idnits 2.17.1 draft-ietf-bess-evpn-usage-08.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 (February 16, 2018) is 2259 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'PE-IP' is mentioned on line 414, but not defined == Missing Reference: 'AS' is mentioned on line 420, but not defined == Outdated reference: A later version (-15) exists of draft-ietf-bess-evpn-inter-subnet-forwarding-03 Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 BESS Workgroup J. Rabadan, Ed. 3 Internet Draft S. Palislamovic 4 W. Henderickx 5 Intended status: Informational Nokia 7 A. Sajassi 8 Cisco 10 J. Uttaro 11 AT&T 13 Expires: August 20, 2018 February 16, 2018 15 Usage and applicability of BGP MPLS based Ethernet VPN 16 draft-ietf-bess-evpn-usage-08 18 Abstract 20 This document discusses the usage and applicability of BGP MPLS based 21 Ethernet VPN (EVPN) in a simple and fairly common deployment 22 scenario. The different EVPN procedures are explained on the example 23 scenario, analyzing the benefits and trade-offs of each option. This 24 document is intended to provide a simplified guide for the deployment 25 of EVPN networks. 27 Status of this Memo 29 This Internet-Draft is submitted in full conformance with the 30 provisions of BCP 78 and BCP 79. 32 Internet-Drafts are working documents of the Internet Engineering 33 Task Force (IETF), its areas, and its working groups. Note that 34 other groups may also distribute working documents as Internet- 35 Drafts. 37 Internet-Drafts are draft documents valid for a maximum of six months 38 and may be updated, replaced, or obsoleted by other documents at any 39 time. It is inappropriate to use Internet-Drafts as reference 40 material or to cite them other than as "work in progress." 42 The list of current Internet-Drafts can be accessed at 43 http://www.ietf.org/ietf/1id-abstracts.txt 45 The list of Internet-Draft Shadow Directories can be accessed at 46 http://www.ietf.org/shadow.html 48 This Internet-Draft will expire on August 20, 2018. 50 Copyright Notice 52 Copyright (c) 2018 IETF Trust and the persons identified as the 53 document authors. All rights reserved. 55 This document is subject to BCP 78 and the IETF Trust's Legal 56 Provisions Relating to IETF Documents 57 (http://trustee.ietf.org/license-info) in effect on the date of 58 publication of this document. Please review these documents 59 carefully, as they describe your rights and restrictions with respect 60 to this document. Code Components extracted from this document must 61 include Simplified BSD License text as described in Section 4.e of 62 the Trust Legal Provisions and are provided without warranty as 63 described in the Simplified BSD License. 65 Table of Contents 67 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 68 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . 3 69 3. Use-case scenario description and requirements . . . . . . . . 4 70 3.1. Service Requirements . . . . . . . . . . . . . . . . . . . 5 71 3.2. Why EVPN is chosen to address this use-case . . . . . . . . 6 72 4. Provisioning Model . . . . . . . . . . . . . . . . . . . . . . 7 73 4.1. Common provisioning tasks . . . . . . . . . . . . . . . . . 7 74 4.1.1. Non-service specific parameters . . . . . . . . . . . . 7 75 4.1.2. Service specific parameters . . . . . . . . . . . . . . 8 76 4.2. Service interface dependent provisioning tasks . . . . . . 9 77 4.2.1. VLAN-based service interface EVI . . . . . . . . . . . 9 78 4.2.2. VLAN-bundle service interface EVI . . . . . . . . . . . 10 79 4.2.3. VLAN-aware bundling service interface EVI . . . . . . . 10 80 5. BGP EVPN NLRI usage . . . . . . . . . . . . . . . . . . . . . . 10 81 6. MAC-based forwarding model use-case . . . . . . . . . . . . . . 11 82 6.1. EVPN Network Startup procedures . . . . . . . . . . . . . . 11 83 6.2. VLAN-based service procedures . . . . . . . . . . . . . . . 12 84 6.2.1. Service startup procedures . . . . . . . . . . . . . . 12 85 6.2.2. Packet walkthrough . . . . . . . . . . . . . . . . . . 13 86 6.3. VLAN-bundle service procedures . . . . . . . . . . . . . . 16 87 6.3.1. Service startup procedures . . . . . . . . . . . . . . 16 88 6.3.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 17 89 6.4. VLAN-aware bundling service procedures . . . . . . . . . . 17 90 6.4.1. Service startup procedures . . . . . . . . . . . . . . 18 91 6.4.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 18 92 7. MPLS-based forwarding model use-case . . . . . . . . . . . . . 19 93 7.1. Impact of MPLS-based forwarding on the EVPN network 94 startup . . . . . . . . . . . . . . . . . . . . . . . . . . 20 95 7.2. Impact of MPLS-based forwarding on the VLAN-based service 96 procedures . . . . . . . . . . . . . . . . . . . . . . . . 20 97 7.3. Impact of MPLS-based forwarding on the VLAN-bundle 98 service procedures . . . . . . . . . . . . . . . . . . . . 21 99 7.4. Impact of MPLS-based forwarding on the VLAN-aware service 100 procedures . . . . . . . . . . . . . . . . . . . . . . . . 21 101 8. Comparison between MAC-based and MPLS-based Egress Forwarding 102 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 103 9. Traffic flow optimization . . . . . . . . . . . . . . . . . . . 23 104 9.1. Control Plane Procedures . . . . . . . . . . . . . . . . . 23 105 9.1.1. MAC learning options . . . . . . . . . . . . . . . . . 23 106 9.1.2. Proxy-ARP/ND . . . . . . . . . . . . . . . . . . . . . 24 107 9.1.3. Unknown Unicast flooding suppression . . . . . . . . . 25 108 9.1.4. Optimization of Inter-subnet forwarding . . . . . . . . 25 109 9.2. Packet Walkthrough Examples . . . . . . . . . . . . . . . . 26 110 9.2.1. Proxy-ARP example for CE2 to CE3 traffic . . . . . . . 26 111 9.2.2. Flood suppression example for CE1 to CE3 traffic . . . 26 112 9.2.3. Optimization of inter-subnet forwarding example for 113 CE3 to CE2 traffic . . . . . . . . . . . . . . . . . . 27 114 10. Security Considerations . . . . . . . . . . . . . . . . . . . 28 115 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 29 116 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 29 117 12.1. Normative References . . . . . . . . . . . . . . . . . . . 29 118 12.2. Informative References . . . . . . . . . . . . . . . . . . 29 119 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 29 120 14. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 30 121 15. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 30 123 1. Introduction 125 This document complements [RFC7432] by discussing the applicability 126 of the technology in a simple and fairly common deployment scenario, 127 which is described in section 3. 129 After describing the topology and requirements of the use-case 130 scenario, section 4 will describe the provisioning model. 132 Once the provisioning model is analyzed, sections 5, 6 and 7 will 133 describe the control plane and data plane procedures in the example 134 scenario, for the two potential disposition/forwarding models: 135 MAC-based and MPLS-based models. While both models can interoperate 136 in the same network, each one has different trade-offs that are 137 analyzed in section 8. 139 Finally, EVPN provides some potential traffic flow optimization tools 140 that are also described in section 9, in the context of the example 141 scenario. 143 2. Terminology 144 The following terminology is used: 146 o VID: VLAN Identifier. 148 o CE: Customer Edge device. 150 o EVI: EVPN Instance. 152 o MAC-VRF: A Virtual Routing and Forwarding table for Media Access 153 Control (MAC) addresses on a PE. 155 o Ethernet Segment (ES): set of links through which a customer site 156 (CE) is connected to one or more PEs. Each ES is identified by an 157 Ethernet Segment Identifier (ESI) in the control plane. 159 o CE-VIDs refer to the VLAN tag identifiers being used at CE1, CE2 160 and CE3 to tag customer traffic sent to the Service Provider E- VPN 161 network 163 o CE1-MAC, CE2-MAC and CE3-MAC refer to source MAC addresses "behind" 164 each CE respectively. Those MAC addresses can belong to the CEs 165 themselves or to devices connected to the CEs. 167 o CE1-IP, CE2-IP and CE3-IP refer to IP addresses associated to the 168 above MAC addresses. 170 o LACP: Link Aggregation Control Protocol. 172 o RD: Route Distinguisher. 174 o RT: Route Target. 176 o PE: Provider Edge router. 178 o AS: Autonomous System. 180 o PE-IP: it refers to the IP address of a given PE. 182 3. Use-case scenario description and requirements 184 Figure 1 depicts the scenario that will be referenced throughout the 185 rest of the document. 