idnits 2.17.1 draft-malhotra-bess-evpn-irb-extended-mobility-01.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 (October 26, 2017) is 2371 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: 'RFC2119' is mentioned on line 183, but not defined == Missing Reference: 'VM-IP1' is mentioned on line 290, but not defined == Missing Reference: 'VM-IP2' is mentioned on line 290, but not defined == Missing Reference: 'VM-IP3' is mentioned on line 290, but not defined == Missing Reference: 'VM-IP4' is mentioned on line 290, but not defined == Missing Reference: 'VM-IP5' is mentioned on line 290, but not defined == Missing Reference: 'VM-IP6' is mentioned on line 290, but not defined == Missing Reference: 'IP1' is mentioned on line 364, but not defined == Missing Reference: 'MAC1' is mentioned on line 295, but not defined == Missing Reference: 'GW1' is mentioned on line 343, but not defined == Missing Reference: 'GW2' is mentioned on line 343, but not defined == Missing Reference: 'GW3' is mentioned on line 420, but not defined == Missing Reference: 'GW4' is mentioned on line 420, but not defined == Missing Reference: 'MAC2' is mentioned on line 299, but not defined == Missing Reference: 'VM-IP1-M1' is mentioned on line 341, but not defined == Missing Reference: 'VM-IP2-M2' is mentioned on line 341, but not defined == Missing Reference: 'VM-IP3-M3' is mentioned on line 341, but not defined == Missing Reference: 'VM-IP4-M4' is mentioned on line 341, but not defined == Missing Reference: 'VM-IP5-M5' is mentioned on line 341, but not defined == Missing Reference: 'VM-IP6-M6' is mentioned on line 341, but not defined == Missing Reference: 'IP7' is mentioned on line 362, but not defined == Missing Reference: 'M1' is mentioned on line 364, but not defined == Missing Reference: 'RFC 7814' is mentioned on line 724, but not defined == Missing Reference: 'RFC 7432' is mentioned on line 893, but not defined == Unused Reference: 'RFC7814' is defined on line 950, but no explicit reference was found in the text == Outdated reference: A later version (-16) exists of draft-ietf-bess-evpn-proxy-arp-nd-02 == Outdated reference: A later version (-15) exists of draft-ietf-bess-evpn-inter-subnet-forwarding-03 ** Downref: Normative reference to an Informational RFC: RFC 7814 Summary: 1 error (**), 0 flaws (~~), 28 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 INTERNET-DRAFT N. Malhotra, Ed. 3 A. Sajassi 4 A. Pattekar 5 Intended Status: Proposed Standard (Cisco) 6 A. Lingala 7 (AT&T) 8 J. Rabadan 9 (Nokia) 10 J. Drake 11 (Juniper Networks) 13 Expires: April 29, 2018 October 26, 2017 15 Extended Mobility Procedures for EVPN-IRB 16 draft-malhotra-bess-evpn-irb-extended-mobility-01 18 Abstract 20 The procedure to handle host mobility in a layer 2 Network with EVPN 21 control plane is defined as part of RFC 7432. EVPN has since evolved 22 to find wider applicability across various IRB use cases that include 23 distributing both MAC and IP reachability via a common EVPN control 24 plane. MAC Mobility procedures defined in RFC 7432 are extensible to 25 IRB use cases if a fixed 1:1 mapping between VM IP and MAC is assumed 26 across VM moves. Generic mobility support for IP and MAC that allows 27 these bindings to change across moves is required to support a 28 broader set of EVPN IRB use cases, and requires further 29 consideration. EVPN all-active multi-homing further introduces 30 scenarios that require additional consideration from mobility 31 perspective. Intent of this draft is to enumerate a set of design 32 considerations applicable to mobility across EVPN IRB use cases and 33 define generic sequence number assignment procedures to address these 34 IRB use cases. 36 Status of this Memo 38 This Internet-Draft is submitted to IETF in full conformance with the 39 provisions of BCP 78 and BCP 79. 41 Internet-Drafts are working documents of the Internet Engineering 42 Task Force (IETF), its areas, and its working groups. Note that 43 other groups may also distribute working documents as 44 Internet-Drafts. 46 Internet-Drafts are draft documents valid for a maximum of six months 47 and may be updated, replaced, or obsoleted by other documents at any 48 time. It is inappropriate to use Internet-Drafts as reference 49 material or to cite them other than as "work in progress." 51 The list of current Internet-Drafts can be accessed at 52 http://www.ietf.org/1id-abstracts.html 54 The list of Internet-Draft Shadow Directories can be accessed at 55 http://www.ietf.org/shadow.html 57 Copyright and License Notice 59 Copyright (c) 2017 IETF Trust and the persons identified as the 60 document authors. All rights reserved. 62 This document is subject to BCP 78 and the IETF Trust's Legal 63 Provisions Relating to IETF Documents 64 (http://trustee.ietf.org/license-info) in effect on the date of 65 publication of this document. Please review these documents 66 carefully, as they describe your rights and restrictions with respect 67 to this document. Code Components extracted from this document must 68 include Simplified BSD License text as described in Section 4.e of 69 the Trust Legal Provisions and are provided without warranty as 70 described in the Simplified BSD License. 72 Table of Contents 74 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 75 1.1 Terminology . . . . . . . . . . . . . . . . . . . . . . . . 5 76 2. Optional MAC only RT-2 . . . . . . . . . . . . . . . . . . . . 5 77 3. Mobility Use Cases . . . . . . . . . . . . . . . . . . . . . . 6 78 3.1 VM MAC+IP Move . . . . . . . . . . . . . . . . . . . . . . 6 79 3.2 VM IP Move to new MAC . . . . . . . . . . . . . . . . . . . 6 80 3.2.1 VM Reload . . . . . . . . . . . . . . . . . . . . . . . 6 81 3.2.2 MAC Sharing . . . . . . . . . . . . . . . . . . . . . . 6 82 3.2.3 Problem . . . . . . . . . . . . . . . . . . . . . . . . 