idnits 2.17.1 draft-ietf-v6ops-nd-cache-init-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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (September 6, 2020) is 875 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: draft-ietf-6man-grand has been published as RFC 9131 -- Obsolete informational reference (is this intentional?): RFC 4941 (Obsoleted by RFC 8981) Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 v6ops J. Linkova 3 Internet-Draft Google 4 Intended status: Informational September 6, 2020 5 Expires: March 10, 2021 7 Neighbor Cache Entries on First-Hop Routers: Operational Considerations 8 draft-ietf-v6ops-nd-cache-init-05 10 Abstract 12 Neighbor Discovery (RFC4861) is used by IPv6 nodes to determine the 13 link-layer addresses of neighboring nodes as well as to discover and 14 maintain reachability information. This document discusses how the 15 neighbor discovery state machine on a first-hop router is causing 16 user-visible connectivity issues when a new (not being seen on the 17 network before) IPv6 address is being used. The various approaches 18 to mitigate the problem are described, with the proposed solution 19 fully documented in I-D.ietf-6man-grand. 21 Status of This Memo 23 This Internet-Draft is submitted in full conformance with the 24 provisions of BCP 78 and BCP 79. 26 Internet-Drafts are working documents of the Internet Engineering 27 Task Force (IETF). Note that other groups may also distribute 28 working documents as Internet-Drafts. The list of current Internet- 29 Drafts is at https://datatracker.ietf.org/drafts/current/. 31 Internet-Drafts are draft documents valid for a maximum of six months 32 and may be updated, replaced, or obsoleted by other documents at any 33 time. It is inappropriate to use Internet-Drafts as reference 34 material or to cite them other than as "work in progress." 36 This Internet-Draft will expire on March 10, 2021. 38 Copyright Notice 40 Copyright (c) 2020 IETF Trust and the persons identified as the 41 document authors. All rights reserved. 43 This document is subject to BCP 78 and the IETF Trust's Legal 44 Provisions Relating to IETF Documents 45 (https://trustee.ietf.org/license-info) in effect on the date of 46 publication of this document. Please review these documents 47 carefully, as they describe your rights and restrictions with respect 48 to this document. Code Components extracted from this document must 49 include Simplified BSD License text as described in Section 4.e of 50 the Trust Legal Provisions and are provided without warranty as 51 described in the Simplified BSD License. 53 Table of Contents 55 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 56 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 4 57 1.2. Terminology . . . . . . . . . . . . . . . . . . . . . . . 4 58 2. Proposed Solution . . . . . . . . . . . . . . . . . . . . . . 5 59 2.1. Solution Requirements . . . . . . . . . . . . . . . . . . 5 60 2.2. Solution Overview . . . . . . . . . . . . . . . . . . . . 5 61 3. Solutions Considered but Discarded . . . . . . . . . . . . . 6 62 3.1. Do Nothing . . . . . . . . . . . . . . . . . . . . . . . 7 63 3.2. Change to the Registration-Based Neighbor Discovery . . . 7 64 3.3. Host Sending NS to the Router Address from Its GUA . . . 7 65 3.4. Host Sending Router Solicitation from its GUA . . . . . . 8 66 3.5. Routers Populating Their Caches by Gleaning From Neighbor 67 Discovery Packets . . . . . . . . . . . . . . . . . . . . 9 68 3.6. Initiating Hosts-to-Routers Communication . . . . . . . . 9 69 3.7. Transit Dataplane Traffic From a New Address Triggering 70 Address Resolution . . . . . . . . . . . . . . . . . . . 10 71 4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 72 5. Security Considerations . . . . . . . . . . . . . . . . . . . 10 73 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 10 74 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 11 75 7.1. Normative References . . . . . . . . . . . . . . . . . . 11 76 7.2. Informative References . . . . . . . . . . . . . . . . . 12 77 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 12 79 1. Introduction 81 The section 7.2.5 of [RFC4861] states: "When a valid Neighbor 82 Advertisement is received (either solicited or unsolicited), the 83 Neighbor Cache is searched for the target's entry. If no entry 84 exists, the advertisement SHOULD be silently discarded. There is no 85 need to create an entry if none exists, since the recipient has 86 apparently not initiated any communication with the target." 