187 +--------------+ 188 | | 189 +----+ +----+ | | +----+ +----+ 190 | CE1|-----| | | | | |---| CE3| 191 +----+ /| PE1| | IP/MPLS | | PE3| +----+ 192 / +----+ | Network | +----+ 193 / | | 194 / +----+ | | 195 +----+/ | | | | 196 | CE2|-----| PE2| | | 197 +----+ +----+ | | 198 +--------------+ 200 Figure 1 EVPN use-case scenario 202 There are three PEs and three CEs considered in this example: PE1, 203 PE2, PE3, as well as CE1, CE2 and CE3. Broadcast Domains must be 204 extended among the three CEs. 206 3.1. Service Requirements 208 The following service requirements are assumed in this scenario: 210 o Redundancy requirements: 212 - CE2 requires multi-homing connectivity to PE1 and PE2, not only 213 for redundancy purposes, but also for adding more 214 upstream/downstream connectivity bandwidth to/from the network. 216 - Fast convergence. For example: if the link between CE2 and PE1 217 goes down, a fast convergence mechanism must be supported so that 218 PE3 can immediately send the traffic to PE2, irrespective of the 219 number of affected services and MAC addresses. 221 o Service interface requirements: 223 - The service definition must be flexible in terms of CE-VID-to- 224 broadcast-domain assignment in the core. 226 - The following three EVI services are required in this example: 228 EVI100 - It uses VLAN-based service interfaces in the three CEs 229 with a 1:1 VLAN-to-EVI mapping. The CE-VIDs at the three CEs can 230 be the same, for example: VID 100, or different at each CE, for 231 instance: VID 101 in CE1, VID 102 in CE2 and VID 103 in CE3. A 232 single broadcast domain needs to be created for EVI100 in any 233 case; therefore CE-VIDs will require translation at the egress 234 PEs if they are not consistent across the three CEs. The case 235 when the same CE-VID is used across the three CEs for EVI100 is 236 referred in [RFC7432] as the "Unique VLAN" EVPN case. This term 237 will be used throughout this document too. 239 EVI200 - It uses VLAN-bundle service interfaces in CE1, CE2 and 240 CE3, based on an N:1 VLAN-to-EVI mapping. The operator needs to 241 pre-configure a range of CE-VIDs and its mapping to the EVI, and 242 this mapping should be consistent in all the PEs (no translation 243 is supported). A single broadcast domain is created for the 244 customer. The customer is responsible of keeping the separation 245 between users in different CE-VIDs. 247 EVI300 - It uses VLAN-aware bundling service interfaces in CE1, 248 CE2 and CE3. As in the EVI200 case, an N:1 VLAN-to-EVI mapping is 249 created at the ingress PEs, however in this case, a separate 250 broadcast domain is required per CE-VID. The CE-VIDs can be 251 different (hence CE-VID translation is required). 253 NOTE: in section 4.2.1, only EVI100 is used as an example of 254 VLAN-based service provisioning. In sections 6.2 and 7.2, 4k 255 VLAN-based EVIs (EVI1 to EVI4k) are used so that the impact of MAC 256 vs. MPLS disposition models in the control plane can be evaluated. In 257 the same way, EVI200 and EVI300 will be described with a 4k:1 mapping 258 (CE-VIDs-to-EVI mapping) in sections 6.3, 6.4, 7.3 and 7.4. 260 o BUM (Broadcast, Unknown unicast, Multicast) optimization 261 requirements: 263 - The solution must support ingress replication or P2MP MPLS LSPs 264 on a per EVI service. 266 - For example, we can use ingress replication for EVI100 and 267 EVI200, assuming those EVIs will not carry much BUM traffic. On 268 the contrary, if EVI300 is presumably carrying a significant 269 amount of multicast traffic, P2MP MPLS LSPs can be used for this 270 service. 272 - The benefit of ingress replication compared to P2MP LSPs is that 273 the core routers will not need to maintain any multicast states. 275 3.2. Why EVPN is chosen to address this use-case 277 VPLS solutions based on [RFC4761], [RFC4762] and [RFC6074] cannot 278 meet the requirements in section 3, whereas EVPN can. 280 For example: 282 o If CE2 has a single CE-VID (or a few CE-VIDs) the current VPLS 283 multi-homing solutions (based on load-balancing per CE-VID or 284 service) do not provide the optimized link utilization required in 285 this example. EVPN provides the flow-based load-balancing 286 multi-homing solution required in this scenario to optimize the 287 upstream/downstream link utilization between CE2 and PE1-PE2. 289 o Also, EVPN provides a fast convergence solution that is independent 290 of the CE-VIDs in the multi-homed PEs. Upon failure on the link 291 between CE2 and PE1, PE3 can immediately send the traffic to PE2, 292 based on a single notification message being sent by PE1. This is 293 not possible with VPLS solutions. 295 o With regard to service interfaces and mapping to broadcast domains, 296 while VPLS might meet the requirements for EVI100 and EVI200, the 297 VLAN-aware bundling service interfaces required by EVI300 are not 298 supported by the current VPLS tools. 300 The rest of the document will describe how EVPN can be used to meet 301 the service requirements described in section 3, and even optimize 302 the network further by: 304 o Providing the user with an option to reduce (and even suppress) 305 ARP-flooding. 307 o Supporting ARP termination and inter-subnet-forwarding. 309 4. Provisioning Model 311 One of the requirements stated in [RFC7209] is the ease of 312 provisioning. BGP parameters and service context parameters should be 313 auto-provisioned so that the addition of a new MAC-VRF to the EVI 314 requires a minimum number of single-sided provisioning touches. 315 However this is possible only in a limited number of cases. This 316 section describes the provisioning tasks required for the services 317 described in section 3, i.e. EVI100 (VLAN-based service interfaces), 318 EVI200 (VLAN-bundle service interfaces) and EVI300 (VLAN-aware 319 bundling service interfaces). 321 4.1. Common provisioning tasks 323 Regardless of the service interface type (VLAN-based, VLAN-bundle or 324 VLAN-aware), the following sub-sections describe the parameters to be 325 provisioned in the three PEs. 327 4.1.1. Non-service specific parameters 328 The multi-homing function in EVPN requires the provisioning of 329 certain parameters that are not service-specific and that are shared 330 by all the MAC-VRFs in the node using the multi-homing capabilities. 331 In our use-case, these parameters are only provisioned or auto- 332 derived in PE1 and PE2, and are listed below: 334 o Ethernet Segment Identifier (ESI): only the ESI associated to CE2 335 needs to be considered in our example. Single-homed CEs such as CE1 336 and CE3 do not require the provisioning of an ESI (the ESI will be 337 coded as zero in the BGP NLRIs). In our example, a LAG is used 338 between CE2 and PE1-PE2 (since all-active multi-homing is a 339 requirement) therefore the ESI can be auto-derived from the LACP 340 information as described in [RFC7432]. Note that the ESI must be 341 unique across all the PEs in the network, therefore the 342 auto-provisioning of the ESI is recommended only in case the CEs 343 are managed by the Operator. Otherwise the ESI should be manually 344 provisioned (type 0 as in [RFC7432]) in order to avoid potential 345 conflicts. 347 o ES-Import Route Target (ES-Import RT): this is the RT that will be 348 sent by PE1 and PE2, along with the ES route. Regardless of how the 349 ESI is provisioned in PE1 and PE2, the ES-Import RT must always be 350 auto-derived from the 6-byte MAC address portion of the ESI value. 352 o Ethernet Segment Route Distinguisher (ES RD): this is the RD to be 353 encoded in the ES route and Ethernet Auto-Discovery (A-D) route to 354 be sent by PE1 and PE2 for the CE2 ESI. This RD should always be 355 auto-derived from the PE IP address, as described in [RFC7432]. 357 o Multi-homing type: the user must be able to provision the 358 multi-homing type to be used in the network. In our use-case, the 359 multi-homing type will be set to all-active for the CE2 ESI. This 360 piece of information is encoded in the ESI Label extended community 361 flags and sent by PE1 and PE2 along with the Ethernet A-D route for 362 the CE2 ESI. 364 In addition, the same LACP parameters will be configured in PE1 and 365 PE2 for the ES so that CE2 can send frames to PE1 and PE2 as though 366 they were forming a single system. 368 4.1.2. Service specific parameters 370 The following parameters must be provisioned in PE1, PE2 and PE3 per 371 EVI service: 373 o EVI identifier: global identifier per EVI that is shared by all the 374 PEs part of the EVI, i.e. PE1, PE2 and PE3 will be provisioned with 375 EVI100, 200 and 300. The EVI identifier can be associated to (or be 376 the same value as) the EVI default Ethernet Tag (4-byte default 377 broadcast domain identifier for the EVI). The Ethernet Tag is 378 different from zero in the EVPN BGP routes only if the service 379 interface type (of the source PE) is VLAN-aware Bundle. 