7 83 3.3 VM MAC move to new IP . . . . . . . . . . . . . . . . . . . 8 84 3.3.1 Problem . . . . . . . . . . . . . . . . . . . . . . . . 8 85 4. EVPN All Active multi-homed ES . . . . . . . . . . . . . . . . 10 86 5. Design Considerations . . . . . . . . . . . . . . . . . . . . 11 87 6. Solution Components . . . . . . . . . . . . . . . . . . . . . 12 88 6.1 Sequence Number Inheritance . . . . . . . . . . . . . . . . 12 89 6.2 MAC Sharing . . . . . . . . . . . . . . . . . . . . . . . . 13 90 6.3 Multi-homing Mobility Synchronization . . . . . . . . . . . 14 91 7. Requirements for Sequence Number Assignment . . . . . . . . . 14 92 7.1 LOCAL MAC-IP learning . . . . . . . . . . . . . . . . . . . 14 93 7.2 LOCAL MAC learning . . . . . . . . . . . . . . . . . . . . 15 94 7.3 Remote MAC OR MAC-IP Update . . . . . . . . . . . . . . . . 15 95 7.4 REMOTE (SYNC) MAC update . . . . . . . . . . . . . . . . . 15 96 7.5 REMOTE (SYNC) MAC-IP update . . . . . . . . . . . . . . . . 16 97 7.6 Inter-op . . . . . . . . . . . . . . . . . . . . . . . . . 16 98 8. Routed Overlay . . . . . . . . . . . . . . . . . . . . . . . . 16 99 9. Duplicate Host Detection . . . . . . . . . . . . . . . . . . . 18 100 9.1 Scenario A . . . . . . . . . . . . . . . . . . . . . . . . . 18 101 9.2 Scenario B . . . . . . . . . . . . . . . . . . . . . . . . . 18 102 9.2.1 Duplicate IP Detection Procedure for Scenario B . . . . 19 103 9.3 Scenario C . . . . . . . . . . . . . . . . . . . . . . . . . 19 104 9.4 Duplicate Host Recovery . . . . . . . . . . . . . . . . . . 20 105 9.4.1 Route Un-freezing Configuration . . . . . . . . . . . . 20 106 9.4.2 Route Clearing Configuration . . . . . . . . . . . . . 21 107 10. Security Considerations . . . . . . . . . . . . . . . . . . . 21 108 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 21 109 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 21 110 12.1 Normative References . . . . . . . . . . . . . . . . . . . 21 111 12.2 Informative References . . . . . . . . . . . . . . . . . . 22 112 13. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 22 113 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 22 114 Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 116 1 Introduction 118 EVPN-IRB enables capability to advertise both MAC and IP routes via a 119 single MAC+IP RT-2 advertisement. MAC is imported into local bridge 120 MAC table and enables L2 bridged traffic across the network overlay. 121 IP is imported into the local ARP table in an asymmetric IRB design 122 OR imported into the IP routing table in a symmetric IRB design, and 123 enables routed traffic across the layer 2 network overlay. Please 124 refer to [EVPN-INTER-SUBNET] more background on EVPN IRB forwarding 125 modes. 127 To support EVPN mobility procedure, a single sequence number mobility 128 attribute is advertised with the combined MAC+IP route. A single 129 sequence number advertised with the combined MAC+IP route to resolve 130 both MAC and IP reachability implicitly assumes a 1:1 fixed mapping 131 between IP and MAC. While a fixed 1:1 mapping between IP and MAC is a 132 common use case that could be addressed via existing MAC mobility 133 procedure, additional IRB scenarios need to be considered, that don't 134 necessarily adhere to this assumption. Following IRB mobility 135 scenarios are considered: 137 o VM move results in VM IP and MAC moving together 139 o VM move results in VM IP moving to a new MAC association 141 o VM move results in VM MAC moving to a new IP association 143 While existing MAC mobility procedure can be leveraged for MAC+IP 144 move in the first scenario, subsequent scenarios result in a new MAC- 145 IP association. As a result, a single sequence number assigned 146 independently per-[MAC, IP] is not sufficient to determine most 147 recent reachability for both MAC and IP, unless the sequence number 148 assignment algorithm is designed to allow for changing MAC-IP 149 bindings across moves. 151 Purpose of this draft is to define additional sequence number 152 assignment and handling procedures to adequately address generic 153 mobility support across EVPN-IRB overlay use cases that allow MAC-IP 154 bindings to change across VM moves and can support mobility for both 155 MAC and IP components carried in an EVPN RT-2 for these use cases. 157 In addition, for hosts on an ESI multi-homed to multiple GW devices, 158 additional procedure is proposed to ensure synchronized sequence 159 number assignments across the multi-homing devices. 161 Content presented in this draft is independent of data plane 162 encapsulation used in the overlay being MPLS or NVO Tunnels. It is 163 also largely independent of the EVPN IRB solution being based on 164 symmetric OR asymmetric IRB design as defined in [EVPN-INTER-SUBNET]. 165 In addition to symmetric and asymmetric IRB, mobility solution for a 166 routed overlay, where traffic to an end host in the overlay is always 167 IP routed using EVPN RT-5 is also presented in section 8. 169 To summarize, this draft covers mobility mobility for the following 170 independent of the overlay encapsulation being MPLS or an NVO Tunnel: 172 o Symmetric EVPN IRB overlay 174 o Asymmetric EVPN IRB overlay 176 o Routed EVPN overlay 178 1.1 Terminology 180 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 181 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 182 document are to be interpreted as described in RFC 2119 [RFC2119]. 184 o ARP is widely referred to in this document. This is simply for 185 ease of reading, and as such, these references are equally 186 applicable to ND (neighbor discovery) as well. 188 o GW: used widely in the document refers to an IRB GW that is 189 doing routing and bridging between an access network and an EVPN 190 enabled overlay network. 