idnits 2.17.1 draft-ietf-bess-evpn-usage-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([RFC7432]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (March 13, 2017) is 2593 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC7432' is mentioned on line 1278, but not defined == Missing Reference: 'RFC4761' is mentioned on line 242, but not defined == Missing Reference: 'RFC4762' is mentioned on line 242, but not defined == Missing Reference: 'RFC6074' is mentioned on line 242, but not defined == Missing Reference: 'RFC7209' is mentioned on line 255, but not defined == Missing Reference: 'PE-IP' is mentioned on line 357, but not defined == Missing Reference: 'AS' is mentioned on line 363, but not defined == Missing Reference: 'RFC2119' is mentioned on line 1271, but not defined == Outdated reference: A later version (-15) exists of draft-ietf-bess-evpn-inter-subnet-forwarding-03 Summary: 1 error (**), 0 flaws (~~), 10 warnings (==), 1 comment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 J. Uttaro A. Sajassi 8 AT&T Cisco 10 K. Patel 11 Arrcus 13 T. Boyes A. Isaac 14 Bloomberg Juniper 16 Expires: September 14, 2017 March 13, 2017 18 Usage and applicability of BGP MPLS based Ethernet VPN 19 draft-ietf-bess-evpn-usage-04 21 Abstract 23 This document discusses the usage and applicability of BGP MPLS based 24 Ethernet VPN (EVPN) in a simple and fairly common deployment 25 scenario. The different EVPN procedures will be explained on the 26 example scenario, analyzing the benefits and trade-offs of each 27 option. Along with [RFC7432], this document is intended to provide a 28 simplified guide for the deployment of EVPN in Service Provider 29 networks. 31 Status of this Memo 33 This Internet-Draft is submitted in full conformance with the 34 provisions of BCP 78 and BCP 79. 36 Internet-Drafts are working documents of the Internet Engineering 37 Task Force (IETF), its areas, and its working groups. Note that 38 other groups may also distribute working documents as Internet- 39 Drafts. 41 Internet-Drafts are draft documents valid for a maximum of six months 42 and may be updated, replaced, or obsoleted by other documents at any 43 time. It is inappropriate to use Internet-Drafts as reference 44 material or to cite them other than as "work in progress." 45 The list of current Internet-Drafts can be accessed at 46 http://www.ietf.org/ietf/1id-abstracts.txt 48 The list of Internet-Draft Shadow Directories can be accessed at 49 http://www.ietf.org/shadow.html 51 This Internet-Draft will expire on September 20, 2016. 53 Copyright Notice 55 Copyright (c) 2017 IETF Trust and the persons identified as the 56 document authors. All rights reserved. 58 This document is subject to BCP 78 and the IETF Trust's Legal 59 Provisions Relating to IETF Documents 60 (http://trustee.ietf.org/license-info) in effect on the date of 61 publication of this document. Please review these documents 62 carefully, as they describe your rights and restrictions with respect 63 to this document. Code Components extracted from this document must 64 include Simplified BSD License text as described in Section 4.e of 65 the Trust Legal Provisions and are provided without warranty as 66 described in the Simplified BSD License. 68 Table of Contents 70 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 71 2. Use-case scenario description . . . . . . . . . . . . . . . . . 4 72 3. Provisioning Model . . . . . . . . . . . . . . . . . . . . . . 6 73 3.1. Common provisioning tasks . . . . . . . . . . . . . . . . . 6 74 3.1.1. Non-service specific parameters . . . . . . . . . . . . 7 75 3.1.2. Service specific parameters . . . . . . . . . . . . . . 7 76 3.2. Service interface dependent provisioning tasks . . . . . . 8 77 3.2.1. VLAN-based service interface EVI . . . . . . . . . . . 8 78 3.2.2. VLAN-bundle service interface EVI . . . . . . . . . . . 9 79 3.2.3. VLAN-aware bundling service interface EVI . . . . . . . 9 80 4. BGP EVPN NLRI usage . . . . . . . . . . . . . . . . . . . . . . 9 81 5. MAC-based forwarding model use-case . . . . . . . . . . . . . . 10 82 5.1. EVPN Network Startup procedures . . . . . . . . . . . . . . 10 83 5.2. VLAN-based service procedures . . . . . . . . . . . . . . . 11 84 5.2.1. Service startup procedures . . . . . . . . . . . . . . 11 85 5.2.2. Packet walkthrough . . . . . . . . . . . . . . . . . . 12 86 5.3. VLAN-bundle service procedures . . . . . . . . . . . . . . 15 87 5.3.1. Service startup procedures . . . . . . . . . . . . . . 15 88 5.3.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 16 89 5.4. VLAN-aware bundling service procedures . . . . . . . . . . 16 90 5.4.1. Service startup procedures . . . . . . . . . . . . . . 16 91 5.4.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 17 92 6. MPLS-based forwarding model use-case . . . . . . . . . . . . . 18 93 6.1. Impact of MPLS-based forwarding on the EVPN network 94 startup . . . . . . . . . . . . . . . . . . . . . . . . . . 19 95 6.2. Impact of MPLS-based forwarding on the VLAN-based service 96 procedures . . . . . . . . . . . . . . . . . . . . . . . . 19 97 6.3. Impact of MPLS-based forwarding on the VLAN-bundle 98 service procedures . . . . . . . . . . . . . . . . . . . . 19 99 6.4. Impact of MPLS-based forwarding on the VLAN-aware service 100 procedures . . . . . . . . . . . . . . . . . . . . . . . . 20 101 7. Comparison between MAC-based and MPLS-based forwarding models . 21 102 8. Traffic flow optimization . . . . . . . . . . . . . . . . . . . 22 103 8.1. Control Plane Procedures . . . . . . . . . . . . . . . . . 22 104 8.1.1. MAC learning options . . . . . . . . . . . . . . . . . 22 105 8.1.2. Proxy-ARP/ND . . . . . . . . . . . . . . . . . . . . . 23 106 8.1.3. Unknown Unicast flooding suppression . . . . . . . . . 23 107 8.1.4. Optimization of Inter-subnet forwarding . . . . . . . . 24 108 8.2. Packet Walkthrough Examples . . . . . . . . . . . . . . . . 24 109 8.2.1. Proxy-ARP example for CE2 to CE3 traffic . . . . . . . 25 110 8.2.2. Flood suppression example for CE1 to CE3 traffic . . . 25 111 8.2.3. Optimization of inter-subnet forwarding example for 112 CE3 to CE2 traffic . . . . . . . . . . . . . . . . . . 26 113 9. Conventions used in this document . . . . . . . . . . . . . . . 27 114 10. Security Considerations . . . . . . . . . . . . . . . . . . . 28 115 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 116 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 28 117 12.1. Normative References . . . . . . . . . . . . . . . . . . . 28 118 12.2. Informative References . . . . . . . . . . . . . . . . . . 29 119 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 29 120 14. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 29 121 14. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 29 123 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 2. 