381 o EVI Route Distinguisher (EVI RD): This RD is a unique value across 382 all the MAC-VRFs in a PE. Auto-derivation of this RD might be 383 possible depending on the service interface type being used in the 384 EVI. Next section discusses the specifics of each service interface 385 type. 387 o EVI Route Target(s) (EVI RT): one or more RTs can be provisioned 388 per MAC-VRF. The RT(s) imported and exported can be equal or 389 different, just as the RT(s) in IP-VPNs. Auto-derivation of this 390 RT(s) might be possible depending on the service interface type 391 being used in the EVI. Next section discusses the specifics of each 392 service interface type. 394 o CE-VID and port/LAG binding to EVI identifier or Ethernet Tag: see 395 section 4.2. 397 4.2. Service interface dependent provisioning tasks 399 Depending on the service interface type being used in the EVI, a 400 specific CE-VID binding provisioning must be specified. 402 4.2.1. VLAN-based service interface EVI 404 In our use-case, EVI100 is a VLAN-based service interface EVI. 406 EVI100 can be a "unique-VLAN" service if the CE-VID being used for 407 this service in CE1, CE2 and CE3 is identical, for example VID 100. 408 In that case, the VID 100 binding must be provisioned in PE1, PE2 and 409 PE3 for EVI100 and the associated port or LAG. The MAC-VRF RD and RT 410 can be auto-derived from the CE-VID: 412 o The auto-derived MAC-VRF RD will be a Type 1 RD, as recommended in 413 [RFC7432], and it will be comprised of [PE-IP]:[zero-padded-VID]; 414 where [PE-IP] is the IP address of the PE (a loopback address) and 415 [zero-padded-VID] is a 2-byte value where the low order 12 bits are 416 the VID (VID 100 in our example) and the high order 4 bits are 417 zero. 419 o The auto-derived MAC-VRF RT will be composed of [AS]:[zero-padded- 420 VID]; where [AS] is the Autonomous System that the PE belongs to 421 and [zero-padded-VID] is a 2 or 4-byte value where the low order 12 422 bits are the VID (VID 100 in our example) and the high order bits 423 are zero. Note that auto-deriving the RT implies supporting a basic 424 any-to-any topology in the EVI and using the same import and export 425 RT in the EVI. 427 If EVI100 is not a "unique-VLAN" instance, each individual CE-VID 428 must be configured in each PE, and MAC-VRF RDs and RTs cannot be 429 auto-derived, hence they must be provisioned by the user. 431 4.2.2. VLAN-bundle service interface EVI 433 Assuming EVI200 is a VLAN-bundle service interface EVI, and VIDs 434 200-250 are assigned to EVI200, the CE-VID bundle 200-250 must be 435 provisioned on PE1, PE2 and PE3. Note that this model does not allow 436 CE-VID translation and the CEs must use the same CE-VIDs for EVI200. 437 No auto-derived EVI RDs or EVI RTs are possible. 439 4.2.3. VLAN-aware bundling service interface EVI 441 If EVI300 is a VLAN-aware bundling service interface EVI, CE-VID 442 binding to EVI300 does not have to match on the three PEs (only on 443 PE1 and PE2, since they are part of the same ES). For example: PE1 444 and PE2 CE-VID binding to EVI300 can be set to the range 300-310 and 445 PE3 to 321-330. Note that each individual CE-VID will be assigned to 446 a different broadcast domain, represented by an Ethernet Tag in the 447 control plane. 449 Therefore, besides the CE-VID bundle range bound to EVI300 in each 450 PE, associations between each individual CE-VID and the corresponding 451 EVPN Ethernet Tag must be provisioned by the user. No auto-derived 452 EVI RDs/RTs are possible. 454 5. BGP EVPN NLRI usage 456 [RFC7432] defines four different route types and four different 457 extended communities. However, not all the PEs in an EVPN network 458 must generate and process all the different routes and extended 459 communities. Table 1 shows the routes that must be exported and 460 imported in the use-case described in this document. "Export", in 461 this context, means that the PE must be capable of generating and 462 exporting a given route, assuming there are no BGP policies to 463 prevent it. In the same way, "Import" means the PE must be capable of 464 importing and processing a given route, assuming the right RTs and 465 policies. "N/A" means neither import nor export actions are required. 467 +-------------------+---------------+---------------+ 468 | BGP EVPN routes | PE1-PE2 | PE3 | 469 +-------------------+---------------+---------------+ 470 | ES | Export/import | N/A | 471 | A-D per ESI | Export/import | Import | 472 | A-D per EVI | Export/import | Import | 473 | MAC | Export/import | Export/import | 474 | Inclusive mcast | Export/import | Export/import | 475 +-------------------+---------------+---------------+ 477 Table 1 - Base EVPN Routes and Export/Import Actions 479 PE3 is required to export only MAC and Inclusive multicast routes and 480 be able to import and process A-D routes, as well as MAC and 481 Inclusive multicast routes. If PE3 did not support importing and 482 processing A-D routes per ESI and per EVI, fast convergence and 483 aliasing functions (respectively) would not be possible in this 484 use-case. 486 6. MAC-based forwarding model use-case 488 This section describes how the BGP EVPN routes are exported and 489 imported by the PEs in our use-case, as well as how traffic is 490 forwarded assuming that PE1, PE2 and PE3 support a MAC-based 491 forwarding model. In order to compare the control and data plane 492 impact in the two forwarding models (MAC-based and MPLS-based) and 493 different service types, we will assume that CE1, CE2 and CE3 need to 494 exchange traffic for up to 4k CE-VIDs. 496 6.1. EVPN Network Startup procedures 498 Before any EVI is provisioned in the network, the following 499 procedures are required: 501 o Infrastructure setup: the proper MPLS infrastructure must be setup 502 among PE1, PE2 and PE3 so that the EVPN services can make use of 503 P2P and P2MP LSPs. In addition to the MPLS transport, PE1 and PE2 504 must be properly configured with the same LACP configuration to 505 CE2. Details are provided in [RFC7432]. Once the LAG is properly 506 setup, the ESI for the CE2 Ethernet Segment, for example ESI12, can 507 be auto-generated by PE1 and PE2 from the LACP information 508 exchanged with CE2 (ESI type 1), as discussed in section 4.1. 509 Alternatively, the ESI can also be manually provisioned on PE1 and 510 PE2 (ESI type 0). PE1 and PE2 will auto-configure a BGP policy that 511 will import any ES route matching the auto-derived ES-import RT for 512 ESI12. 514 o Ethernet Segment route exchange and DF election: PE1 and PE2 will 515 advertise a BGP Ethernet Segment route for ESI12, where the ESI RD 516 and ES-Import RT will be auto-generated as discussed in section 517 4.1.1. PE1 and PE2 will import the ES routes of each other and will 518 run the DF election algorithm for any existing EVI (if any, at this 519 point). PE3 will simply discard the route. Note that the DF 520 election algorithm can support service carving, so that the 521 downstream BUM traffic from the network to CE2 can be load-balanced 522 across PE1 and PE2 on a per-service basis. 524 At the end of this process, the network infrastructure is ready to 525 start deploying EVPN services. PE1 and PE2 are aware of the existence 526 of a shared Ethernet Segment, i.e. ESI12. 528 6.2. VLAN-based service procedures 530 Assuming that the EVPN network must carry traffic among CE1, CE2 and 531 CE3 for up to 4k CE-VIDs, the Service Provider can decide to 532 implement VLAN-based service interface EVIs to accomplish it. In this 533 case, each CE-VID will be individually mapped to a different EVI. 534 While this means a total number of 4k MAC-VRFs is required per PE, 535 the advantages of this approach are the auto-provisioning of most of 536 the service parameters if no VLAN translation is needed (see section 537 4.2.1) and great control over each individual customer broadcast 538 domain. We assume in this section that the range of EVIs from 1 to 4k 539 is provisioned in the network. 541 6.2.1. Service startup procedures 543 As soon as the EVIs are created in PE1, PE2 and PE3, the following 544 control plane actions are carried out: 546 o Flooding tree setup per EVI (4k routes): Each PE will send one 547 Inclusive Multicast Ethernet Tag route per EVI (up to 4k routes per 548 PE) so that the flooding tree per EVI can be setup. Note that 549 ingress replication or P2MP LSPs can optionally be signaled in the 550 PMSI Tunnel attribute and the corresponding tree be created. 