192 o RT-2: EVPN route type 2 carrying both MAC and IP reachability 194 o RT-5: EVPN route type 5 carrying IP prefix reachability 196 o ES: EVPN Ethernet Segment 198 o MAC-IP: IP association for a MAC, referred to in this document 199 may be IPv4, IPv6 or both. 201 2. Optional MAC only RT-2 203 In an EVPN IRB scenario, where a single MAC+IP RT-2 advertisement 204 carries both IP and MAC routes, a MAC only RT-2 advertisement is 205 redundant for host MACs that are advertised via MAC+IP RT-2. As a 206 result, a MAC only RT-2 is an optional route that may not be 207 advertised from or received at an IRB GW. This is an important 208 consideration for mobility scenarios discussed in subsequent 209 sections. 211 MAC only RT-2 may still be advertised for non-IP host MACs that are 212 not advertised via MAC+IP RT-2. 214 3. Mobility Use Cases 216 This section describes the IRB mobility use cases considered in this 217 document. Procedures to address them are covered later in section 6 218 and section 7. 220 o VM move results in VM IP and MAC moving together 222 o VM move results in VM IP moving to a new MAC association 224 o VM move results in VM MAC moving to a new IP association 226 3.1 VM MAC+IP Move 228 This is the baseline case, wherein a VM move results in both VM MAC 229 and IP moving together with no change in MAC-IP binding across a 230 move. Existing MAC mobility defined in RFC 7432 may be leveraged to 231 apply to corresponding MAC+IP route to support this mobility 232 scenario. 234 3.2 VM IP Move to new MAC 236 This is the case, where a VM move results in VM IP moving to a new 237 MAC binding. 239 3.2.1 VM Reload 241 A VM reload or an orchestrated VM move that results in VM being re- 242 spawned at a new location may result in VM getting a new MAC 243 assignment, while maintaining existing IP address. This results in a 244 VM IP move to a new MAC binding: 246 IP-a, MAC-a ---> IP-a, MAC-b 248 3.2.2 MAC Sharing 250 This takes into account scenarios, where multiple hosts, each with a 251 unique IP, may share a common MAC binding, and a host move results in 252 a new MAC binding for the host IP. 254 As an example, host VMs running on a single physical server, each 255 with a unique IP, may share the same physical server MAC. In yet 256 another scenario, an L2 access network may be behind a firewall, such 257 that all hosts IPs on the access network are learnt with a common 258 firewall MAC. In all such "shared MAC" use cases, multiple local MAC- 259 IP ARP entries may be learnt with the same MAC. A VM IP move, in such 260 scenarios (for e.g., to a new physical server), could result in new 261 MAC association for the VM IP. 263 3.2.3 Problem 265 In both of the above scenarios, a combined MAC+IP EVPN RT-2 266 advertised with a single sequence number attribute implicitly assumes 267 a fixed IP to MAC mapping. A host IP move to a new MAC breaks this 268 assumption and results in a new MAC+IP route. If this new MAC+IP 269 route is independently assigned a new sequence number, the sequence 270 number can no longer be used to determine most recent host IP 271 reachability in a symmetric EVPN-IRB design OR the most recent IP to 272 MAC binding in an asymmetric EVPN-IRB design. 274 +------------------------+ 275 | Underlay Network Fabric| 276 +------------------------+ 278 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 279 | GW1 | | GW2 | | GW3 | | GW4 | | GW5 | | GW6 | 280 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 281 \ / \ / \ / 282 \ ESI-1 / \ ESI-2 / \ ESI-3 / 283 \ / \ / \ / 284 +\---/+ +\---/+ +\---/+ 285 | \ / | | \ / | | \ / | 286 +--+--+ +--+--+ +--+--+ 287 | | | 288 Server-MAC1 Server-MAC2 Server-MAC3 289 | | | 290 [VM-IP1, VM-IP2] [VM-IP3, VM-IP4] [VM-IP5, VM-IP6] 292 Figure 1 294 As an example, consider a topology shown in Figure 1, with host VMs 295 sharing the physical server MAC. In steady state, [IP1, MAC1] route 296 is learnt at [GW1, GW2] and advertised to remote GWs with a sequence 297 number N. Now, VM-IP1 is moved to Server-MAC2. ARP or ND based local 298 learning at [GW3, GW4] would now result in a new [IP1, MAC2] route 299 being learnt. If route [IP1, MAC2] is learnt as a new MAC+IP route 300 and assigned a new sequence number of say 0, mobility procedure for 301 VM-IP1 will not trigger across the overlay network. 303 A clear sequence number assignment procedure needs to be defined to 304 unambiguously determine the most recent IP reachability, IP to MAC 305 binding, and MAC reachability for such a MAC sharing scenario. 307 3.3 VM MAC move to new IP 309 This is a scenario where host move or re-provisioning behind a new 310 gateway location may result in the same VM MAC getting a new IP 311 address assigned. 313 3.3.1 Problem 315 Complication with this scenario is that MAC reachability could be 316 carried via a combined MAC+IP route while a MAC only route may not be 317 advertised at all. A single sequence number association with the 318 MAC+IP route again implicitly assumes a fixed mapping between MAC and 319 IP. A MAC move resulting in a new IP association for the host MAC 320 breaks this assumption and results in a new MAC+IP route. If this new 321 MAC+IP route independently assumes a new sequence number, this 322 mobility attribute can no longer be used to determine most recent 323 host MAC reachability as opposed to the older existing MAC 324 reachability. 