129 After describing the topology of the use-case scenario and the 130 characteristics of the service to be deployed, section 3 will 131 describe the provisioning model, comparing the EVPN procedures with 132 the provisioning tasks required for other VPN technologies, such as 133 VPLS or IP-VPN. 135 Once the provisioning model is analyzed, sections 4, 5 and 6 will 136 describe the control plane and data plane procedures in the example 137 scenario, for the two potential disposition/forwarding models: 138 MAC-based and MPLS-based models. While both models can interoperate 139 in the same network, each one has different trade-offs that are 140 analyzed in section 7. 142 Finally, EVPN provides some potential traffic flow optimization tools 143 that are also described in section 8, in the context of the example 144 scenario. 146 2. Use-case scenario description 148 The following figure depicts the scenario that will be referenced 149 throughout the rest of the document. 151 +--------------+ 152 | | 153 +----+ +----+ | | +----+ +----+ 154 | CE1|-----| | | | | |---| CE3| 155 +----+ /| PE1| | IP/MPLS | | PE3| +----+ 156 / +----+ | Network | +----+ 157 / | | 158 / +----+ | | 159 +----+/ | | | | 160 | CE2|-----| PE2| | | 161 +----+ +----+ | | 162 +--------------+ 164 Figure 1 EVPN use-case scenario 166 There are three PEs and three CEs considered in this example: PE1, 167 PE2, PE3, as well as CE1, CE2 and CE3. Layer-2 traffic must be 168 extended among the three CEs. The following service requirements are 169 assumed in this scenario: 171 o Redundancy requirements: CE1 and CE3 are single-homed to PE1 and 172 PE3 respectively. CE2 requires multi-homing connectivity to PE1 and 173 PE2, not only for redundancy purposes, but also for adding more 174 upstream/downstream connectivity bandwidth to/from the network. If 175 CE2 has a single CE-VID (or a few CE-VIDs) the current VPLS 176 multi-homing solutions (based on load-balancing per CE-VID or 177 service) do not provide the optimized link utilization required in 178 this example. Another redundancy requirement that must be met is 179 fast convergence. E.g.: if the link between CE2 and PE1 goes down, 180 a fast convergence mechanism must be supported so that PE3 can 181 immediately send the traffic to PE2, irrespectively of the number 182 of affected services and MAC addresses. EVPN provides the 183 flow-based load-balancing multi-homing solution required in this 184 scenario to optimize the upstream/downstream link utilization 185 between CE2 and PE1-PE2. EVPN also provides a fast convergence 186 solution so that PE3 can immediately send the traffic to PE2 upon 187 failure on the link between CE2 and PE1. 189 o Service interface requirements: service definition must be flexible 190 in terms of CE-VID-to-broadcast-domain assignment and service 191 contexts in the core. The following three services are required in 192 this example: 194 EVI100 - It will use VLAN-based service interfaces in the three CEs 195 with a 1:1 mapping (VLAN-to-EVI). The CE-VIDs at the three CEs can 196 be the same, e.g.: VID 100, or different at each CE, e.g.: VID 101 197 in CE1, VID 102 in CE2 and VID 103 in CE3. A single broadcast 198 domain needs to be created for EVI100 in any case; therefore CE- 199 VIDs will require translation at the egress PEs if they are not 200 consistent across the three CEs. The case when the same CE-VID is 201 used across the three CEs for EVI100 is referred in [RFC7432] as 202 the "Unique VLAN" EVPN case. This term will be used throughout this 203 document too. 205 EVI200 - It will use VLAN-bundle service interfaces in CE1, CE2 and 206 CE3, based on an N:1 VLAN-to-EVI mapping. In this case, the service 207 provider just needs to assign a pre-configured number of CE-VIDs on 208 the ingress PE to EVI200, and send the customer frames with the 209 original CE-VIDs. The Service Provider will build a single 210 broadcast domain for the customer. The customer will be responsible 211 for the CE-VID handling. 213 EVI300 - It will use VLAN-aware bundling service interfaces in CE1, 214 CE2 and CE3. At the ingress PE, an N:1 VLAN-to-EVI mapping will be 215 done, however and as opposed to EVI200, a separate core broadcast 216 domain is required per CE-VID. In addition to that, the CE-VIDs can 217 be different (hence CE-VID translation is required). Note that, 218 while the requirements stated for EVI100 and EVI200 might be met 219 with the current VPLS solutions, the VLAN-aware bundling service 220 interfaces required by EVI300 are not supported by the current VPLS 221 tools. 223 NOTE: in section 3.2.1, only EVI100 is used as an example of 224 VLAN-based service provisioning. In sections 5.2 and 6.2, 4k 225 VLAN-based EVIs (EVI1 to EVI4k) are used so that the impact of MAC 226 vs. MPLS disposition models in the control plane can be evaluated. In 227 the same way, EVI200 and EVI300 will be described with a 4k:1 mapping 228 (CE-VIDs-to-EVI mapping) in sections 5.3-4 and 6.3-4. 230 o BUM (Broadcast, Unknown unicast, Multicast) optimization 231 requirements: The solution must be able to support ingress 232 replication and P2MP MPLS LSPs and the user must be able to decide 233 what kind of provider tree will be used by each EVI service. For 234 example, if we assume that EVI100 and EVI200 will not carry much 235 BUM traffic, we can use ingress replication for those service 236 instances. The benefit is that the core will not need to maintain 237 any states for the multicast trees associated to EVI100 and EVI200. 238 On the contrary, if EVI300 is presumably carrying a significant 239 amount of multicast traffic, P2MP MPLS LSPs can be used for this 240 service. 242 The current VPLS solutions, based on [RFC4761][RFC4762][RFC6074], 243 cannot meet all the above set of requirements and therefore a new 244 solution is needed. The rest of the document will describe how EVPN 245 can be used to meet those service requirements and even optimize the 246 network further by: 248 o Providing the user with an option to reduce (and even suppress) the 249 ARP-flooding. 251 o Supporting ARP termination for inter-subnet forwarding 253 3. Provisioning Model 255 One of the requirements stated in [RFC7209] is the ease of 256 provisioning. BGP parameters and service context parameters should be 257 auto-provisioned so that the addition of a new MAC-VRF to the EVI 258 requires a minimum number of single-sided provisioning touches. 259 However this is only possible in a limited number of cases. This 260 section describes the provisioning tasks required for the services 261 described in section 2, i.e. EVI100 (VLAN-based service interfaces), 262 EVI200 (VLAN-bundle service interfaces) and EVI300 (VLAN-aware 263 bundling service interfaces). 