552 o Ethernet A-D routes per ESI (a set of routes for ESI12): A set of 553 A-D routes with a total list of 4k RTs (one per EVI) for ESI12 will 554 be issued from PE1 and PE2 (it has to be a set of routes so that 555 the total number of RTs can be conveyed). As per [RFC7432], each 556 Ethernet A-D route per ESI is differentiated from the other routes 557 in the set by a different Route Distinguisher (ES RD). This set 558 will also include ESI Label extended communities with the active- 559 standby flag set to zero (all-active multi-homing type) and an ESI 560 Label different from zero (used for split-horizon functions). These 561 routes will be imported by the three PEs, since the RTs match the 562 EVI RTs locally configured. The A-D routes per ESI will be used for 563 fast convergence and split-horizon functions, as discussed in 565 [RFC7432]. 567 o Ethernet A-D routes per EVI (4k routes): An A-D route per EVI will 568 be sent by PE1 and PE2 for ESI12. Each individual route includes 569 the corresponding EVI RT and an MPLS label to be used by PE3 for 570 the aliasing function. These routes will be imported by the three 571 PEs. 573 6.2.2. Packet walkthrough 575 Once the services are setup, the traffic can start flowing. Assuming 576 there are no MAC addresses learned yet and that MAC learning at the 577 access is performed in the data plane in our use-case, this is the 578 process followed upon receiving frames from each CE (example for 579 EVI1). 581 (1) BUM frame example from CE1: 583 a) An ARP-request with CE-VID=1 is issued from source MAC CE1-MAC 584 (MAC address coming from CE1 or from a device connected to CE1) to 585 find the MAC address of CE3-IP. 587 b) Based on the CE-VID, the frame is identified to be forwarded in 588 the MAC-VRF-1 (EVI1) context. A source MAC lookup is done in the 589 MAC FIB and the sender's CE1-IP in the proxy-ARP table within the 590 MAC-VRF-1 (EVI1) context. If CE1-MAC/CE1-IP are unknown in both 591 tables, three actions are carried out (assuming the source MAC is 592 accepted by PE1): 594 (1) Forwarding state is added for CE1-MAC associated to the 595 corresponding port and CE-VID, 597 (2) the ARP-request is snooped and the tuple CE1-MAC/CE1-IP is 598 added to the proxy-ARP table and 600 (3) a BGP MAC advertisement route is triggered from PE1 containing 601 the EVI1 RD and RT, ESI=0, Ethernet-Tag=0 and CE1-MAC/CE1-IP 602 along with an MPLS label assigned to MAC-VRF-1 from the PE1 603 label space. Note that depending on the implementation, the 604 MAC FIB and proxy-ARP learning processes can independently 605 send two BGP MAC advertisements instead of one (one containing 606 only the CE1-MAC and another one containing CE1-MAC/CE1-IP). 608 Since we assume a MAC forwarding model, a label per MAC-VRF is 609 normally allocated and signaled by the three PEs for MAC 610 advertisement routes. Based on the RT, the route is imported by 611 PE2 and PE3 and the forwarding state plus ARP entry are added to 612 their MAC-VRF-1 context. From this moment on, any ARP request from 613 CE2 or CE3 destined to CE1-IP, can be directly replied by PE1, PE2 614 or PE3 and ARP flooding for CE1-IP is not needed in the core. 616 c) Since the ARP frame is a broadcast frame, it is forwarded by PE1 617 using the Inclusive multicast tree for EVI1 (CE-VID=1 tag should 618 be kept if translation is required). Depending on the type of 619 tree, the label stack may vary. For example assuming ingress 620 replication, the packet is replicated to PE2 and PE3 with the 621 downstream allocated labels and the P2P LSP transport labels. No 622 other labels are added to the stack. 624 d) Assuming PE1 is the DF for EVI1 on ESI12, the frame is locally 625 replicated to CE2. 627 e) The MPLS-encapsulated frame gets to PE2 and PE3. Since PE2 is non- 628 DF for EVI1 on ESI12, and there is no other CE connected to PE2, 629 the frame is discarded. At PE3, the frame is de-encapsulated, CE- 630 VID translated if needed and forwarded to CE3. 632 Any other type of BUM frame from CE1 would follow the same 633 procedures. BUM frames from CE3 would follow the same procedures too. 635 (2) BUM frame example from CE2: 637 a) An ARP-request with CE-VID=1 is issued from source MAC CE2-MAC to 638 find the MAC address of CE3-IP. 640 b) CE2 will hash the frame and will forward it to for example PE2. 641 Based on the CE-VID, the frame is identified to be forwarded in 642 the EVI1 context. A source MAC lookup is done in the MAC FIB and 643 the sender's CE2-IP in the proxy-ARP table within the MAC-VRF-1 644 context. If both are unknown, three actions are carried out 645 (assuming the source MAC is accepted by PE2): 647 (1) Forwarding state is added for CE2-MAC associated to the 648 corresponding LAG/ESI and CE-VID, 650 (2) the ARP-request is snooped and the tuple CE2-MAC/CE2-IP is 651 added to the proxy-ARP table and 653 (3) a BGP MAC advertisement route is triggered from PE2 containing 654 the EVI1 RD and RT, ESI=12, Ethernet-Tag=0 and CE2-MAC/CE2-IP 655 along with an MPLS label assigned from the PE2 label space 656 (one label per MAC-VRF). Again, depending on the 657 implementation, the MAC FIB and proxy-ARP learning processes 658 can independently send two BGP MAC advertisements instead of 659 one. 661 Note that, since PE3 is not part of ESI12, it will install 662 forwarding state for CE2-MAC as long as the A-D routes for ESI12 663 are also active on PE3. On the contrary, PE1 is part of ESI12, 664 therefore PE1 will not modify the forwarding state for CE2-MAC if 665 it has previously learnt CE2-MAC locally attached to ESI12. 666 Otherwise it will add forwarding state for CE2-MAC associated to 667 the local ESI12 port. 669 c) Assuming PE2 does not have the ARP information for CE3-IP yet, and 670 since the ARP is a broadcast frame and PE2 the non-DF for EVI1 on 671 ESI12, the frame is forwarded by PE2 in the Inclusive multicast 672 tree for EVI1, adding the ESI label for ESI12 at the bottom of the 673 stack. The ESI label has been previously allocated and signaled by 674 the A-D routes for ESI12. Note that, as per [RFC7432], if the 675 result of the CE2 hashing is different and the frame sent to PE1, 676 PE1 should add the ESI label too (PE1 is the DF for EVI1 on 677 ESI12). 679 d) The MPLS-encapsulated frame gets to PE1 and PE3. PE1 680 de-encapsulates the Inclusive multicast tree label(s) and based on 681 the ESI label at the bottom of the stack, it decides to not 682 forward the frame to the ESI12. It will pop the ESI label and will 683 replicate it to CE1 though, since CE1 is not part of the ESI 684 identified by the ESI label. At PE3, the Inclusive multicast tree 685 label is popped and the frame forwarded to CE3. If a P2MP LSP is 686 used as Inclusive multicast tree for EVI1, PE3 will find an ESI 687 label after popping the P2MP LSP label. The ESI label will simply 688 be popped, since CE3 is not part of ESI12. 690 (3) Unicast frame example from CE3 to CE1: 692 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 693 and destination MAC CE1-MAC (we assume PE3 has previously resolved 694 an ARP request from CE3 to find the MAC of CE1-IP, and has added 695 CE3-MAC/CE3-IP to its proxy-ARP table). 697 b) Based on the CE-VID, the frame is identified to be forwarded in 698 the EVI1 context. A source MAC lookup is done in the MAC FIB 699 within the MAC-VRF-1 context and this time, since we assume CE3- 700 MAC is known, no further actions are carried out as a result of 701 the source lookup. A destination MAC lookup is performed next and 702 the label stack associated to the MAC CE1-MAC is found (including 703 the label associated to MAC-VRF-1 in PE1 and the P2P LSP label to 704 get to PE1). The unicast frame is then encapsulated and forwarded 705 to PE1. 707 c) At PE1, the packet is identified to be part of EVI1 and a 708 destination MAC lookup is performed in the MAC-VRF-1 context. The 709 labels are popped and the frame forwarded to CE1 with CE-VID=1. 711 Unicast frames from CE1 to CE3 or from CE2 to CE3 follow the same 712 procedures described above. 714 (4) Unicast frame example from CE3 to CE2: 716 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 717 and destination MAC CE2-MAC (we assume PE3 has previously resolved 718 an ARP request from CE3 to find the MAC of CE2-IP). 