88 This approach is perfectly suitable for host-to-host communications, 89 which are in most cases bi-directional, and it could be expected that 90 if a host A has an neighbor cache entry for the host B IPv6 address, 91 host B also has the corresponding entry for the host A address in its 92 cache. However when a host communicates to off-link destinations via 93 its first-hop router, that logic does not apply. The most typical 94 scenario when the problem may arise is a host joining the network, 95 forming a new address and using that address for accessing the 96 Internet: 98 1. A host joins the network and receives a Router Advertisement (RA) 99 packet from the first-hop router (either a periodic unsolicited 100 RA or a response to a Router Solicitation sent by the host). The 101 RA contains information the host needs to perform Stateless 102 Address Autoconfiguration ([RFC4862]) and to configure its 103 network stack. As in most cases the RA also contains the link- 104 layer address of the router, the host can populate its Neighbor 105 Cache with the router's link-local and link-layer addresses. 107 2. The host starts opening connections to off-link destinations. A 108 very common use case is a mobile device sending probes to detect 109 the Internet connectivity and/or the presence of a captive portal 110 on the network. To speed up that process many implementations 111 use Optimistic Duplicate Address Detection [RFC4429] which allows 112 them to send probes before the Duplicate Address Detection (DAD) 113 process is completed. At that moment the device neighbor cache 114 contains all information required to send those probes (such as 115 the default router link-local the link-layer addresses). The 116 router neighbor cache, however, might contain an entry for the 117 device link-local address (if the device has been performing the 118 address resolution for the router link-local address), but there 119 are no entries for the device global addresses. 121 3. Return traffic is received by the first-hop router. As the 122 router does not have any cache entry for the host global address 123 yet, the router starts the neighbor discovery process by creating 124 an INCOMPLETE cache entry and then sending a Neighbor Solictation 125 to the Solicited Node Multicast Address. Most router 126 implementations buffer only one data packet while resolving the 127 packet destination address, so it would drop all subsequent 128 packets for the host global address, until the address resolution 129 process is completed. 131 4. If the host sends multiple probes in parallel, it would consider 132 all but one of them failed. That leads to user-visible delay in 133 connecting to the network, especially if the host implements some 134 form of backoff mechanism and does not retransmit the probes as 135 soon as possible. 137 This scenario illustrates the problem occurring when the device 138 connects to the network for the first time or after a timeout long 139 enough for the device address to be removed from the router's 140 neighbor cache. However, the same sequence of events happen when the 141 host starts using a new global address previously unseen by the 142 router, such as a new privacy address [RFC4941] or if the router's 143 Neighbor Cache has been flushed. 145 While in dual-stack networks this problem might be hidden by Happy 146 Eyeballs [RFC8305] it manifests quite clearly in IPv6-only 147 environments, especially wireless ones, leading to poor user 148 experience and contributing to a negative perception of IPv6-only 149 solutions as unstable and non-deployable. 151 This document discusses the operational implications of not 152 proactively creating Neighbor Cache entries on first-hop routers and 153 summarizes various approaches to mitigate the problem. The document 154 provides an overview of the proposed solution which is fully 155 described in [I-D.ietf-6man-grand]. 157 1.1. Requirements Language 159 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 160 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 161 "OPTIONAL" in this document are to be interpreted as described in BCP 162 14 [RFC2119] [RFC8174] when, and only when, they appear in all 163 capitals, as shown here. 165 1.2. Terminology 167 ND: Neighbor Discovery, [RFC4861]. 169 SLAAC: IPv6 Stateless Address Autoconfiguration, [RFC4862]. 171 NS: Neighbor Solicitation, [RFC4861]. 173 NA: Neighbor Advertisement, [RFC4861]. 175 RS: Router Solicitation, [RFC4861]. 177 RA: Router Advertisement, [RFC4861]. 179 SLLAO: Source link-layer Address Option, an option in the ND packets 180 containing the link-layer address of the sender of the packet, 181 [RFC4861]. 183 GUA: Global Unicast Address, [RFC4291]. 185 DAD: Duplicate Address Detection, [RFC4862]. 187 Optimistic DAD: a modification of DAD, [RFC4429]. 189 FCFS SAVI: First-Come, First-Served Source Address Validation, 190 [RFC6620]. 192 2. Proposed Solution 194 2.1. Solution Requirements 196 It would be highly desirable to improve the Neighbor Discovery 197 mechanics so routers have a usable cache entry for a host address by 198 the time the router receives the first packet for that address. In 199 particular: 201 o If the router does not have a Neighbor Cache entry for the 202 address, a STALE entry needs to be created. 