326 +------------------------+ 327 | Underlay Network Fabric| 328 +------------------------+ 329 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 330 | GW1 | | GW2 | | GW3 | | GW4 | | GW5 | | GW6 | 331 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 332 \ / \ / \ / 333 \ ESI-1 / \ ESI-2 / \ ESI-3 / 334 \ / \ / \ / 335 +\---/+ +\---/+ +\---/+ 336 | \ / | | \ / | | \ / | 337 +--+--+ +--+--+ +--+--+ 338 | | | 339 Server1 Server2 Server3 340 | | | 341 [VM-IP1-M1, VM-IP2-M2] [VM-IP3-M3, VM-IP4-M4] [VM-IP5-M5, VM-IP6-M6] 343 As an example, IP1-M1 is learnt locally at [GW1, GW2] and currently 344 advertised to remote hosts with a sequence number N. Consider a 345 scenario where a VM with MAC M1 is re-provisioned at server 2, 346 however, as part of this re-provisioning, assigned a different IP 347 address say IP7. [IP7, M1] is learnt as a new route at [GW3, GW4] and 348 advertised to remote GWs with a sequence number of 0. As a result, L3 349 reachability to IP7 would be established across the overlay, however, 350 MAC mobility procedure for MAC1 will not trigger as a result of this 351 MAC-IP route advertisement. If an optional MAC only route is also 352 advertised, sequence number associated with the MAC only route would 353 trigger MAC mobility as per [RFC7432]. However, in the absence of an 354 additional MAC only route advertisement, a single sequence number 355 advertised with a combined MAC+IP route would not be sufficient to 356 update MAC reachability across the overlay. 358 A MAC-IP sequence number assignment procedure needs to be defined to 359 unambiguously determine the most recent MAC reachability in such a 360 scenario without a MAC only route being advertised. 362 Further, GW1/GW2, on learning new reachability for [IP7, M1] via 363 GW3/GW4 MUST probe and delete any local IPs associated with MAC M1, 364 such as [IP1, M1] in the above example. 366 Arguably, MAC mobility sequence number defined in [RFC7432], could be 367 interpreted to apply only to the MAC part of MAC-IP route, and would 368 hence cover this scenario. It could hence be interpreted as a 369 clarification to [RFC7432] and one of the considerations for a common 370 sequence number assignment procedure across all MAC-IP mobility 371 scenarios detailed in this document. 373 4. EVPN All Active multi-homed ES 375 +------------------------+ 376 | Underlay Network Fabric| 377 +------------------------+ 379 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 380 | GW1 | | GW2 | | GW3 | | GW4 | | GW5 | | GW6 | 381 +-----+ +-----+ +-----+ +-----+ +-----+ +-----+ 382 \ / \ / \ / 383 \ ESI-1 / \ ESI-2 / \ ESI-3 / 384 \ / \ / \ / 385 +\---/+ +\---/+ +\---/+ 386 | \ / | | \ / | | \ / | 387 +--+--+ +--+--+ +--+--+ 388 | | | 389 Server-1 Server-2 Server-3 391 Figure 2 393 Consider an EVPN-IRB overlay network shown in Figure 2, with hosts 394 multi-homed to two or more leaf GW devices via an all-active multi- 395 homed ES. MAC and ARP entries learnt on a local ESI may also be 396 synchronized across the multi-homing GW devices sharing this ESI. 397 This MAC and ARP SYNC enables local switching of intra and inter 398 subnet ECMP traffic flows from remote hosts. In other words, local 399 MAC and ARP entries on a given Ethernet segment (ES) may be learnt 400 via local learning and / or sync from another GW device sharing the 401 same ES. 403 For a host that is multi-homed to multiple GW devices via an all- 404 active ES interface, local learning of host MAC and MAC-IP at each GW 405 device is an independent asynchronous event, that is dependent on 406 traffic flow and or ARP / ND response from the host hashing to a 407 directly connected GW on the MC-LAG interface. As a result, sequence 408 number mobility attribute value assigned to a locally learnt MAC or 409 MAC-IP route (as per RFC 7432) at each device may not always be the 410 same, depending on transient states on the device at the time of 411 local learning. 413 As an example, consider a host VM that is deleted from ESI-2 and 414 moved to ESI-1. It is possible for host to be learnt on say, GW1 415 following deletion of the remote route from [GW3, GW4], while being 416 learnt on GW2 prior to deletion of remote route from [GW3, GW4]. If 417 so, GW1 would process local host route learning as a new route and 418 assign a sequence number of 0, while GW2 would process local host 419 route learning as a remote to local move and assign a sequence number 420 of N+1, N being the existing sequence number assigned at [GW3, GW4]. 421 Inconsistent sequence numbers advertised from multi-homing devices 422 introduces ambiguity with respect to sequence number based mobility 423 procedures across the overlay. 425 o Ambiguity with respect to how the remote ToRs should handle 426 paths with same ESI and different sequence numbers. A remote ToR 427 may not program ECMP paths if it receives routes with different 428 sequence numbers from a set of multi-homing GWs sharing the same 429 ESI. 431 o Breaks consistent route versioning across the network overlay 432 that is needed for EVPN mobility procedures to work. 434 As an example, in this inconsistent state, GW2 would drop a remote 435 route received for the same host with sequence number N (as its local 436 sequence number is N+1), while GW1 would install it as the best route 437 (as its local sequence number is 0). 439 There is need for a mechanism to ensure consistency of sequence 440 numbers advertised from a set of multi-homing devices for EVPN 441 mobility to work reliably. 443 In order to support mobility for multi-homed hosts using the sequence 444 number mobility attribute, local MAC and MAC-IP routes MUST be 445 advertised with the same sequence number by all GW devices that the 446 ESI is multi-homed to. In other words, there is need for a mechanism 447 to ensure consistency of sequence numbers advertised from a set of 448 multi-homing devices for EVPN mobility to work reliably. 450 5. Design Considerations 452 To summarize, sequence number assignment scheme and implementation 453 must take following considerations into account: 455 o MAC+IP may be learnt on an ESI multi-homed to multiple GW 456 devices, hence requires sequence numbers to be synchronized 457 across multi-homing GW devices. 