/RFC6762, February 2013, 1486 . 1488 [RFC6763] Cheshire, S. and M. Krochmal, "DNS-Based Service 1489 Discovery", RFC 6763, DOI 10.17487/RFC6763, February 2013, 1490 . 1492 [RFC7404] Behringer, M. and E. Vyncke, "Using Only Link-Local 1493 Addressing inside an IPv6 Network", RFC 7404, 1494 DOI 10.17487/RFC7404, November 2014, 1495 . 1497 [RFC7575] Behringer, M., Pritikin, M., Bjarnason, S., Clemm, A., 1498 Carpenter, B., Jiang, S., and L. Ciavaglia, "Autonomic 1499 Networking: Definitions and Design Goals", RFC 7575, 1500 DOI 10.17487/RFC7575, June 2015, 1501 . 1503 [RFC7576] Jiang, S., Carpenter, B., and M. Behringer, "General Gap 1504 Analysis for Autonomic Networking", RFC 7576, 1505 DOI 10.17487/RFC7576, June 2015, 1506 . 1508 Appendix A. Background on the choice of routing protocol 1510 In a pre-standard implementation, the "IPv6 Routing Protocol for Low- 1511 Power and Lossy Networks (RPL, [RFC6550] was chosen. This 1512 Appendix explains the reasoning behind that decision. 1514 Requirements for routing in the ACP are: 1516 o Self-management: The ACP must build automatically, without human 1517 intervention. Therefore routing protocol must also work 1518 completely automatically. RPL is a simple, self-managing 1519 protocol, which does not require zones or areas; it is also self- 1520 configuring, since configuration is carried as part of the 1521 protocol (see Section 6.7.6 of [RFC6550]). 1523 o Scale: The ACP builds over an entire domain, which could be a 1524 large enterprise or service provider network. The routing 1525 protocol must therefore support domains of 100,000 nodes or more, 1526 ideally without the need for zoning or separation into areas. RPL 1527 has this scale property. This is based on extensive use of 1528 default routing. RPL also has other scalability improvements, 1529 such as selecting only a subset of peers instead of all possible 1530 ones, and trickle support for information synchronisation. 1532 o Low resource consumption: The ACP supports traditional network 1533 infrastructure, thus runs in addition to traditional protocols. 1534 The ACP, and specifically the routing protocol must have low 1535 resource consumption both in terms of memory and CPU requirements. 1536 Specifically, at edge nodes, where memory and CPU are scarce, 1537 consumption should be minimal. RPL builds a destination-oriented 1538 directed acyclic graph (DODAG), where the main resource 1539 consumption is at the root of the DODAG. The closer to the edge 1540 of the network, the less state needs to be maintained. This 1541 adapts nicely to the typical network design. Also, all changes 1542 below a common parent node are kept below that parent node. 1544 o Support for unstructured address space: In the Autonomic 1545 Networking Infrastructure, node addresses are identifiers, and may 1546 not be assigned in a topological way. Also, nodes may move 1547 topologically, without changing their address. Therefore, the 1548 routing protocol must support completely unstructured address 1549 space. RPL is specifically made for mobile ad-hoc networks, with 1550 no assumptions on topologically aligned addressing. 1552 o Modularity: To keep the initial implementation small, yet allow 1553 later for more complex methods, it is highly desirable that the 1554 routing protocol has a simple base functionality, but can import 1555 new functional modules if needed. RPL has this property with the 1556 concept of "objective function", which is a plugin to modify 1557 routing behaviour. 1559 o Extensibility: Since the Autonomic Networking Infrastructure is a 1560 new concept, it is likely that changes in the way of operation 1561 will happen over time. RPL allows for new objective functions to 1562 be introduced later, which allow changes to the way the routing 1563 protocol creates the DAGs. 1565 o Multi-topology support: It may become necessary in the future to 1566 support more than one DODAG for different purposes, using 1567 different objective functions. RPL allow for the creation of 1568 several parallel DODAGs, should this be required. This could be 1569 used to create different topologies to reach different roots. 1571 o No need for path optimisation: RPL does not necessarily compute 1572 the optimal path between any two nodes. However, the ACP does not 1573 require this today, since it carries mainly non-delay-sensitive 1574 feedback loops. It is possible that different optimisation 1575 schemes become necessary in the future, but RPL can be expanded 1576 (see point "Extensibility" above). 1578 Authors' Addresses 1580 Michael H. Behringer (editor) 1581 Cisco Systems 1582 Building D, 45 Allee des Ormes 1583 Mougins 06250 1584 France 1586 Email: mbehring@cisco.com 1588 Toerless Eckert 1590 Email: tte+ietf@cs.fau.de 1592 Steinthor Bjarnason 1593 Arbor Networks 1594 2727 South State Street, Suite 200 1595 Ann Arbor MI 48104 1596 United States 1598 Email: sbjarnason@arbor.net