720 b) Based on the CE-VID, the frame is identified to be forwarded in 721 the MAC-VRF-1 context. We assume CE3-MAC is known. A destination 722 MAC lookup is performed next and PE3 finds CE2-MAC associated to 723 PE2 on ESI12, an Ethernet Segment for which PE3 has two active A-D 724 routes per ESI (from PE1 and PE2) and two active A-D routes for 725 EVI1 (from PE1 and PE2). Based on a hashing function for the 726 frame, PE3 may decide to forward the frame using the label stack 727 associated to PE2 (label received from the MAC advertisement 728 route) or the label stack associated to PE1 (label received from 729 the A-D route per EVI for EVI1). Either way, the frame is 730 encapsulated and sent to the remote PE. 732 c) At PE2 (or PE1), the packet is identified to be part of EVI1 based 733 on the bottom label, and a destination MAC lookup is performed. At 734 either PE (PE2 or PE1), the FIB lookup yields a local ESI12 port 735 to which the frame is sent. 737 Unicast frames from CE1 to CE2 follow the same procedures. 739 6.3. VLAN-bundle service procedures 741 Instead of using VLAN-based interfaces, the Operator can choose to 742 implement VLAN-bundle interfaces to carry the traffic for the 4k CE- 743 VIDs among CE1, CE2 and CE3. If that is the case, the 4k CE-VIDs can 744 be mapped to the same EVI, for example EVI200, at each PE. The main 745 advantage of this approach is the low control plane overhead (reduced 746 number of routes and labels) and easiness of provisioning, at the 747 expense of no control over the customer broadcast domains, i.e. a 748 single inclusive multicast tree for all the CE-VIDs and no CE-VID 749 translation in the Provider network. 751 6.3.1. Service startup procedures 753 As soon as the EVI200 is created in PE1, PE2 and PE3, the following 754 control plane actions are carried out: 756 o Flooding tree setup per EVI (one route): Each PE will send one 757 Inclusive Multicast Ethernet Tag route per EVI (hence only one 758 route per PE) so that the flooding tree per EVI can be setup. Note 759 that ingress replication or P2MP LSPs can optionally be signaled 760 in the PMSI Tunnel attribute and the corresponding tree be 761 created. 763 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 764 route for ESI12 will be issued from PE1 and PE2. This route will 765 include a single RT (RT for EVI200), an ESI Label extended 766 community with the active-standby flag set to zero (all-active 767 multi-homing type) and an ESI Label different from zero (used by 768 the non-DF for split-horizon functions). This route will be 769 imported by the three PEs, since the RT matches the EVI200 RT 770 locally configured. The A-D routes per ESI will be used for fast 771 convergence and split-horizon functions, as described in 772 [RFC7432]. 774 o Ethernet A-D routes per EVI (one route): An A-D route (EVI200) will 775 be sent by PE1 and PE2 for ESI12. This route includes the EVI200 776 RT and an MPLS label to be used by PE3 for the aliasing function. 777 This route will be imported by the three PEs. 779 6.3.2. Packet Walkthrough 781 The packet walkthrough for the VLAN-bundle case is similar to the one 782 described for EVI1 in the VLAN-based case except for the way the 783 CE-VID is handled by the ingress PE and the egress PE: 785 o No VLAN translation is allowed and the CE-VIDs are kept untouched 786 from CE to CE, i.e. the ingress CE-VID must be kept at the 787 imposition PE and at the disposition PE. 789 o The frame is identified to be forwarded in the MAC-VRF-200 context 790 as long as its CE-VID belongs to the VLAN-bundle defined in the 791 PE1/PE2/PE3 port to CE1/CE2/CE3. Our example is a special VLAN- 792 bundle case, since the entire CE-VID range is defined in the 793 ports, therefore any CE-VID would be part of EVI200. 795 Please refer to section 6.2.2 for more information about the control 796 plane and forwarding plane interaction for BUM and unicast traffic 797 from the different CEs. 799 6.4. VLAN-aware bundling service procedures 801 The last potential service type analyzed in this document is 802 VLAN-aware bundling. When this type of service interface is used to 803 carry the 4k CE-VIDs among CE1, CE2 and CE3, all the CE-VIDs will be 804 mapped to the same EVI, for example EVI300. The difference, compared 805 to the VLAN-bundle service type in the previous section, is that each 806 incoming CE-VID will also be mapped to a different "normalized" 807 Ethernet-Tag in addition to EVI300. If no translation is required, 808 the Ethernet-tag will match the CE-VID. Otherwise a translation 809 between CE-VID and Ethernet-tag will be needed at the imposition PE 810 and at the disposition PE. The main advantage of this approach is the 811 ability to control customer broadcast domains while providing a 812 single EVI to the customer. 814 6.4.1. Service startup procedures 816 As soon as the EVI300 is created in PE1, PE2 and PE3, the following 817 control plane actions are carried out: 819 o Flooding tree setup per EVI per Ethernet-Tag (4k routes): Each PE 820 will send one Inclusive Multicast Ethernet Tag route per EVI and 821 per Ethernet-Tag (hence 4k routes per PE) so that the flooding 822 tree per customer broadcast domain can be setup. Note that ingress 823 replication or P2MP LSPs can optionally be signaled in the PMSI 824 Tunnel attribute and the corresponding tree be created. In the 825 described use-case, since all the CE-VIDs and Ethernet-Tags are 826 defined on the three PEs, multicast tree aggregation might make 827 sense in order to save forwarding states. 829 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 830 route for ESI12 will be issued from PE1 and PE2. This route will 831 include a single RT (RT for EVI300), an ESI Label extended 832 community with the active-standby flag set to zero (all-active 833 multi-homing type) and an ESI Label different than zero (used by 834 the non-DF for split-horizon functions). This route will be 835 imported by the three PEs, since the RT matches the EVI300 RT 836 locally configured. The A-D routes per ESI will be used for fast 837 convergence and split-horizon functions, as described in 838 [RFC7432]. 840 o Ethernet A-D routes per EVI: a single A-D route (EVI300) may be 841 sent by PE1 and PE2 for ESI12, in case no CE-VID translation is 842 required. This route includes the EVI300 RT and an MPLS label to 843 be used by PE3 for the aliasing function. This route will be 844 imported by the three PEs. Note that if CE-VID translation is 845 required, an A-D per EVI route is required per Ethernet-Tag (4k). 847 6.4.2. Packet Walkthrough 849 The packet walkthrough for the VLAN-aware case is similar to the one 850 described before. Compared to the other two cases, VLAN-aware 851 services allow for CE-VID translation and for an N:1 CE-VID to EVI 852 mapping. Both things are not supported at once in either of the two 853 other service interfaces. Some differences compared to the packet 854 walkthrough described in section 6.2.2 are: 856 o At the ingress PE, the frames are identified to be forwarded in the 857 EVI300 context as long as their CE-VID belong to the range defined 858 in the PE port to the CE. In addition to it, CE-VID=x is mapped to 859 a "normalized" Ethernet-Tag=y at the MAC-VRF-300 (where x and y 860 might be equal if no translation is needed). Qualified learning is 861 now required (a different Bridge Table is allocated within MAC- 862 VRF-300 for each Ethernet-Tag). Potentially the same MAC could be 863 learned in two different Ethernet-Tag Bridge Tables of the same 864 MAC-VRF. 866 o Any new locally learned MAC on the MAC-VRF-300/Ethernet-Tag=y 867 interface is advertised by the ingress PE in a MAC advertisement 868 route, using now the Ethernet-Tag field (Ethernet-Tag=y) so that 869 the remote PE learns the MAC associated to the MAC-VRF- 870 300/Ethernet-Tag=y FIB. Note that the Ethernet-Tag field is not 871 used in advertisements of MACs learned on VLAN-based or VLAN- 872 bundle service interfaces. 874 o At the ingress PE, BUM frames are sent to the corresponding 875 flooding tree for the particular Ethernet-Tag they are mapped to. 876 Each individual Ethernet-Tag can have a different flooding tree 877 within the same EVI300. For instance, Ethernet-Tag=y can use 878 ingress replication to get to the remote PEs whereas Ethernet- 879 Tag=z can use a p2mp LSP. 881 o At the egress PE, Ethernet-Tag=y, for a given broadcast domain 882 within MAC-VRF-300, can be translated to egress CE-VID=x. That is 883 not possible for VLAN-bundle interfaces. It is possible for VLAN- 884 based interfaces, but it requires a separate MAC-VRF per CE-VID. 886 7. MPLS-based forwarding model use-case 888 EVPN supports an alternative forwarding model, usually referred to as 889 MPLS-based forwarding or disposition model as opposed to the 890 MAC-based forwarding or disposition model described in section 6. 