459 o MAC only RT-2 is optional in an IRB scenario and may not 460 necessarily be advertised in addition to MAC+IP RT-2 462 o Single MAC may be associated with multiple IPs, i.e., multiple 463 host IPs may share a common MAC 465 o Host IP move could result in host moving to a new MAC, resulting 466 in a new IP to MAC association and a new MAC+IP route. 468 o Host MAC move to a new location could result in host MAC being 469 associated with a different IP address, resulting in a new MAC to 470 IP association and a new MAC+IP route 472 o LOCAL MAC-IP learn via ARP would always accompanied by a LOCAL 473 MAC learn event resulting from the ARP packet. MAC and MAC-IP 474 learning, however, could happen in any order 476 o Use cases discussed earlier that do not maintain a constant 1:1 477 MAC-IP mapping across moves could potentially be addressed by 478 using separate sequence numbers associated with MAC and IP 479 components of MAC+IP route. Maintaining two separate sequence 480 numbers however adds significant overhead with respect to 481 complexity, debugability, and backward compatibility. It is 482 therefore goal of solution presented here to address these 483 requirements via a single sequence number attribute. 485 6. Solution Components 487 This section goes over main components of the EVPN IRB mobility 488 solution proposed in this draft. Later sections will go over exact 489 sequence number assignment procedures resulting from concepts 490 described in this section. 492 6.1 Sequence Number Inheritance 494 Main idea presented here is to view a LOCAL MAC-IP route as a child 495 of the corresponding LOCAL MAC only route that inherits the sequence 496 number attribute from the parent LOCAL MAC only route: 498 Mx-IPx -----> Mx (seq# = N) 500 As a result, both parent MAC and child MAC-IP routes share one common 501 sequence number associated with the parent MAC route. Doing so 502 ensures that a single sequence number attribute carried in a combined 503 MAC+IP route represents sequence number for both a MAC only route as 504 well as a MAC+IP route, and hence makes the MAC only route truly 505 optional. As a result, optional MAC only route with its own sequence 506 number is not required to establish most recent reachability for a 507 MAC in the overlay network. Specifically, this enables a MAC to 508 assume a different IP address on a move, and still be able to 509 establish most recent reachability to the MAC across the overlay 510 network via mobility attribute associated with the MAC+IP route 511 advertisement. As an example, when Mx moves to a new location, it 512 would result in LOCAL Mx being assigned a higher sequence number at 513 its new location as per RFC 7432. If this move results in Mx assuming 514 a different IP address, IPz, LOCAL Mx+IPz route would inherit the new 515 sequence number from Mx. 517 LOCAL MAC and LOCAL MAC-IP routes would typically be sourced from 518 data plane learning and ARP learning respectively, and could get 519 learnt in control plane in any order. Implementation could either 520 replicate inherited sequence number in each MAC-IP entry OR maintain 521 a single attribute in the parent MAC by creating a forward reference 522 LOCAL MAC object for cases where a LOCAL MAC-IP is learnt before the 523 LOCAL MAC. 525 Arguably, this inheritance may be assumed from RFC 7432, in which 526 case, the above may be interpreted as a clarification with respect to 527 interpretation of a MAC sequence number in a MAC-IP route. 529 6.2 MAC Sharing 531 Further, for the shared MAC scenario, this would result in multiple 532 LOCAL MAC-IP siblings inheriting sequence number attribute from a 533 common parent MAC route: 535 Mx-IP1 ----- 536 | | 537 Mx-IP2 ----- 538 . | 539 . +---> Mx (seq# = N) 540 . | 541 Mx-IPw ----- 542 | | 543 Mx-IPx ----- 545 In such a case, a host-IP move to a different physical server would 546 result in IP moving to a new MAC binding. A new MAC-IP route 547 resulting from this move must now be advertised with a sequence 548 number that is higher than the previous MAC-IP route for this IP, 549 advertised from the prior location. As an example, consider a route 550 Mx-IPx that is currently advertised with sequence number N from GW1. 551 IPx moving to a new physical server behind GW2 results in IPx being 552 associated with MAC Mz. A new local Mz-IPx route resulting from this 553 move at GW2 must now be advertised with a sequence number higher than 554 N. This is so that GW devices, including GW1, GW2, and other remote 555 GW devices that are part of the overlay can clearly determine and 556 program the most recent MAC binding and reachability for the IP. GW1, 557 on receiving this new Mz-IPx route with sequence number say, N+1, for 558 symmetric IRB case, would update IPx reachability via GW2 in 559 forwarding, for asymmetric IRB case, would update IPx's ARP binding 560 to Mz. In addition, GW1 would clear and withdraw the stale Mx-IPx 561 route with the lower sequence number. 563 This also implies that sequence number associated with local MAC Mz 564 and all local MAC-IP children of Mz at GW2 must now be incremented to 565 N+1, and re-advertised across the overlay. While this re- 566 advertisement of all local MAC-IP children routes affected by the 567 parent MAC route is an overhead, it avoids the need for two separate 568 sequence number attributes to be maintained and advertised for IP and 569 MAC components of MAC+IP RT-2. Implementation would need to be able 570 to lookup MAC-IP routes for a given IP and update sequence number for 571 it's parent MAC and its MAC-IP children. 573 6.3 Multi-homing Mobility Synchronization 575 In order to support mobility for multi-homed hosts, local MAC and 576 MAC-IP routes learnt on the shared ESI MUST be advertised with the 577 same sequence number by all GW devices that the ESI is multi-homed 578 to. This also applies to local MAC only routes. LOCAL MAC and MAC-IP 579 may be learnt natively via data plane and ARP/ND respectively as well 580 as via SYNC from another multi-homing GW to achieve local switching. 