265 3.1. Common provisioning tasks 266 Regardless of the service interface type (VLAN-based, VLAN-bundle or 267 VLAN-aware), the following sub-sections describe the parameters to be 268 provisioned in the three PEs. 270 3.1.1. Non-service specific parameters 272 The multi-homing function in EVPN requires the provisioning of 273 certain parameters which are not service-specific and that are shared 274 by all the MAC-VRFs in the node using the multi-homing capabilities. 275 In our use-case, these parameters are only provisioned in PE1 and 276 PE2, and are listed below: 278 o Ethernet Segment Identifier (ESI): only the ESI associated to CE2 279 needs to be considered in our example. Single-homed CEs such as CE1 280 and CE3 do not require the provisioning of an ESI (the ESI will be 281 coded as zero in the BGP NLRIs). In our example, a LAG is used 282 between CE2 and PE1-PE2 (since all-active multi-homing is a 283 requirement) therefore the ESI can be auto-derived from the LACP 284 information as described in [RFC7432]. Note that the ESI MUST be 285 unique across all the PEs in the network, therefore the 286 auto-provisioning of the ESI is only recommended in case the CEs 287 are managed by the Service Provider. Otherwise the ESI should be 288 manually provisioned (type 0 as in [RFC7432]) in order to avoid 289 potential conflicts. 291 o ES-Import Route Target (ES-Import RT): this is the RT that will be 292 sent by PE1 and PE2, along with the ES route. Regardless of how the 293 ESI is provisioned in PE1 and PE2, the ES-Import RT must always be 294 auto-derived from the 6-byte MAC address portion of the ESI value. 296 o Ethernet Segment Route Distinguisher (ES RD): this is the RD to be 297 encoded in the ES route and Ethernet Auto-Discovery (A-D) route to 298 be sent by PE1 and PE2 for the CE2 ESI. This RD should always be 299 auto-derived from the PE IP address, as described in [RFC7432]. 301 o Multi-homing type: the user must be able to provision the 302 multi-homing type to be used in the network. In our use-case, the 303 multi-homing type will be set to all-active for the CE2 ESI. This 304 piece of information is encoded in the ESI Label extended community 305 flags and sent by PE1 and PE2 along with the Ethernet A-D route for 306 the CE2 ESI. 308 In our use-case, besides the above parameters, the same LACP 309 parameters will be configured in PE1 and PE2 for the ESI, so that CE2 310 can send different flows to PE1 and PE2 for the same CE-VID as though 311 they were forming a single system from the CE2 perspective. 313 3.1.2. Service specific parameters 314 The following parameters must be provisioned in PE1, PE2 and PE3 per 315 EVI service: 317 o EVI identifier: global identifier per EVI that is shared by all the 318 PEs part of the EVI, i.e. PE1, PE2 and PE3 will be provisioned with 319 EVI100, 200 and 300. The EVI identifier can be associated to (or be 320 the same value as) the EVI default Ethernet Tag (4-byte default 321 broadcast domain identifier for the EVI). The Ethernet Tag is 322 different from zero in the EVPN BGP routes only if the service 323 interface type (of the source PE) is VLAN-aware. 325 o EVI Route Distinguisher (EVI RD): This RD is a unique value across 326 all the MAC-VRFs in a PE. Auto-derivation of this RD might be 327 possible depending on the service interface type being used in the 328 EVI. Next section discusses the specifics of each service interface 329 type. 331 o EVI Route Target(s) (EVI RT): one or more RTs can be provisioned 332 per MAC-VRF. The RT(s) imported and exported can be equal or 333 different, just as the RT(s) in IP-VPNs. Auto-derivation of this 334 RT(s) might be possible depending on the service interface type 335 being used in the EVI. Next section discusses the specifics of each 336 service interface type. 338 o CE-VID and port/LAG binding to EVI identifier or Ethernet Tag: see 339 section 3.2. 341 3.2. Service interface dependent provisioning tasks 343 Depending on the service interface type being used in the EVI, a 344 specific CE-VID binding provisioning must be specified. 346 3.2.1. VLAN-based service interface EVI 348 In our use-case, EVI100 is a VLAN-based service interface EVI. 350 EVI100 can be a "unique-VLAN" EVPN if the CE-VID being used for this 351 service in CE1, CE2 and CE3 is equal, e.g. VID 100. In that case, the 352 VID 100 binding must be provisioned in PE1, PE2 and PE3 for EVI100 353 and the associated port or LAG. The MAC-VRF RD and RT can be auto- 354 derived from the CE-VID: 356 o The auto-derived MAC-VRF RD will be a Type 1 RD, as recommended in 357 [RFC7432], and it will be comprised of [PE-IP]:[zero-padded-VID]; 358 where PE-IP is the IP address of the PE (a loopback address) and 359 [zero-padded-VID] is a 2-byte value where the low order 12 bits are 360 the VID (VID 100 in our example) and the high order 4 bits are 361 zero. 363 o The auto-derived MAC-VRF RT will be composed of [AS]:[zero-padded- 364 VID]; where AS is the Autonomous System that the PE belongs to and 365 [zero-padded-VID] is a 2 or 4-byte value where the low order 12 366 bits are the VID (VID 100 in our example) and the high order bits 367 are zero. Note that auto-deriving the RT implies supporting a basic 368 any-to-any topology in the EVI and using the same import and export 369 RT in the EVI. 371 If EVI100 is not a "unique-VLAN" EVPN, each individual CE-VID must be 372 configured in each PE, and MAC-VRF RDs and RTs cannot be auto- 373 derived, hence they must be provisioned by the user. 375 3.2.2. VLAN-bundle service interface EVI 377 Assuming EVI200 is a VLAN-bundle service interface EVI, and VIDs 378 200-250 are assigned to EVI200, the CE-VID bundle 200-250 must be 379 provisioned on PE1, PE2 and PE3. Note that this model does not allow 380 CE-VID translation and the CEs must use the same CE-VIDs for EVI200. 381 No auto-derived EVI RDs or EVI RTs are possible. 383 3.2.3. VLAN-aware bundling service interface EVI 385 If EVI300 is a VLAN-aware bundling service interface EVI, CE-VID 386 binding to EVI300 does not have to match on the three PEs (only on 387 PE1 and PE2, since they are part of the same ES). E.g.: PE1 and PE2 388 CE-VID binding to EVI300 can be set to the range 300-310 and PE3 to 389 321-330. Note that each individual CE-VID will be assigned to a core 390 broadcast domain, i.e. Ethernet Tag, which will be encoded in the BGP 391 EVPN routes. 393 Therefore, besides the CE-VID bundle range bound to EVI300 in each 394 PE, associations between each individual CE-VID and the EVPN Ethernet 395 Tag must be provisioned by the user. No auto-derived EVI RDs/RTs are 396 possible. 398 4. BGP EVPN NLRI usage 400 [RFC7432] defines four different types of routes and four different 401 extended communities advertised along with the different routes. 