891 Using MPLS-based forwarding model instead of MAC-based model might 892 have an impact on: 894 o The number of forwarding states required. 896 o The FIB where the forwarding states are handled: MAC FIB or MPLS 897 LFIB. 899 The MPLS-based forwarding model avoids the destination MAC lookup at 900 the egress PE MAC FIB, at the expense of increasing the number of 901 next-hop forwarding states at the egress MPLS LFIB. This also has an 902 impact on the control plane and the label allocation model, since an 903 MPLS-based disposition PE must send as many routes and labels as 904 required next-hops in the egress MAC-VRF. This concept is equivalent 905 to the forwarding models supported in IP-VPNs at the egress PE, where 906 an IP lookup in the IP-VPN FIB might be necessary or not depending on 907 the available next-hop forwarding states in the LFIB. 909 The following sub-sections highlight the impact on the control and 910 data plane procedures described in section 6 when and MPLS-based 911 forwarding model is used. 913 Note that both forwarding models are compatible and interoperable in 914 the same network. The implementation of either model in each PE is a 915 local decision to the PE node. 917 7.1. Impact of MPLS-based forwarding on the EVPN network startup 919 The MPLS-based forwarding model has no impact on the procedures 920 explained in section 6.1. 922 7.2. Impact of MPLS-based forwarding on the VLAN-based service 923 procedures 925 Compared to the MAC-based forwarding model, the MPLS-based forwarding 926 model has no impact in terms of number of routes, when all the 927 service interfaces are VLAN-based. The differences for the use-case 928 described in this document are summarized in the following list: 930 o Flooding tree setup per EVI (4k routes per PE): no impact compared 931 to the MAC-based model. 933 o Ethernet A-D routes per ESI (one set of routes for ESI12 per PE): 934 no impact compared to the MAC-based model. 936 o Ethernet A-D routes per EVI (4k routes per PE/ESI): no impact 937 compared to the MAC-based model. 939 o MAC-advertisement routes: instead of allocating and advertising the 940 same MPLS label for all the new MACs locally learnt on the same 941 MAC-VRF, a different label must be advertised per CE next-hop or 942 MAC so that no MAC FIB lookup is needed at the egress PE. In 943 general, this means that a different label at least per CE must be 944 advertised, although the PE can decide to implement a label per MAC 945 if more granularity (hence less scalability) is required in terms 946 of forwarding states. For example if CE2 sends traffic from two 947 different MACs to PE1, CE2-MAC1 and CE2-MAC2, the same MPLS label=x 948 can be re-used for both MAC advertisements since they both share 949 the same source ESI12. It is up to the PE1 implementation to use a 950 different label per individual MAC within the same ES Segment (even 951 if only one label per ESI is enough). 953 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 954 learning new local CE MAC addresses on the data plane, but will 955 rather add forwarding states to the MPLS LFIB. 957 7.3. Impact of MPLS-based forwarding on the VLAN-bundle service 958 procedures 960 Compared to the MAC-based forwarding model, the MPLS-based forwarding 961 model has no impact in terms of number of routes when all the service 962 interfaces are VLAN-bundle type. The differences for the use-case 963 described in this document are summarized in the following list: 965 o Flooding tree setup per EVI (one route): no impact compared to the 966 MAC-based model. 968 o Ethernet A-D routes per ESI (one route for ESI12 per PE): no impact 969 compared to the MAC-based model. 971 o Ethernet A-D routes per EVI (one route per PE/ESI): no impact 972 compared to the MAC-based model since no VLAN translation is 973 required. 975 o MAC-advertisement routes: instead of allocating and advertising the 976 same MPLS label for all the new MACs locally learnt on the same 977 MAC-VRF, a different label must be advertised per CE next-hop or 978 MAC so that no MAC FIB lookup is needed at the egress PE. In 979 general, this means that a different label at least per CE must be 980 advertised, although the PE can decide to implement a label per MAC 981 if more granularity (hence less scalability) is required in terms 982 of forwarding states. It is up to the PE1 implementation to use a 983 different label per individual MAC within the same ES Segment (even 984 if only one label per ESI is enough). 986 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 987 learning new local CE MAC addresses on the data plane, but will 988 rather add forwarding states to the MPLS LFIB. 990 7.4. Impact of MPLS-based forwarding on the VLAN-aware service 991 procedures 993 Compared to the MAC-based forwarding model, the MPLS-based forwarding 994 model has no impact in terms of number of A-D routes when all the 995 service interfaces are VLAN-aware bundle type. The differences for 996 the use-case described in this document are summarized in the 997 following list: 999 o Flooding tree setup per EVI (4k routes per PE): no impact compared 1000 to the MAC-based model. 1002 o Ethernet A-D routes per ESI (one route for ESI12 per PE): no impact 1003 compared to the MAC-based model. 1005 o Ethernet A-D routes per EVI (1 route per ESI or 4k routes per 1006 PE/ESI): PE1 and PE2 may send one route per ESI if no CE-VID 1007 translation is needed. However, 4k routes normally sent for EVI300, 1008 one per tuple. This will allow the egress PE 1009 to find out all the forwarding information in the MPLS LFIB and 1010 even support Ethernet-Tag to CE-VID translation at the egress. 1012 o MAC-advertisement routes: instead of allocating and advertising the 1013 same MPLS label for all the new MACs locally learnt on the same 1014 MAC-VRF, a different label must be advertised per CE next-hop or 1015 MAC so that no MAC FIB lookup is needed at the egress PE. In 1016 general, this means that a different label at least per CE must be 1017 advertised, although the PE can decide to implement a label per MAC 1018 if more granularity (hence less scalability) is required in terms 1019 of forwarding states. It is up to the PE1 implementation to use a 1020 different label per individual MAC within the same ES Segment. Note 1021 that the Ethernet-Tag will be set to a non-zero value for the MAC- 1022 advertisement routes. The same MAC address can be announced with 1023 different Ethernet-Tag value. This will make the advertising PE 1024 install two different forwarding states in the MPLS LFIB. 1026 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 1027 learning new local CE MAC addresses on the data plane, but will 1028 rather add forwarding states to the MPLS LFIB. 1030 8. Comparison between MAC-based and MPLS-based Egress Forwarding Models 1032 Both forwarding models are possible in a network deployment and each 1033 one has its own trade-offs. 1035 Both forwarding models can save A-D routes per EVI when VLAN-aware 1036 bundling services are deployed and no CE-VID translation is required. 1037 While this saves a significant amount of routes, customers normally 1038 require CE-VID translation, hence we assume an A-D per EVI route per 1039 is needed. 1041 The MAC-based model saves a significant amount of MPLS labels 1042 compared to the MPLS-based forwarding model. All the MACs and A-D 1043 routes for the same EVI can signal the same MPLS label, saving labels 1044 from the local PE space. A MAC FIB lookup at the egress PE is 1045 required in order to do so. 1047 The MPLS-based forwarding model can save forwarding states at the 1048 egress PEs if labels per next hop CE (as opposed to per MAC) are 1049 implemented. No egress MAC lookup is required. Also, a different 1050 label per next-hop CE per MAC-VRF is consumed, as opposed to a single 1051 label per MAC-VRF. 1053 Table 2 summarizes the resource implementation details of both 1054 models. 1056 +-----------------------------+----------------+----------------+ 1057 | Resources | MAC-based | MPLS-based | 1058 | | Model | Model | 1059 +-----------------------------+----------------+----------------+ 1060 | MPLS labels consumed | 1 per MAC-VRF | 1 per CE/EVI | 1061 | Egress PE Forwarding states | 1 per MAC | 1 per next-hop | 1062 | Egress PE Lookups | 2 (MPLS+MAC) | 1 (MPLS) | 1063 +-----------------------------+----------------+----------------+ 1065 Table 2 - Resource Comparison Between MAC-based and MPLS-based Models 1067 The egress forwarding model is an implementation local to the egress 1068 PE and is independent of the model supported on the rest of the PEs, 1069 i.e. in our use-case, PE1, PE2 and PE3 could have either egress 1070 forwarding model without any dependencies. 