581 Local and SYNC route learning can happen in any order. Local MAC-IP 582 routes advertised by all multi-homing GW devices sharing the ESI must 583 carry the same sequence number, independent of the order in which 584 they are learnt. This implies: 586 o On local or sync MAC-IP route learning, sequence number for the 587 local MAC-IP route MUST be compared and updated to the higher 588 value. 590 o On local or sync MAC route learning, sequence number for the 591 local MAC route MUST be compared and updated to the higher value. 593 If an update to local MAC-IP sequence number is required as a result 594 of above comparison with sync MAC-IP route, it would essentially 595 amount to a sequence number update on the parent local MAC, resulting 596 in the inherited sequence number update on the MAC-IP route. 598 7. Requirements for Sequence Number Assignment 600 Following sections summarize sequence number assignment procedure 601 needed on local and sync MAC and MAC-IP route learning events in 602 order to accomplish the above. 604 7.1 LOCAL MAC-IP learning 606 A local Mx-IPx learning via ARP or ND should result in computation OR 607 re-computation of parent MAC Mx's sequence number, following which 608 the MAC-IP route Mx-IPx would simply inherit parent MAC's sequence 609 number. Parent MAC Mx Sequence number should be computed as follows: 611 o MUST be higher than any existing remote MAC route for Mx, as per 612 RFC 7432. 614 o MUST be at least equal to corresponding SYNC MAC sequence number 615 if one is present. 617 o If the IP is also associated with a different remote MAC "Mz", 618 MUST be higher than "Mz" sequence number 620 Once new sequence number for MAC route Mx is computed as per above, 621 all LOCAL MAC-IPs associated with MAC Mx MUST inherit the updated 622 sequence number. 624 7.2 LOCAL MAC learning 626 Local MAC Mx Sequence number should be computed as follows: 628 o MUST be higher than any existing remote MAC route for Mx, as per 629 RFC 7432. 631 o MUST be at least equal to corresponding SYNC MAC sequence number 632 if one is present. 634 o Once new sequence number for MAC route Mx is computed as per 635 above, all LOCAL MAC-IPs associated with MAC Mx MUST inherit the 636 updated sequence number. 638 Note that the local MAC sequence number might already be present if 639 there was a local MAC-IP learnt prior to the local MAC, in which case 640 the above may not result in any change in local MAC's sequence 641 number. 643 7.3 Remote MAC OR MAC-IP Update 645 On receiving a remote MAC OR MAC-IP route update associated with a 646 MAC Mx with a sequence number that is higher than a LOCAL route for 647 MAC Mx: 649 o GW MUST trigger probe and deletion procedure for all LOCAL IPs 650 associated with MAC Mx 652 o GW MUST trigger deletion procedure for LOCAL MAC route for Mx 654 7.4 REMOTE (SYNC) MAC update 656 Corresponding local MAC Mx (if present) Sequence number should be re- 657 computed as follows: 659 o If the current sequence number is less than the received SYNC 660 MAC sequence number, it MUST be increased to be equal to received 661 SYNC MAC sequence number. 663 o If a LOCAL MAC sequence number is updated as a result of the 664 above, all LOCAL MAC-IPs associated with MAC Mx MUST inherit the 665 updated sequence number. 667 7.5 REMOTE (SYNC) MAC-IP update 669 If this is a SYNCed MAC-IP on a local ESI, it would also result in a 670 derived SYNC MAC Mx route entry, as MAC only RT-2 advertisement is 671 optional. Corresponding local MAC Mx (if present) Sequence number 672 should be re-computed as follows: 674 o If the current sequence number is less than the received SYNC 675 MAC sequence number, it MUST be increased to be equal to received 676 SYNC MAC sequence number. 678 o If a LOCAL MAC sequence number is updated as a result of the 679 above, all LOCAL MAC-IPs associated with MAC Mx MUST inherit the 680 updated sequence number. 682 7.6 Inter-op 684 In general, if all GW nodes in the overlay network follow the above 685 sequence number assignment procedure, and the GW is advertising both 686 MAC+IP and MAC routes, sequence number advertised with the MAC and 687 MAC+IP routes with the same MAC would always be the same. However, an 688 inter-op scenario with a different implementation could arise, where 689 a GW implementation non-compliant with this document or with RFC 7432 690 assigns and advertises independent sequence numbers to MAC and MAC+IP 691 routes. To handle this case, if different sequence numbers are 692 received for remote MAC+IP and corresponding remote MAC routes from a 693 remote GW, sequence number associated with the remote MAC route 694 should be computed as: 696 o Highest of the all received sequence numbers with remote MAC+IP 697 and MAC routes with the same MAC. 699 o MAC sequence number would be re-computed on a MAC or MAC+IP 700 route withdraw as per above. 702 A MAC and / or IP move to the local GW would now result in the MAC 703 (and hence all MAC-IP) sequence numbers incremented from the above 704 computed remote MAC sequence number. 706 8. Routed Overlay 707 An additional use case is possible, such that traffic to an end host 708 in the overlay is always IP routed. In a purely routed overlay such 709 as this: 711 o A host MAC is never advertised in EVPN overlay control plane 713 o Host /32 or /128 IP reachability is distributed across the 714 overlay via EVPN route type 5 (RT-5) along with a zero or non- 715 zero ESI 717 o An overlay IP subnet may still be stretched across the underlay 718 fabric, however, intra-subnet traffic across the stretched 719 overlay is never bridged 721 o Both inter-subnet and intra-subnet traffic, in the overlay is 722 IP routed at the EVPN GW. 724 Please refer to [RFC 7814] for more details. 