402 However not all the PEs in a network must generate and process all 403 the different routes and extended communities. The following table 404 shows the routes that must be exported and imported in the use-case 405 described in this document. "Export", in this context, means that the 406 PE must be capable of generating and exporting a given route, 407 assuming there are no BGP policies to prevent it. In the same way, 408 "Import" means the PE must be capable of importing and processing a 409 given route, assuming the right RTs and policies. "N/A" means neither 410 import nor export actions are required. 412 +-------------------+---------------+---------------+ 413 | BGP EVPN routes | PE1-PE2 | PE3 | 414 +-------------------+---------------+---------------+ 415 | ES | Export/import | N/A | 416 | A-D per ESI | Export/import | Import | 417 | A-D per EVI | Export/import | Import | 418 | MAC | Export/import | Export/import | 419 | Inclusive mcast | Export/import | Export/import | 420 +-------------------+---------------+---------------+ 422 PE3 is only required to export MAC and Inclusive multicast routes and 423 be able to import and process A-D routes, as well as MAC and 424 Inclusive multicast routes. If PE3 did not support importing and 425 processing A-D routes per ESI and per EVI, fast convergence and 426 aliasing functions (respectively) would not be possible in this 427 use-case. 429 5. MAC-based forwarding model use-case 431 This section describes how the BGP EVPN routes are exported and 432 imported by the PEs in our use-case, as well as how traffic is 433 forwarded assuming that PE1, PE2 and PE3 support a MAC-based 434 forwarding model. In order to compare the control and data plane 435 impact in the two forwarding models (MAC-based and MPLS-based) and 436 different service types, we will assume that CE1, CE2 and CE3 need to 437 exchange traffic for up to 4k CE-VIDs. 439 5.1. EVPN Network Startup procedures 441 Before any EVI is provisioned in the network, the following 442 procedures are required: 444 o Infrastructure setup: the proper MPLS infrastructure must be setup 445 among PE1, PE2 and PE3 so that the EVPN services can make use of 446 P2P and P2MP LSPs. In addition to the MPLS transport, PE1 and PE2 447 must be properly configured with the same LACP configuration to 448 CE2. Details are provided in [RFC7432]. Once the LAG is properly 449 setup, the ESI for the CE2 Ethernet Segment, e.g. ESI12, can be 450 auto-generated by PE1 and PE2 from the LACP information exchanged 451 with CE2 (ESI type 1), as discussed in section 3.1. Alternatively, 452 the ESI can also be manually provisioned on PE1 and PE2 (ESI type 453 0). PE1 and PE2 will auto-configure a BGP policy that will import 454 any ES route matching the auto-derived ES-import RT for ESI12. 456 o Ethernet Segment route exchange and DF election: PE1 and PE2 will 457 advertise a BGP Ethernet Segment route for ESI12, where the ESI RD 458 and ES-Import RT will be auto-generated as discussed in section 459 3.1.1. PE1 and PE2 will import the ES routes of each other and will 460 run the DF election algorithm for any existing EVI (if any, at this 461 point). PE3 will simply discard the route. Note that the DF 462 election algorithm can support service carving, so that the 463 downstream BUM traffic from the network to CE2 can be load-balanced 464 across PE1 and PE2 on a per-service basis. 466 At the end of this process, the network infrastructure is ready to 467 start deploying EVPN services. PE1 and PE2 are aware of the existence 468 of a shared Ethernet Segment, i.e. ESI12. 470 5.2. VLAN-based service procedures 472 Assuming that the EVPN network must carry traffic among CE1, CE2 and 473 CE3 for up to 4k CE-VIDs, the Service Provider can decide to 474 implement VLAN-based service interface EVIs to accomplish it. In this 475 case, each CE-VID will be individually mapped to a different EVI. 476 While this means a total number of 4k MAC-VRFs is required per PE, 477 the advantages of this approach are the auto-provisioning of most of 478 the service parameters if no VLAN translation is needed (see section 479 3.2.1) and great control over each individual customer broadcast 480 domain. We assume in this section that the range of EVIs from 1 to 4k 481 is provisioned in the network. 483 5.2.1. Service startup procedures 485 As soon as the EVIs are created in PE1, PE2 and PE3, the following 486 control plane actions are carried out: 488 o Flooding tree setup per EVI (4k routes): Each PE will send one 489 Inclusive Multicast Ethernet Tag route per EVI (up to 4k routes per 490 PE) so that the flooding tree per EVI can be setup. Note that 491 ingress replication or P2MP LSPs can optionally be signaled in the 492 PMSI Tunnel attribute and the corresponding tree be created. 494 o Ethernet A-D routes per ESI (a set of routes for ESI12): A set of 495 A-D routes with a list of 4k RTs (one per EVI) for ESI12 will be 496 issued from PE1 and PE2 (it has to be a set of routes so that the 497 total number of RTs can be conveyed). As per [RFC7432], each 498 Ethernet A-D route per ESI is differentiated from the other routes 499 in the set by a different Route Distinguisher (ES RD). This set 500 will also include ESI Label extended communities with the active- 501 standby flag set to zero (all-active multi-homing type) and an ESI 502 Label different from zero (used for split-horizon functions). These 503 routes will be imported by the three PEs, since the RTs match the 504 EVI RTs locally configured. The A-D routes per ESI will be used for 505 fast convergence and split-horizon functions, as discussed in 506 [RFC7432]. 508 o Ethernet A-D routes per EVI (4k routes): An A-D route per EVI will 509 be sent by PE1 and PE2 for ESI12. Each individual route includes 510 the corresponding EVI RT and an MPLS label to be used by PE3 for 511 the aliasing function. These routes will be imported by the three 512 PEs. 514 5.2.2. Packet walkthrough 516 Once the services are setup, the traffic can start flowing. Assuming 517 there are no MAC addresses learnt yet and that MAC learning at the 518 access is performed in the data plane in our use-case, this is the 519 process followed upon receiving frames from each CE (example for 520 EVI1). 