204 o The solution needs to work for Optimistic addresses as well. 205 Devices implementing the Optimistic DAD usually attempt to 206 minimize the delay in connecting to the network and therefore are 207 more likely to be affected by the problem described in this 208 document. 210 o In case of duplicate addresses present in the network, the 211 proposed solution MUST NOT override the existing entry. 213 o In topologies with multiple first-hop routers the cache needs to 214 be updated on all of them, as traffic might be asymmetric: 215 outgoing flows leaving the network via one router while the return 216 traffic enters the segment via another one. 218 In addition the solution MUST NOT exacerbate issues described in 219 [RFC6583] and MUST be compatible with the recommendations provided in 220 [RFC6583]. 222 2.2. Solution Overview 224 The Neighbor Discovery is designed to allow IPv6 nodes to discover 225 neighboring nodes' reachability and learn IPv6 to link-layer 226 addresses mapping. Therefore ND seems to be the most appropriate 227 tool to inform the first-hop routers about addresses the host is 228 going to use. 230 Section 4.4 of [RFC4861] says: 232 "A node sends Neighbor Advertisements in response to Neighbor 233 Solicitations and sends unsolicited Neighbor Advertisements in order 234 to (unreliably) propagate new information quickly." 236 Propagating information about new GUA as quickly as possible is 237 exactly what is required to solve the problem outlined in this 238 document. Therefore the host might send an unsolicited NA with the 239 target link-layer address option to advertise its GUA as soon as the 240 said address enters Optimistic or Preferred state. 242 The proposed solution is discussed in [I-D.ietf-6man-grand]. In 243 summary, the following changes to [RFC4861] are suggested: 245 o A node SHOULD send up to MAX_NEIGHBOR_ADVERTISEMENT unsolicited NA 246 packets with the Override flag cleared to all-routers multicast 247 address (ff02::2) as soon as one of the following events happens: 249 * (if Optimistic DAD is used): a new Optimistic address is 250 assigned to the node interface. 252 * (if Optimistic DAD is not used): an address changes the state 253 from tentative to preferred. 255 o Routers SHOULD create a new STALE ND cache entry upon receiving 256 unsolicited NAs. 258 It should be noted that some routing and switching platforms have 259 implemented such behaviour already. Administrators could enable the 260 creation of neighbor discovery cache entries based on unsolicited NA 261 packets sent from the previously unknown neighbors on that interface. 263 Network devices implementing FCFS SAVI might drop Neighbor 264 Advertisements received through a Validating Port which is in the 265 TENTATIVE state (see Section 2.3.2 of[RFC6620]). Therefore hosts 266 using Optimistic DAD might not benefit from the proposed solution if 267 FCFS SAVI is implemented on the network infrastructure. 268 [I-D.ietf-6man-grand] discusses in more details how the proposed 269 solution interacts with SAVI. 271 3. Solutions Considered but Discarded 273 The problem could be addressed from different angles. Possible 274 approaches are: 276 o Just do nothing. 278 o Migrate from the "reactive" Neighbor Discovery ([RFC4861]) to the 279 registration-based mechanisms ([RFC8505]). 281 o The router creates new entries in its Neighbor Cache by gleaning 282 from Neighbor Discovery DAD messages. 284 o The host initiates bidirectional communication to the router using 285 the host GUA. 287 o Making the probing logic on hosts more robust. 289 o Increasing the buffer size on routers. 291 o Transit dataplane traffic from an unknown address (an address w/o 292 the corresponding neighbor cache entry) triggers an address 293 resolution process on the router. 295 It should be noted that some of those options are already implemented 296 by some vendors. The following sections discuss those approaches and 297 the reasons they were discarded. 299 3.1. Do Nothing 301 One of the possible approaches might be to declare that everything is 302 working as intended and let the upper-layer protocols to deal with 303 packet loss. The obvious drawbacks include: 305 o Unhappy users. 307 o Many support tickets. 309 o More resistance to deploy IPv6 and IPv6-Only networks. 311 3.2. Change to the Registration-Based Neighbor Discovery 313 The most radical approach would be to move away from the reactive ND 314 as defined in [RFC4861] and expand the registration-based ND 315 ([RFC6775], [RFC8505]) used in Low-Power Wireless Personal Area 316 Networks (6LoWPANs) to the rest of IPv6 deployments. This option 317 requires some investigation and discussions and seems to be excessive 318 for the problem described in this document. 