726 Host mobility within the stretched subnet would still need to be 727 supported for this use. In the absence of any host MAC routes, 728 sequence number mobility EXT-COMM specified in [RFC7432], section 7.7 729 may be associated with a /32 OR /128 host IP prefix advertised via 730 EVPN route type 5. MAC mobility procedures defined in RFC 7432 can 731 now be applied as is to host IP prefixes: 733 o On LOCAL learning of a host IP, on a new ESI, host IP MUST be 734 advertised with a sequence number attribute that is higher than 735 what is currently advertised with the old ESI 737 o on receiving a host IP route advertisement with a higher 738 sequence number, a PE MUST trigger ARP/ND probe and deletion 739 procedure on any LOCAL route for that IP with a lower sequence 740 number. A PE would essentially move the forwarding entry to point 741 to the remote route with a higher sequence number and send an 742 ARP/ND PROBE for the local IP route. If the IP has indeed moved, 743 PROBE would timeout and the local IP host route would be deleted. 745 Note that there is still only one sequence number associated with a 746 host route at any time. For earlier use cases where a host MAC is 747 advertised along with the host IP, a sequence number is only 748 associated with a MAC. Only if the MAC is not advertised at all, as 749 in this use case, is a sequence number associated with a host IP. 751 Note that this mobility procedure would not apply to "anycast IPv6" 752 hosts advertised via NA messages with 0-bit=0. Please refer to [EVPN- 753 PROXY-ARP]. 755 9. Duplicate Host Detection 757 Duplicate host detection scenarios across EVPN IRB can be classified 758 as follows: 760 o Scenario A: where two hosts have the same MAC (host IPs may or 761 may not be duplicate) 763 o Scenario B: where two hosts have the same IP but different MACs 765 o Scenario C: where two hosts have the same IP and host MAC is not 766 advertised at all 768 Duplicate detection procedures for scenario B and C would not apply 769 to "anycast IPv6" hosts advertised via NA messages with 0-bit=0. 770 Please refer to [EVPN-PROXY-ARP]. 772 9.1 Scenario A 774 For all use cases where duplicate hosts have the same MAC, MAC is 775 detected as duplicate via duplicate MAC detection procedure described 776 in RFC 7432. Corresponding MAC-IP routes with the same MAC do not 777 require duplicate detection and MUST simply inherit the DUPLICATE 778 property from the corresponding MAC route. In other words, if a MAC 779 route is in DUPLICATE state, all corresponding MAC-IP routes MUST 780 also be treated as DUPLICATE. Duplicate detection procedure need only 781 be applied to MAC routes. 783 9.2 Scenario B 785 Due to misconfiguration, a situation may arise where hosts with 786 different MACs are configured with the same IP. This scenario would 787 not be detected by existing duplicate MAC detection procedure and 788 would result in incorrect forwarding of routed traffic destined to 789 this IP. 791 Such a situation, on LOCAL MAC-IP learning, would be detected as a 792 move scenario via the following local MAC sequence number computation 793 procedure described earlier in section 5.1: 795 o If the IP is also associated with a different remote MAC "Mz", 796 MUST be higher than "Mz" sequence number 798 Such a move that results in sequence number increment on local MAC 799 because of a remote MAC-IP route associated with a different MAC MUST 800 be counted as an "IP move" against the "IP" independent of MAC. 801 Duplicate detection procedure described in RFC 7432 can now be 802 applied to an "IP" entity independent of MAC. Once an IP is detected 803 as DUPLICATE, corresponding MAC-IP route should be treated as 804 DUPLICATE. Associated MAC routes and any other MAC-IP routes 805 associated with this MAC should not be affected. 807 9.2.1 Duplicate IP Detection Procedure for Scenario B 809 Duplicate IP detection procedure for such a scenario is specified in 810 [EVPN-PROXY-ARP]. What counts as an "IP move" in this scenario is 811 further clarified as follows: 813 o On learning a LOCAL MAC-IP route Mx-IPx, check if there is an 814 existing REMOTE OR LOCAL route for IPx with a different MAC 815 association, say, Mz-IPx. If so, count this as an "IP move" count 816 for IPx, independent of the MAC 818 o On learning a REMOTE MAC-IP route Mz-IPx, check if there is an 819 existing LOCAL route for IPx with a different MAC association, 820 say, Mx-IPx. If so, count this as an "IP move" count for IPx, 821 independent of the MAC 823 A MAC-IP route SHOULD be treated as DUPLICATE if either of the 824 following two conditions are met: 826 o Corresponding MAC route is marked as DUPLICATE via existing 827 duplicate detection procedure 829 o Corresponding IP is marked as DUPLICATE via extended procedure 830 described above 832 9.3 Scenario C 834 For a purely routed overlay scenario described in section 8, where 835 only a host IP is advertised via EVPN RT-5, together with a sequence 836 number mobility attribute, duplicate MAC detection procedures 837 specified in RFC 7432 can be intuitively applied to IP only host 838 routes for the purpose of duplicate IP detection. 840 o On learning a LOCAL host IP route IPx, check if there is an 841 existing REMOTE OR LOCAL route for IPx with a different ESI 842 association. If so, count this as an "IP move" count for IPx. 844 o On learning a REMOTE host IP route IPx, check if there is an 845 existing LOCAL route for IPx with a different ESI association. If 846 so, count this as an "IP move" count for IPx 848 o With configurable parameters "N" and "M", If "N" IP moves are 849 detected within "M" seconds for IPx, treat IPx as DUPLICATE 851 9.4 Duplicate Host Recovery 853 Once a MAC or IP is marked as DUPLICATE and FROZEN, corrective action 854 must be taken to un-provision one of the duplicate MAC or IP. Un- 855 provisioning a duplicate MAC or IP in this context refers to a 856 corrective action taken on the host side. Once one of the duplicate 857 MAC or IP is un-provisioned, normal operation would not resume until 858 the duplicate MAC or IP ages out, following this correction, unless 859 additional action is taken to speed up recovery. 