522 (1) BUM frame example from CE1: 524 a) An ARP-request with CE-VID=1 is issued from source MAC CE1-MAC 525 (MAC address coming from CE1 or from a device connected to CE1) to 526 find the MAC address of CE3-IP. 528 b) Based on the CE-VID, the frame is identified to be forwarded in 529 the MAC-VRF-1 (EVI1) context. A source MAC lookup is done in the 530 MAC FIB and the sender's CE1-IP in the proxy-ARP table within the 531 MAC-VRF-1 (EVI1) context. If CE1-MAC/CE1-IP are unknown in both 532 tables, three actions are carried out (assuming the source MAC is 533 accepted by PE1): (1) a forwarding state is added for CE1-MAC 534 associated to the corresponding port and CE-VID, (2) the ARP- 535 request is snooped and the tuple CE1-MAC/CE1-IP is added to the 536 proxy-ARP table and (3) a BGP MAC advertisement route is triggered 537 from PE1 containing the EVI1 RD and RT, ESI=0, Ethernet-Tag=0 and 538 CE1-MAC/CE1-IP along with an MPLS label assigned to MAC-VRF-1 from 539 the PE1 label space. Note that depending on the implementation, 540 the MAC FIB and proxy-ARP learning processes can independently 541 send two BGP MAC advertisements instead of one (one containing 542 only the CE1-MAC and another one containing CE1-MAC/CE1-IP). 544 Since we assume a MAC forwarding model, a label per MAC-VRF is 545 normally allocated and signaled by the three PEs for MAC 546 advertisement routes. Based on the RT, the route is imported by 547 PE2 and PE3 and the forwarding state plus ARP entry are added to 548 their MAC-VRF-1 context. From this moment on, any ARP request from 549 CE2 or CE3 destined to CE1-IP, can be directly replied by PE1, PE2 550 or PE3 and ARP flooding for CE1-IP is not needed in the core. 552 c) Since the ARP frame is a broadcast frame, it is forwarded by PE1 553 using the Inclusive multicast tree for EVI1 (CE-VID=1 should be 554 kept if translation is required). Depending on the type of tree, 555 the label stack may vary. E.g. assuming ingress replication, the 556 packet is replicated to PE2 and PE3 with the downstream allocated 557 labels and the P2P LSP transport labels. No other labels are added 558 to the stack. 560 d) Assuming PE1 is the DF for EVI1 on ESI12, the frame is locally 561 replicated to CE2. 563 e) The MPLS-encapsulated frame gets to PE2 and PE3. Since PE2 is non- 564 DF for EVI1 on ESI12, and there is no other CE connected to PE2, 565 the frame is discarded. At PE3, the frame is de-encapsulated, CE- 566 VID translated if needed and replicated to CE3. 568 Any other type of BUM frame from CE1 would follow the same 569 procedures. BUM frames from CE3 would follow the same procedures too. 571 (2) BUM frame example from CE2: 573 a) An ARP-request with CE-VID=1 is issued from source MAC CE2-MAC to 574 find the MAC address of CE3-IP. 576 b) CE2 will hash the frame and will forward it to e.g. PE2. Based on 577 the CE-VID, the frame is identified to be forwarded in the EVI1 578 context. A source MAC lookup is done in the MAC FIB and the 579 sender's CE2-IP in the proxy-ARP table within the MAC-VRF-1 580 context. If both are unknown, three actions are carried out 581 (assuming the source MAC is accepted by PE2): (1) a forwarding 582 state is added for CE2-MAC associated to the corresponding LAG/ESI 583 and CE-VID, (2) the ARP-request is snooped and the tuple CE2- 584 MAC/CE2-IP is added to the proxy-ARP table and (3) a BGP MAC 585 advertisement route is triggered from PE2 containing the EVI1 RD 586 and RT, ESI=12, Ethernet-Tag=0 and CE2-MAC/CE2-IP along with an 587 MPLS label assigned from the PE2 label space (one label per MAC- 588 VRF). Again, depending on the implementation, the MAC FIB and 589 proxy-ARP learning processes can independently send two BGP MAC 590 advertisements instead of one. 592 Note that, since PE3 is not part of ESI12, it will install a 593 forwarding state for CE2-MAC as long as the A-D routes for ESI12 594 are also active on PE3. On the contrary, PE1 is part of ESI12, 595 therefore PE1 will not modify the forwarding state for CE2-MAC if 596 it has previously learnt CE2-MAC locally attached to ESI12. 597 Otherwise it will add forwarding state for CE2-MAC associated to 598 the local ESI12 port. 600 c) Assuming PE2 does not have the ARP information for CE3-IP yet, and 601 since the ARP is a broadcast frame and PE2 the non-DF for EVI1 on 602 ESI12, the frame is forwarded by PE2 in the Inclusive multicast 603 tree for EVI1, adding the ESI label for ESI12 at the bottom of the 604 stack. The ESI label has been previously allocated and signaled by 605 the A-D routes for ESI12. Note that, as per [RFC7432], if the 606 result of the CE2 hashing is different and the frame sent to PE1, 607 PE1 SHOULD add the ESI label too (PE1 is the DF for EVI1 on 608 ESI12). 610 d) The MPLS-encapsulated frame gets to PE1 and PE3. PE1 611 de-encapsulate the Inclusive multicast tree label(s) and based on 612 the ESI label at the bottom of the stack, it decides to not 613 forward the frame to the ESI12. It will pop the ESI label and will 614 replicate it to CE1 though, since CE1 is not part of the ESI 615 identified by the ESI label. At PE3, the Inclusive multicast tree 616 label is popped and the frame forwarded to CE3. If a P2MP LSP is 617 used as Inclusive multicast tree for EVI1, PE3 will find an ESI 618 label after popping the P2MP LSP label. The ESI label will simply 619 be ignored and popped, since CE3 is not part of ESI12. 621 (3) Unicast frame example from CE3 to CE1: 623 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 624 and destination MAC CE1-MAC (we assume PE3 has previously resolved 625 an ARP request from CE3 to find the MAC of CE1-IP, and has added 626 CE3-MAC/CE3-IP to its proxy-ARP table). 628 b) Based on the CE-VID, the frame is identified to be forwarded in 629 the EVI1 context. A source MAC lookup is done in the MAC FIB 630 within the MAC-VRF-1 context and this time, since we assume CE3- 631 MAC is known, no further actions are carried out as a result of 632 the source lookup. A destination MAC lookup is performed next and 633 the label stack associated to the MAC CE1-MAC is found (including 634 the label associated to MAC-VRF-1 in PE1 and the P2P LSP label to 635 get to PE1). The unicast frame is then encapsulated and forwarded 636 to PE1. 638 c) At PE1, the packet is identified to be part of EVI1 and a 639 destination MAC lookup is performed in the MAC-VRF-1 context. The 640 labels are popped and the frame forwarded to CE1 with CE-VID=1. 642 Unicast frames from CE1 to CE3 or from CE2 to CE3 follow the same 643 procedures described above. 645 (4) Unicast frame example from CE3 to CE2: 647 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 648 and destination MAC CE2-MAC (we assume PE3 has previously resolved 649 an ARP request from CE3 to find the MAC of CE2-IP). 