320 3.3. Host Sending NS to the Router Address from Its GUA 322 The host could force creating a STALE entry for its GUA in the router 323 ND cache by sending the following Neighbor Solicitation message: 325 o The NS source address is the host GUA. 327 o The destination address is the default router IPv6 address. 329 o The Source Link-Layer Address option contains the host link-layer 330 address. 332 o The target address is the host default router address (the default 333 router address the host received in the RA). 335 The main disadvantages of this approach are: 337 o Would not work for Optimistic addresses as section 2.2 of 338 [RFC4429] explicitly prohibits sending Neighbor Solicitations from 339 an Optimistic Address. 341 o If first-hop redundancy is deployed in the network, the NS would 342 reach the active router only, so all backup routers (or all active 343 routers except one) would not get their neighbor cache updated. 345 o Some wireless devices are known to alter ND packets and perform 346 various non-obvious forms of ND proxy actions. In some cases, 347 unsolicited NAs might not even reach the routers. 349 3.4. Host Sending Router Solicitation from its GUA 351 The host could send a router solicitation message to 'all routers' 352 multicast address, using its GUA as a source. If the host link-layer 353 address is included in the Source Link-Layer Address option, the 354 router would create a STALE entry for the host GUA as per the section 355 6.2.6 of [RFC4861]. However, this approach can not be used if the 356 GUA is in optimistic state: section 2.2 of [RFC4429] explicitly 357 prohibits using an Optimistic Address as the source address of a 358 Router Solicitation with a SLLAO as it might disrupt the rightful 359 owner of the address in the case of a collision. So for the 360 optimistic addresses the host can send an RS without SLLAO included. 361 In that case the router may respond with either a multicast or a 362 unicast RA (only the latter would create a cache entry). 364 This approach has the following drawbacks: 366 o If the address is in the Optimistic state the RS can not contain 367 SLLAO. As a result the router would only create a cache entry if 368 the solicited RAs is sent as as a unicast. Routers sending 369 solicited RAs as multicast would not create a new cache entry as 370 they do not need to send a unicast packet back to the host. 372 o There might be a random delay between receiving an RS and sending 373 a unicast RA back (and creating a cache entry) which might 374 undermine the idea of creating the cache entry proactively. 376 o Some wireless devices are known to intercept ND packets and 377 perform various non-obvious forms of ND proxy actions. In some 378 cases the RS might not even reach the routers. 380 3.5. Routers Populating Their Caches by Gleaning From Neighbor 381 Discovery Packets 383 Routers may be able to learn about new addresses by gleaning from the 384 DAD Neighbor Solicitation messages. The router could listen to all 385 solicited node multicast address groups and upon receiving a Neighbor 386 Solicitation from the unspecified address search its Neighbor Cache 387 for the solicitation's Target Address. If no entry exists, the 388 router may create an entry, set its reachability state to 389 'INCOMPLETE' and start the address resolution for that entry. 391 The same solution was proposed in 392 [I-D.halpern-6man-nd-pre-resolve-addr]. Some routing vendors support 393 such optimization already. However, this approach has a number of 394 drawbacks and therefore should not be used as the only solution: 396 o Routers need to receive all multicast Neighbor Discovery packets 397 which might negatively impact the routers CPU. 399 o If the router starts the address resolution as soon as it receives 400 the DAD Neighbor Solicitation the host might be still performing 401 DAD and the target address might be tentative. In that case, the 402 host SHOULD silently ignore the received Neighbor Solicitation 403 from the router as per the Section 5.4.3 of [RFC4862]. As a 404 result the router might not be able to complete the address 405 resolution before the return traffic arrives. 407 3.6. Initiating Hosts-to-Routers Communication 409 The host may force the router to start address resolution by sending 410 a data packet such as ping or traceroute to its default router link- 411 local address, using the GUA as a source address. As the RTT to the 412 default router is lower than RTT to any off-link destinations it's 413 quite likely that the router would start the neighbor discovery 414 process for the host GUA before the first packet of the returning 415 traffic arrives. 417 This approach has the following drawbacks: 419 o Data packets to the router link-local address could be blocked by 420 security policy or control plane protection mechanism. 422 o It introduces an additional overhead for routers control plane (in 423 addition to processing ND packets, the data packet needs to be 424 processed as well). 