861 This section lists possible additional corrective actions that could 862 be taken to achieve faster recovery to normal operation. 864 9.4.1 Route Un-freezing Configuration 866 Unfreezing the DUPLICATE OR FROZEN MAC or IP via a CLI can be 867 leveraged to recover from DUPLICATE and FROZEN state following 868 corrective un-provisioning of the duplicate MAC or IP. 870 Unfreezing the frozen MAC or IP via a CLI at a GW should result in 871 that MAC OR IP being advertised with a sequence number that is higher 872 than the sequence number advertised from the other location of that 873 MAC or IP. 875 Two possible corrective un-provisioning scenarios exist: 877 o Scenario A: A duplicate MAC or IP may have been un-provisioned 878 at the location where it was NOT marked as DUPLICATE and FROZEN 880 o Scenario B: A duplicate MAC or IP may have been un-provisioned 881 at the location where it was marked as DUPLICATE and FROZEN 883 Unfreezing the DUPLICATE and FROZEN MAC or IP, following the above 884 corrective un-provisioning scenarios would result in recovery to 885 steady state as follows: 887 o Scenario A: If the duplicate MAC or IP was un-provisioned at 888 the location where it was NOT marked as DUPLICATE, unfreezing the 889 route at the FROZEN location will result in the route being 890 advertised with a higher sequence number. This would in-turn 891 result in automatic clearing of local route at the GW location, 892 where the host was un-provisioned via ARP/ND PROBE and DELETE 893 procedure specified earlier in section 8 and in [RFC 7432]. 895 o Scenario B: If the duplicate host is un-provisioned at the 896 location where it was marked as DUPLICATE, unfreezing the route 897 will trigger an advertisement with a higher sequence number to 898 the other location. This would in-turn trigger re-learning of 899 local route at the remote location, resulting in another 900 advertisement with a higher sequence number from the remote 901 location. Route at the local location would now be cleared on 902 receiving this remote route advertisement, following the ARP/ND 903 PROBE. 905 9.4.2 Route Clearing Configuration 907 In addition to the above, route clearing CLIs may also be leveraged 908 to clear the local MAC or IP route, to be executed AFTER the 909 duplicate host is un-provisioned: 911 o clear mac CLI: A clear MAC CLI can be leveraged to clear a 912 DUPLICATE MAC route, to recover from a duplicate MAC scenario 914 o clear ARP/ND: A clear ARP/ND CLI may be leveraged to clear a 915 DUPLICATE IP route to recover from a duplicate IP scenario 917 Note that the route unfreeze CLI may still need to be run if the 918 route was un-provisioned and cleared from the NON-DUPLICATE / NON- 919 FROZEN location. Given that unfreezing of the route via the un-freeze 920 CLI would any ways result in auto-clearing of the route from the "un- 921 provisioned" location, as explained in the prior section, need for a 922 route clearing CLI for recovery from DUPLICATE / FROZEN state is 923 truly optional. 925 10. Security Considerations 927 11. IANA Considerations 929 12. References 931 12.1 Normative References 933 [RFC7432] Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 934 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based 935 Ethernet VPN", RFC 7432, DOI 10.17487/RFC7432, February 936 2015, . 938 [EVPN-PROXY-ARP] Rabadan et al., "Operational Aspects of Proxy- 939 ARP/ND in EVPN Networks", draft-ietf-bess-evpn-proxy-arp- 940 nd-02, work in progress, April 2017, 941 . 944 [EVPN-INTER-SUBNET] Sajassi et al., "Integrated Routing and Bridging 945 in EVPN", draft-ietf-bess-evpn-inter-subnet-forwarding-03, 946 work in progress, Feb 2017, 947 . 950 [RFC7814] Xu, X., Jacquenet, C., Raszuk, R., Boyes, T., Fee, B., 951 "Virtual Subnet: A BGP/MPLS IP VPN-Based Subnet Extension 952 Solution", RFC 7814, March 2016, 953 . 955 12.2 Informative References 957 13. Acknowledgements 959 Authors would like to thank Vibov Bhan and Patrice Brisset for 960 feedback and comments through the process. 962 Authors' Addresses 964 Neeraj Malhotra (Editor) 965 Cisco 966 EMail: nmalhotr@cisco.com 968 Ali Sajassi 969 Cisco 970 EMail: sajassi@cisco.com 972 Aparna Pattekar 973 Cisco 974 Email: apjoshi@cisco.com 976 Avinash Lingala 977 AT&T 978 Email: ar977m@att.com 980 Jorge Rabadan 981 Nokia 982 Email: jorge.rabadan@nokia.com 984 John Drake 985 Juniper Networks 986 EMail: jdrake@juniper.net 988 Appendix A 990 An alternative approach considered was to associate two independent 991 sequence number attributes with MAC and IP components of a MAC-IP 992 route. However, the approach of enabling IRB mobility procedures 993 using a single sequence number associated with a MAC, as specified in 994 this document was preferred for the following reasons: 996 o Procedural overhead and complexity associated with maintaining 997 two separate sequence numbers all the time, only to address 998 scenarios with changing MAC-IP bindings is a big overhead for 999 topologies where MAC-IP bindings never change. 1001 o Using a single sequence number associated with MAC is much 1002 simpler and adds no overhead for topologies where MAC-IP bindings 1003 never change. 1005 o Using a single sequence number associated with MAC is aligned 1006 with existing MAC mobility implementations. On other words, it is 1007 an easier implementation extension to existing MAC mobility 1008 procedure.