651 b) Based on the CE-VID, the frame is identified to be forwarded in 652 the MAC-VRF-1 context. We assume CE3-MAC is known. A destination 653 MAC lookup is performed next and PE3 finds CE2-MAC associated to 654 PE2 on ESI12, an Ethernet Segment for which PE3 has two active A-D 655 routes per ESI (from PE1 and PE2) and two active A-D routes for 656 EVI1 (from PE1 and PE2). Based on a hashing function for the 657 frame, PE3 may decide to forward the frame using the label stack 658 associated to PE2 (label received from the MAC advertisement 659 route) or the label stack associated to PE1 (label received from 660 the A-D route per EVI for EVI1). Either way, the frame is 661 encapsulated and sent to the remote PE. 663 c) At PE2 (or PE1), the packet is identified to be part of EVI1 based 664 on the bottom label, and a destination MAC lookup is performed. At 665 either PE (PE2 or PE1), the FIB lookup yields a local ESI12 port 666 to which the frame is sent. 668 Unicast frames from CE1 to CE2 follow the same procedures. Aliasing 669 is possible in this case too, since ESI12 is local to PE1 and load 670 balancing through PE1 and PE2 may happen. 672 5.3. VLAN-bundle service procedures 674 Instead of using VLAN-based interfaces, the Service Provider can 675 choose to implement VLAN-bundle interfaces to carry the traffic for 676 the 4k CE-VIDs among CE1, CE2 and CE3. If that is the case, the 4k 677 CE-VIDs can be mapped to the same EVI, e.g. EVI200, at each PE. The 678 main advantage of this approach is the low control plane overhead 679 (reduced number of routes and labels) and easiness of provisioning, 680 at the expense of no control over the customer broadcast domains, 681 i.e. a single inclusive multicast tree for all the CE-VIDs and no CE- 682 VID translation in the Provider network. 684 5.3.1. Service startup procedures 686 As soon as the EVI200 is created in PE1, PE2 and PE3, the following 687 control plane actions are carried out: 689 o Flooding tree setup per EVI (one route): Each PE will send one 690 Inclusive Multicast Ethernet Tag route per EVI (hence only one 691 route per PE) so that the flooding tree per EVI can be setup. Note 692 that ingress replication or P2MP LSPs can optionally be signaled 693 in the PMSI Tunnel attribute and the corresponding tree be 694 created. 696 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 697 route for ESI12 will be issued from PE1 and PE2. This route will 698 include a single RT (RT for EVI200), an ESI Label extended 699 community with the active-standby flag set to zero (all-active 700 multi-homing type) and an ESI Label different from zero (used by 701 the non-DF for split-horizon functions). This route will be 702 imported by the three PEs, since the RT matches the EVI200 RT 703 locally configured. The A-D routes per ESI will be used for fast 704 convergence and split-horizon functions, as described in 705 [RFC7432]. 707 o Ethernet A-D routes per EVI (one route): An A-D route (EVI200) will 708 be sent by PE1 and PE2 for ESI12. This route includes the EVI200 709 RT and an MPLS label to be used by PE3 for the aliasing function. 710 This route will be imported by the three PEs. 712 5.3.2. Packet Walkthrough 714 The packet walkthrough for the VLAN-bundle case is similar to the one 715 described for EVI1 in the VLAN-based case except for the way the 716 CE-VID is handled by the ingress PE and the egress PE: 718 o No VLAN translation is allowed and the CE-VIDs are kept untouched 719 from CE to CE, i.e. the ingress CE-VID MUST be kept at the 720 imposition PE and at the disposition PE. 722 o The frame is identified to be forwarded in the MAC-VRF-200 context 723 as long as its CE-VID belongs to the VLAN-bundle defined in the 724 PE1/PE2/PE3 port to CE1/CE2/CE3. Our example is a special VLAN- 725 bundle case, since the entire CE-VID range is defined in the 726 ports, therefore any CE-VID would be part of EVI200. 728 Please refer to section 5.2.2 for more information about the control 729 plane and forwarding plane interaction for BUM and unicast traffic 730 from the different CEs. 732 5.4. VLAN-aware bundling service procedures 734 The last potential service type analyzed in this document is 735 VLAN-aware bundling. When this type of service interface is used to 736 carry the 4k CE-VIDs among CE1, CE2 and CE3, all the CE-VIDs will be 737 mapped to the same EVI, e.g. EVI300. The difference, compared to the 738 VLAN-bundle service type in the previous section, is that each 739 incoming CE-VID will also be mapped to a different "normalized" 740 Ethernet-Tag in addition to EVI300. If no translation is required, 741 the Ethernet-tag will match the CE-VID. Otherwise a translation 742 between CE-VID and Ethernet-tag will be needed at the imposition PE 743 and at the disposition PE. The main advantage of this approach is the 744 ability to control customer broadcast domains while providing a 745 single EVI to the customer. 747 5.4.1. Service startup procedures 748 As soon as the EVI300 is created in PE1, PE2 and PE3, the following 749 control plane actions are carried out: 751 o Flooding tree setup per EVI per Ethernet-Tag (4k routes): Each PE 752 will send one Inclusive Multicast Ethernet Tag route per EVI and 753 per Ethernet-Tag (hence 4k routes per PE) so that the flooding 754 tree per customer broadcast domain can be setup. Note that ingress 755 replication or P2MP LSPs can optionally be signaled in the PMSI 756 Tunnel attribute and the corresponding tree be created. In the 757 described use-case, since all the CE-VIDs and Ethernet-Tags are 758 defined on the three PEs, multicast tree aggregation might make 759 sense in order to save forwarding states. 761 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 762 route for ESI12 will be issued from PE1 and PE2. This route will 763 include a single RT (RT for EVI300), an ESI Label extended 764 community with the active-standby flag set to zero (all-active 765 multi-homing type) and an ESI Label different than zero (used by 766 the non-DF for split-horizon functions). This route will be 767 imported by the three PEs, since the RT matches the EVI300 RT 768 locally configured. The A-D routes per ESI will be used for fast 769 convergence and split-horizon functions, as described in 770 [RFC7432]. 772 o Ethernet A-D routes per EVI (one route): An A-D route (EVI300) will 773 be sent by PE1 and PE2 for ESI12. This route includes the EVI300 774 RT and an MPLS label to be used by PE3 for the aliasing function. 775 This route will be imported by the three PEs. 777 5.4.2. Packet Walkthrough 779 The packet walkthrough for the VLAN-aware case is similar to the one 780 described before. Compared to the other two cases, VLAN-aware 781 services allow for CE-VID translation and for an N:1 CE-VID to EVI 782 mapping. Both things are not supported at once in either of the two 783 other service interfaces. Note that this model requires qualified 784 learning on the MAC FIBs. Some differences compared to the packet 785 walkthrough described in section 5.2.2 are: 787 o At the ingress PE, the frames are identified to be forwarded in the 788 EVI300 context as long as their CE-VID belong to the range defined 789 in the PE port to the CE. In addition to it, CE-VID=x is mapped to 790 a "normalized" Ethernet-Tag=y at the MAC-VRF-300 (where x and y 791 might be equal if no translation is needed). Qualified learning is 792 now required (a different FIB space is allocated within MAC-VRF- 793 300 for each Ethernet-Tag). Potentially the same MAC could be 794 learnt in two different Ethernet-Tag bridge domains of the same 795 MAC-VRF. 797 o Any new locally learnt MAC on the MAC-VRF-300/Ethernet-Tag=y 798 interface is advertised by the ingress PE in a MAC advertisement 799 route, using now the Ethernet-Tag field (Ethernet-Tag=y) so that 800 the remote PE learns the MAC associated to the MAC-VRF- 801 300/Ethernet-Tag=y FIB. Note that the Ethernet-Tag field is not 802 used in advertisements of MACs learnt on VLAN-based or VLAN-bundle 803 service interfaces. 