426 o Unless the data packet is sent to 'all routers' ff02::2 multicast 427 address, if the network provides a first-hop redundancy then only 428 the active router would create a new cache entry. 430 3.7. Transit Dataplane Traffic From a New Address Triggering Address 431 Resolution 433 When a router receives a transit packet, it might check the presence 434 of the neighbor cache entry for the packet source address and if the 435 entry does not exist start address resolution process. This approach 436 does ensure that a Neighbor Cache entry is proactively created every 437 time a new, previously unseen GUA is used for sending offlink 438 traffic. However this approach has a number of limitations, in 439 particular: 441 o If traffic flows are asymmetrical the return traffic might not 442 transit the same router as the original traffic which triggered 443 the address resolution. So the neighbor cache entry is created on 444 the "wrong" router, not the one which actually needs the neighbor 445 cache entry for the host address. 447 o The functionality needs to be limited to explicitly configured 448 networks/interfaces, as the router needs to distinguish between 449 onlink addresses (ones the router needs to have Neighbor Cache 450 entries for) and the rest of the address space. 452 o Implementing such functionality is much more complicated than all 453 other solutions as it would involve complex data-control planes 454 interaction. 456 4. IANA Considerations 458 This memo asks the IANA for no new parameters. 460 5. Security Considerations 462 This memo documents the operational issue and does not introduce any 463 new security considerations. Security considerations of the proposed 464 solution are discussed in the corresponding section of 465 [I-D.ietf-6man-grand]. 467 6. Acknowledgements 469 Thanks to the following people (in alphabetical order) for their 470 review and feedback: Mikael Abrahamsson, Stewart Bryant, Lorenzo 471 Colitti, Owen DeLong, Igor Gashinsky, Fernando Gont, Tatuya Jinmei, 472 Erik Kline, Warren Kumari, Barry Leiba, Jordi Palet Martinez, Michael 473 Richardson, Dave Thaler, Pascal Thubert, Loganaden Velvindron, Eric 474 Vyncke. 476 7. References 478 7.1. Normative References 480 [I-D.ietf-6man-grand] 481 Linkova, J., "Gratuitous Neighbor Discovery: Creating 482 Neighbor Cache Entries on First-Hop Routers", draft-ietf- 483 6man-grand-01 (work in progress), July 2020. 485 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 486 Requirement Levels", BCP 14, RFC 2119, 487 DOI 10.17487/RFC2119, March 1997, 488 . 490 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 491 Architecture", RFC 4291, DOI 10.17487/RFC4291, February 492 2006, . 494 [RFC4429] Moore, N., "Optimistic Duplicate Address Detection (DAD) 495 for IPv6", RFC 4429, DOI 10.17487/RFC4429, April 2006, 496 . 498 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 499 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 500 DOI 10.17487/RFC4861, September 2007, 501 . 503 [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless 504 Address Autoconfiguration", RFC 4862, 505 DOI 10.17487/RFC4862, September 2007, 506 . 508 [RFC6583] Gashinsky, I., Jaeggli, J., and W. Kumari, "Operational 509 Neighbor Discovery Problems", RFC 6583, 510 DOI 10.17487/RFC6583, March 2012, 511 . 513 [RFC6620] Nordmark, E., Bagnulo, M., and E. Levy-Abegnoli, "FCFS 514 SAVI: First-Come, First-Served Source Address Validation 515 Improvement for Locally Assigned IPv6 Addresses", 516 RFC 6620, DOI 10.17487/RFC6620, May 2012, 517 . 519 [RFC6775] Shelby, Z., Ed., Chakrabarti, S., Nordmark, E., and C. 520 Bormann, "Neighbor Discovery Optimization for IPv6 over 521 Low-Power Wireless Personal Area Networks (6LoWPANs)", 522 RFC 6775, DOI 10.17487/RFC6775, November 2012, 523 . 525 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 526 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 527 May 2017, . 529 [RFC8305] Schinazi, D. and T. Pauly, "Happy Eyeballs Version 2: 530 Better Connectivity Using Concurrency", RFC 8305, 531 DOI 10.17487/RFC8305, December 2017, 532 . 534 [RFC8505] Thubert, P., Ed., Nordmark, E., Chakrabarti, S., and C. 535 Perkins, "Registration Extensions for IPv6 over Low-Power 536 Wireless Personal Area Network (6LoWPAN) Neighbor 537 Discovery", RFC 8505, DOI 10.17487/RFC8505, November 2018, 538 . 540 7.2. Informative References 542 [I-D.halpern-6man-nd-pre-resolve-addr] 543 Chen, I. and J. Halpern, "Triggering ND Address Resolution 544 on Receiving DAD-NS", draft-halpern-6man-nd-pre-resolve- 545 addr-00 (work in progress), January 2014. 547 [RFC4941] Narten, T., Draves, R., and S. Krishnan, "Privacy 548 Extensions for Stateless Address Autoconfiguration in 549 IPv6", RFC 4941, DOI 10.17487/RFC4941, September 2007, 550 . 552 Author's Address 554 Jen Linkova 555 Google 556 1 Darling Island Rd 557 Pyrmont, NSW 2009 558 AU 560 Email: furry@google.com