1072 9. Traffic flow optimization 1074 In addition to the procedures described across sections 3 through 8, 1075 EVPN [RFC7432] procedures allow for optimized traffic handling in 1076 order to minimize unnecessary flooding across the entire 1077 infrastructure. Optimization is provided through specific ARP 1078 termination and the ability to block unknown unicast flooding. 1079 Additionally, EVPN procedures allow for intelligent, close to the 1080 source, inter-subnet forwarding and solves the commonly known sub- 1081 optimal routing problem. Besides the traffic efficiency, ingress 1082 based inter-subnet forwarding also optimizes packet forwarding rules 1083 and implementation at the egress nodes as well. Details of these 1084 procedures are outlined in sections 9.1 and 9.2. 1086 9.1. Control Plane Procedures 1088 9.1.1. MAC learning options 1090 The fundamental premise of [RFC7432] is the notion of a different 1091 approach to MAC address learning compared to traditional IEEE 802.1 1092 bridge learning methods; specifically EVPN differentiates between 1093 data and control plane driven learning mechanisms. 1095 Data driven learning implies that there is no separate communication 1096 channel used to advertise and propagate MAC addresses. Rather, MAC 1097 addresses are learned through IEEE defined bridge-learning procedures 1098 as well as by snooping on DHCP and ARP requests. As different MAC 1099 addresses show up on different ports, the L2 FIB is populated with 1100 the appropriate MAC addresses. 1102 Control plane driven learning implies a communication channel that 1103 could be either a control-plane protocol or a management-plane 1104 mechanism. In the context of EVPN, two different learning procedures 1105 are defined, i.e. local and remote procedures: 1107 o Local learning defines the procedures used for learning the MAC 1108 addresses of network elements locally connected to a MAC-VRF. 1109 Local learning could be implemented through all three learning 1110 procedures: control plane, management plane as well as data plane. 1111 However, the expectation is that for most of the use cases, local 1112 learning through data plane should be sufficient. 1114 o Remote learning defines the procedures used for learning MAC 1115 addresses of network elements remotely connected to a MAC-VRF, 1116 i.e. far-end PEs. Remote learning procedures defined in [RFC7432] 1117 advocate using only control plane learning; specifically BGP. 1118 Through the use of BGP EVPN NLRIs, the remote PE has the 1119 capability of advertising all the MAC addresses present in its 1120 local FIB. 1122 9.1.2. Proxy-ARP/ND 1124 In EVPN, MAC addresses are advertised via the MAC/IP Advertisement 1125 Route, as discussed in [RFC7432]. Optionally an IP address can be 1126 advertised along with the MAC address advertisement. However, there 1127 are certain rules put in place in terms of IP address usage: if the 1128 MAC/IP Route contains an IP address, this particular IP address 1129 correlates directly with the advertised MAC address. Such 1130 advertisement allows us to build a proxy-ARP/ND table populated with 1131 the IP<->MAC bindings received from all the remote nodes. 1133 Furthermore, based on these bindings, a local MAC-VRF can now provide 1134 Proxy-ARP/ND functionality for all ARP requests and ND solicitations 1135 directed to the IP address pool learned through BGP. Therefore, the 1136 amount of unnecessary L2 flooding, ARP/ND requests/solicitations in 1137 this case, can be further reduced by the introduction of Proxy-ARP/ND 1138 functionality across all EVI MAC-VRFs. 1140 9.1.3. Unknown Unicast flooding suppression 1142 Given that all locally learned MAC addresses are advertised through 1143 BGP to all remote PEs, suppressing flooding of any Unknown Unicast 1144 traffic towards the remote PEs is a feasible network optimization. 1146 The assumption in the use case is made that any network device that 1147 appears on a remote MAC-VRF will somehow signal its presence to the 1148 network. This signaling can be done through for example gratuitous 1149 ARPs. Once the remote PE acknowledges the presence of the node in the 1150 MAC-VRF, it will do two things: install its MAC address in its local 1151 FIB and advertise this MAC address to all other BGP speakers via EVPN 1152 NLRI. Therefore, we can assume that any active MAC address is 1153 propagated and learnt through the entire EVI. Given that MAC 1154 addresses become pre-populated - once nodes are alive on the network 1155 - there is no need to flood any unknown unicast towards the remote 1156 PEs. If the owner of a given destination MAC is active, the BGP route 1157 will be present in the local RIB and FIB, assuming that the BGP 1158 import policies are successfully applied; otherwise, the owner of 1159 such destination MAC is not present on the network. 1161 It is worth noting that unless: a) control or management plane 1162 learning is performed through the entire EVI or b) all the EVI- 1163 attached devices signal their presence when they come up (GARPs or 1164 similar), unknown unicast flooding must be enabled. 1166 9.1.4. Optimization of Inter-subnet forwarding 1168 In a scenario in which both L2 and L3 services are needed over the 1169 same physical topology, some interaction between EVPN and IP-VPN is 1170 required. A common way of stitching the two service planes is through 1171 the use of an IRB interface, which allows for traffic to be either 1172 routed or bridged depending on its destination MAC address. If the 1173 destination MAC address is the one of the IRB interface, traffic 1174 needs to be passed through a routing module and potentially be either 1175 routed to a remote PE or forwarded to a local subnet. If the 1176 destination MAC address is not the one of the IRB, the MAC-VRF 1177 follows standard bridging procedures. 1179 A typical example of EVPN inter-subnet forwarding would be a scenario 1180 in which multiple IP subnets are part of a single or multiple EVIs, 1181 and they all belong to a single IP-VPN. In such topologies, it is 1182 desired that inter-subnet traffic can be efficiently routed without 1183 any tromboning effects in the network. Due to the overlapping 1184 physical and service topology in such scenarios, all inter-subnet 1185 connectivity will be locally routed through the IRB interface. 1187 In addition to optimizing the traffic patterns in the network, local 1188 inter-subnet forwarding also optimizes greatly the amount of 1189 processing needed to cross the subnets. Through EVPN MAC 1190 advertisements, the local PE learns the real destination MAC address 1191 associated with the remote IP address and the inter-subnet forwarding 1192 can happen locally. When the packet is received at the egress PE, it 1193 is directly mapped to an egress MAC-VRF, bypassing any egress IP-VPN 1194 processing. 1196 Please refer to [EVPN-INTERSUBNET] for more information about the IP 1197 inter-subnet forwarding procedures in EVPN. 1199 9.2. Packet Walkthrough Examples 1201 Assuming that the services are setup according to figure 1 in section 1202 3, the following flow optimization processes will take place in terms 1203 of creating, receiving and forwarding packets across the network. 1205 9.2.1. Proxy-ARP example for CE2 to CE3 traffic 1207 Using Figure 1 in section 3, consider EVI 400 residing on PE1, PE2 1208 and PE3 connecting CE2 and CE3 networks. Also, consider that PE1 and 1209 PE2 are part of the all-active multi-homing ES for CE2, and that PE2 1210 is elected designated-forwarder for EVI400. We assume that all the 1211 PEs implement the proxy-ARP functionality in the MAC-VRF-400 context. 1213 In this scenario, PE3 will not only advertise the MAC addresses 1214 through the EVPN MAC Advertisement Route but also IP addresses of 1215 individual hosts, i.e. /32 prefixes, behind CE3. Upon receiving the 1216 EVPN routes, PE1 and PE2 will install the MAC addresses in the MAC- 1217 VRF-400 FIB and based on the associated received IP addresses, PE1 1218 and PE2 can now build a proxy-ARP table within the context of MAC- 1219 VRF-400. 