805 o At the ingress PE, BUM frames are sent to the corresponding 806 flooding tree for the particular Ethernet-Tag they are mapped to. 807 Each individual Ethernet-Tag can have a different flooding tree 808 within the same EVI300. For instance, Ethernet-Tag=y can use 809 ingress replication to get to the remote PEs whereas Ethernet- 810 Tag=z can use a p2mp LSP. 812 o At the egress PE, Ethernet-Tag=y, for a given broadcast domain 813 within MAC-VRF-300, can be translated to egress CE-VID=x. That is 814 not possible for VLAN-bundle interfaces. It is possible for VLAN- 815 based interfaces, but it requires a separate EVI per CE-VID. 817 6. MPLS-based forwarding model use-case 819 EVPN supports an alternative forwarding model, usually referred to as 820 MPLS-based forwarding or disposition model as opposed to the 821 MAC-based forwarding or disposition model described in section 5. 822 Using MPLS-based forwarding model instead of MAC-based model might 823 have an impact on: 825 o The number of forwarding states required 827 o The FIB where the forwarding states are handled: MAC FIB or MPLS 828 LFIB. 830 The MPLS-based forwarding model avoids the destination MAC lookup at 831 the egress PE MAC FIB, at the expense of increasing the number of 832 next-hop forwarding states at the egress MPLS LFIB. This also has an 833 impact on the control plane and the label allocation model, since an 834 MPLS-based disposition PE MUST send as many routes and labels as 835 required next-hops in the egress MAC-VRF. This concept is equivalent 836 to the forwarding models supported in IP-VPNs at the egress PE, where 837 an IP lookup in the IP-VPN FIB might be necessary or not depending on 838 the available next-hop forwarding states in the LFIB. 840 The following sub-sections highlight the impact on the control and 841 data plane procedures described in section 5 when and MPLS-based 842 forwarding model is used. 844 Note that both forwarding models are compatible and interoperable in 845 the same network. The implementation of either model in each PE is a 846 local decision to the PE node. 848 6.1. Impact of MPLS-based forwarding on the EVPN network startup 850 The MPLS-based forwarding model has no impact on the procedures 851 explained in section 5.1. 853 6.2. Impact of MPLS-based forwarding on the VLAN-based service 854 procedures 856 Compared to the MAC-based forwarding model, the MPLS-based forwarding 857 model has no impact in terms of number of routes, when all the 858 service interfaces are VLAN-based. The differences for the use-case 859 described in this document are summarized in the following list: 861 o Flooding tree setup per EVI (4k routes per PE): no impact compared 862 to the MAC-based model. 864 o Ethernet A-D routes per ESI (one set of routes for ESI12 per PE): 865 no impact compared to the MAC-based model. 867 o Ethernet A-D routes per EVI (4k routes per PE/ESI): no impact 868 compared to the MAC-based model. 870 o MAC-advertisement routes: instead of allocating and advertising the 871 same MPLS label for all the new MACs locally learnt on the same 872 MAC-VRF, a different label MUST be advertised per CE next-hop or 873 MAC so that no MAC FIB lookup is needed at the egress PE. In 874 general, this means that a different label at least per CE must be 875 advertised, although the PE can decide to implement a label per 876 MAC if more granularity (hence less scalability) is required in 877 terms of forwarding states. E.g. if CE2 sends traffic from two 878 different MACs to PE1, CE2-MAC1 and CE2-MAC2, the same MPLS 879 label=x can be re-used for both MAC advertisements since they both 880 share the same source ESI12. It is up to the PE1 implementation to 881 use a different label per individual MAC within the same ES 882 Segment (even if only one label per ESI is enough). 884 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 885 learning new local CE MAC addresses on the data plane, but will 886 rather add forwarding states to the MPLS LFIB. 888 6.3. Impact of MPLS-based forwarding on the VLAN-bundle service 889 procedures 891 Compared to the MAC-based forwarding model, the MPLS-based forwarding 892 model has no impact in terms of number of routes when all the service 893 interfaces are VLAN-bundle type. The differences for the use-case 894 described in this document are summarized in the following list: 896 o Flooding tree setup per EVI (one route): no impact compared to the 897 MAC-based model. 899 o Ethernet A-D routes per ESI (one route for ESI12 per PE): no impact 900 compared to the MAC-based model. 902 o Ethernet A-D routes per EVI (one route per PE/ESI): no impact 903 compared to the MAC-based model since no VLAN translation is 904 required. 906 o MAC-advertisement routes: instead of allocating and advertising the 907 same MPLS label for all the new MACs locally learnt on the same 908 MAC-VRF, a different label MUST be advertised per CE next-hop or 909 MAC so that no MAC FIB lookup is needed at the egress PE. In 910 general, this means that a different label at least per CE must be 911 advertised, although the PE can decide to implement a label per 912 MAC if more granularity (hence less scalability) is required in 913 terms of forwarding states. It is up to the PE1 implementation to 914 use a different label per individual MAC within the same ES 915 Segment (even if only one label per ESI is enough). 917 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 918 learning new local CE MAC addresses on the data plane, but will 919 rather add forwarding states to the MPLS LFIB. 921 6.4. Impact of MPLS-based forwarding on the VLAN-aware service 922 procedures 924 Compared to the MAC-based forwarding model, the MPLS-based forwarding 925 model has definitively an impact in terms of number of A-D routes 926 when all the service interfaces are VLAN-aware bundle type. The 927 differences for the use-case described in this document are 928 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 route for ESI12 per PE): no impact 934 compared to the MAC-based model. 