1221 From the forwarding perspective, when a node behind CE2 sends a frame 1222 destined to a node behind CE3, it will first send an ARP request to 1223 for example PE2 (based on the result of the CE2 hashing). Assuming 1224 that PE2 has populated its proxy-ARP table for all active nodes 1225 behind the CE3, and that the IP address in the ARP message matches 1226 the entry in the table, PE2 will respond to the ARP request with the 1227 actual MAC address on behalf of the node behind CE3. 1229 Once the nodes behind CE2 learn the actual MAC address of the nodes 1230 behind CE3, all the MAC-to-MAC communications between the two 1231 networks will be unicast. 1233 9.2.2. Flood suppression example for CE1 to CE3 traffic 1235 Using Figure 1 in section 3, consider EVI 500 residing on PE1 and PE3 1236 connecting CE1 and CE3 networks. Consider that both PE1 and PE3 have 1237 disabled unknown unicast flooding for this specific EVI context. Once 1238 the network devices behind CE3 come online they will learn their MAC 1239 addresses and create local FIB entries for these devices. Note that 1240 local FIB entries could also be created through either a control or 1241 management plane between PE and CE as well. Consequently, PE3 will 1242 automatically create EVPN Type 2 MAC Advertisement Routes and 1243 advertise all locally learned MAC addresses. The routes will also 1244 include the corresponding MPLS label. 1246 Given that PE1 automatically learns and installs all MAC addresses 1247 behind CE3, its MAC-VRF FIB will already be pre-populated with the 1248 respective next-hops and label assignments associated with the MAC 1249 addresses behind CE3. As such, as soon as the traffic sent by CE1 to 1250 nodes behind CE3 is received into the context of EVI 500, PE1 will 1251 push the MPLS Label(s) onto the original Ethernet frame and send the 1252 packet to the MPLS network. As usual, once PE3 receives this packet, 1253 and depending on the forwarding model, PE3 will either do a next-hop 1254 lookup in the EVI 500 context, or will just forward the traffic 1255 directly to the CE3. In the case that PE1 MAC-VRF-500 does not have a 1256 MAC entry for a specific destination that CE1 is trying to reach, PE1 1257 will drop the frame since unknown unicast flooding is disabled. 1259 Based on the assumption that all the MAC entries behind the CEs are 1260 pre-populated through gratuitous-ARP and/or DHCP requests, if one 1261 specific MAC entry is not present in the MAC-VRF-500 FIB on PE1, the 1262 owner of that MAC is not alive on the network behind the CE3, hence 1263 the traffic can be dropped at PE1 instead of be flooded and consume 1264 network bandwidth. 1266 9.2.3. Optimization of inter-subnet forwarding example for CE3 to CE2 1267 traffic 1269 Using Figure 1 in section 3 consider that there is an IP-VPN 666 1270 context residing on PE1, PE2 and PE3 which connects CE1, CE2 and CE3 1271 into a single IP-VPN domain. Also consider that there are two EVIs 1272 present on the PEs, EVI 600 and EVI 60. Each IP subnet is associated 1273 to a different MAC-VRF context. Thus there is a single subnet, subnet 1274 600, between CE1 and CE3 that is established through EVI 600. 1275 Similarly, there is another subnet, subnet 60, between CE2 and CE3 1276 that is established through EVI 60. Since both subnets are part of 1277 the same IP VPN, there is a mapping of each EVI (or individual 1278 subnet) to a local IRB interface on the three PEs. 1280 If a node behind CE2 wants to communicate with a node on the same 1281 subnet seating behind CE3, the communication flow will follow the 1282 standard EVPN procedures, i.e. FIB lookup within the PE1 (or PE2) 1283 after adding the corresponding EVPN label to the MPLS label stack 1284 (downstream label allocation from PE3 for EVI 60). 1286 When it comes to crossing the subnet boundaries, the ingress PE 1287 implements local inter-subnet forwarding. For example, when a node 1288 behind CE2 (EVI 60) sends a packet to a node behind CE1 (EVI 600) the 1289 destination IP address will be in the subnet 600, but the destination 1290 MAC address will be the address of source node's default gateway, 1291 which in this case will be an IRB interface on PE1 (connecting EVI 60 1292 to IP-VPN 666). Once PE1 sees the traffic destined to its own MAC 1293 address, it will route the packet to EVI 600, i.e. it will change the 1294 source MAC address to the one of the IRB interface in EVI 600 and 1295 change the destination MAC address to the address belonging to the 1296 node behind CE1, which is already populated in the MAC-VRF-600 FIB, 1297 either through data or control plane learning. 1299 An important optimization to be noted is the local inter-subnet 1300 forwarding in lieu of IP VPN routing. If the node from subnet 60 1301 (behind CE2) is sending a packet to the remote end node on subnet 600 1302 (behind CE3), the mechanism in place still honors the local inter- 1303 subnet (inter-EVI) forwarding. 1305 In our use-case, therefore, when node from subnet 60 behind CE2 sends 1306 traffic to the node on subnet 600 behind CE3, the destination MAC 1307 address is the PE1 MAC-VRF-60 IRB MAC address. However, once the 1308 traffic locally crosses EVIs, to EVI 600, via the IRB interface on 1309 PE1, the source MAC address is changed to that of the IRB interface 1310 and the destination MAC address is changed to the one advertised by 1311 PE3 via EVPN and already installed in MAC-VRF-600. The rest of the 1312 forwarding through PE1 is using the MAC-VRF-600 forwarding context 1313 and label space. 1315 Another very relevant optimization is due to the fact that traffic 1316 between PEs is forwarded through EVPN, rather than through IP-VPN. In 1317 the example described above for traffic from EVI 60 on CE2 to EVI 600 1318 on CE3, there is no need for IP-VPN processing on the egress PE3. 1319 Traffic is forwarded either to the EVI 600 context in PE3 for further 1320 MAC lookup and next-hop processing, or directly to the node behind 1321 CE3, depending on the egress forwarding model being used. 1323 10. Security Considerations 1325 Please refer to the "Security Considerations" section in [RFC7432]. 1326 The standards produced by the SIDR WG address secure route origin 1327 authentication (e.g., RFCs 6480-93) and route advertisement security 1328 (e.g., RFCs 8205-11). They protect the integrity and authenticity of 1329 IP address advertisements and ASN/IP prefix bindings. This document, 1330 and [RFC7432], use BGP to convey other info, e.g., MAC addresses, 1331 and thus the protections offered by the SIDR WG RFCs are not 1332 applicable in this context. 1334 11. IANA Considerations 1336 No IANA considerations are needed. 1338 12. References 1340 12.1. Normative References 1342 [RFC7209] Sajassi, A., Aggarwal, R., Uttaro, J., Bitar, N., 1343 Henderickx, W., and A. Isaac, "Requirements for Ethernet VPN (EVPN)", 1344 RFC 7209, DOI 10.17487/RFC7209, May 2014, . 1347 [RFC7432] Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 1348 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based Ethernet 1349 VPN", RFC 7432, DOI 10.17487/RFC7432, February 2015, . 1352 12.2. Informative References 1354 [EVPN-INTERSUBNET] Sajassi et al., "IP Inter-subnet forwarding in 1355 EVPN", draft-ietf-bess-evpn-inter-subnet-forwarding-03.txt 1357 [RFC4761] Kompella, K., Ed., and Y. Rekhter, Ed., "Virtual Private 1358 LAN Service (VPLS) Using BGP for Auto-Discovery and Signaling", 1359 RFC 4761, DOI 10.17487/RFC4761, January 2007, . 1362 [RFC4762] Lasserre, M., Ed., and V. Kompella, Ed., "Virtual Private 1363 LAN Service (VPLS) Using Label Distribution Protocol (LDP) 1364 Signaling", RFC 4762, DOI 10.17487/RFC4762, January 2007, 1365 . 1367 [RFC6074] Rosen, E., Davie, B., Radoaca, V., and W. Luo, 1368 "Provisioning, Auto-Discovery, and Signaling in Layer 2 Virtual 1369 Private Networks (L2VPNs)", RFC 6074, DOI 10.17487/RFC6074, January 1370 2011, . 1372 13. Acknowledgments 1374 The authors want to thank Giles Heron for his detailed review of the 1375 document. We also thank Stefan Plug, and Eric Wunan for their 1376 comments. 1378 14. Contributors 1380 In addition to the authors listed on the front page, the following 1381 co-authors have also contributed to this document: 1383 Florin Balus 1384 Keyur Patel 1385 Aldrin Isaac 1386 Truman Boyes 1388 15. Authors' Addresses 1390 Jorge Rabadan 1391 Nokia 1392 777 E. Middlefield Road 1393 Mountain View, CA 94043 USA 1394 Email: jorge.rabadan@nokia.com 1396 Senad Palislamovic 1397 Nokia 1398 Email: senad.palislamovic@nokia.com 1400 Wim Henderickx 1401 Nokia 1402 Email: wim.henderickx@nokia.com 1404 Ali Sajassi 1405 Cisco 1406 Email: sajassi@cisco.com 1408 James Uttaro 1409 AT&T 1410 Email: uttaro@att.com