936 o Ethernet A-D routes per EVI (4k routes per PE/ESI): PE1 and PE2 937 will send 4k routes for EVI300, one per 938 tuple. This will allow the egress PE to find out all the 939 forwarding information in the MPLS LFIB and even support Ethernet- 940 Tag to CE-VID translation at the egress. The MAC-based forwarding 941 model would allow the PEs to send a single route per PE/ESI for 942 EVI300, since the packet with the embedded Ethernet-Tag would be 943 used to perform a MAC lookup and find out the egress CE-VID. 945 o MAC-advertisement routes: instead of allocating and advertising the 946 same MPLS label for all the new MACs locally learnt on the same 947 MAC-VRF, a different label MUST be advertised per CE next-hop or 948 MAC so that no MAC FIB lookup is needed at the egress PE. In 949 general, this means that a different label at least per CE must be 950 advertised, although the PE can decide to implement a label per 951 MAC if more granularity (hence less scalability) is required in 952 terms of forwarding states. It is up to the PE1 implementation to 953 use a different label per individual MAC within the same ES 954 Segment. Note that, in this model, the Ethernet-Tag will be set to 955 a non-zero value for the MAC-advertisement routes. The same MAC 956 address can be announced with different Ethernet-Tag value. This 957 will make the advertising PE install two different forwarding 958 states in the MPLS LFIB. 960 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 961 learning new local CE MAC addresses on the data plane, but will 962 rather add forwarding states to the MPLS LFIB. 964 7. Comparison between MAC-based and MPLS-based forwarding models 966 Both forwarding models are possible in a network deployment and each 967 one has its own trade-offs. 969 The MAC-based forwarding model can save A-D routes per EVI when VLAN- 970 aware bundling services are deployed and therefore reduce the control 971 plane overhead. This model also saves a significant amount of MPLS 972 labels compared to the MPLS-based forwarding model. All the MACs and 973 A-D routes for the same EVI can signal the same MPLS label, saving 974 labels from the local PE space. A MAC FIB lookup at the egress PE is 975 required in order to do so. 977 The MPLS-based forwarding model can save forwarding states at the 978 egress PEs if labels per next hop CE (as opposed to per MAC) are 979 implemented. No egress MAC lookup is required. An A-D route per is required for VLAN-aware services, as opposed to an 981 A-D route per EVI. Also, a different label per next-hop CE per MAC- 982 VRF is consumed, as opposed to a single label per MAC-VRF. 984 The following table summarizes the implementation details of both 985 models for the VLAN-aware bundling service type. 987 +-----------------------------+----------------+----------------+ 988 | 4k CE-VID VLANs | MAC-based | MPLS-based | 989 | | Model | Model | 990 +-----------------------------+----------------+----------------+ 991 | A-D routes/EVI | 1 per ESI/EVI | 4k per ESI/EVI | 992 | MPLS labels consumed | 1 per MAC-VRF | 1 per CE/EVI | 993 | Egress PE Forwarding states | 1 per MAC | 1 per next-hop | 994 | Egress PE Lookups | 2 (MPLS+MAC) | 1 (MPLS) | 995 +-----------------------------+----------------+----------------+ 997 The egress forwarding model is an implementation local to the egress 998 PE and is independent of the model supported on the rest of the PEs, 999 i.e. in our use-case, PE1, PE2 and PE3 could have either egress 1000 forwarding model without any dependencies. 1002 8. Traffic flow optimization 1004 In addition to the procedures described across sections 1 through 7, 1005 EVPN [RFC7432] procedures allow for optimized traffic handling in 1006 order to minimize unnecessary flooding across the entire 1007 infrastructure. Optimization is provided through specific ARP 1008 termination and the ability to block unknown unicast flooding. 1009 Additionally, EVPN procedures allow for intelligent, close to the 1010 source, inter-subnet forwarding and solves the commonly known sub- 1011 optimal routing problem. Besides the traffic efficiency, ingress 1012 based inter-subnet forwarding also optimizes packet forwarding rules 1013 and implementation at the egress nodes as well. Details of these 1014 procedures are outlined in sections 8.1 and 8.2. 1016 8.1. Control Plane Procedures 1018 8.1.1. MAC learning options 1020 The fundamental premise of [RFC7432] is the notion of a different 1021 approach to MAC address learning compared to traditional IEEE 802.1 1022 bridge learning methods; specifically EVPN differentiates between 1023 data and control plane driven learning mechanisms. 1025 Data driven learning implies that there is no separate communication 1026 channel used to advertise and propagate MAC addresses. Rather, MAC 1027 addresses are learned through IEEE defined bridge-learning procedures 1028 as well as by snooping on DHCP and ARP requests. As different MAC 1029 addresses show up on different ports, the L2 FIB is populated with 1030 the appropriate MAC addresses. 1032 Control plane driven learning implies a communication channel that 1033 could be either a control-plane protocol or a management-plane 1034 mechanism. In the context of EVPN, two different learning procedures 1035 are defined, i.e. local and remote procedures: 1037 o Local learning defines the procedures used for learning the MAC 1038 addresses of network elements locally connected to a MAC-VRF. 1039 Local learning could be implemented through all three learning 1040 procedures: control plane, management plane as well as data plane. 1041 However, the expectation is that for most of the use cases, local 1042 learning through data plane should be sufficient. 1044 o Remote learning defines the procedures used for learning MAC 1045 addresses of network elements remotely connected to a MAC-VRF, 1046 i.e. far-end PEs. Remote learning procedures defined in [RFC7432] 1047 advocate using only control plane learning; specifically BGP. 1048 Through the use of BGP EVPN NLRIs, the remote PE has the 1049 capability of advertising all the MAC addresses present in its 1050 local FIB. 1052 8.1.2. Proxy-ARP/ND 1054 In EVPN, MAC addresses are advertised via the MAC/IP Advertisement 1055 Route, as discussed in [RFC7432]. Optionally an IP address can be 1056 advertised along with the MAC address announcement. However, there 1057 are certain rules put in place in terms of IP address usage: if the 1058 MAC Advertisement Route contains an IP address, and the IP Address 1059 Length is 32 bits (or 128 in the IPv6 case), this particular IP 1060 address correlates directly with the advertised MAC address. Such 1061 advertisement allows us to build a proxy-ARP/ND table populated with 1062 the IP<->MAC bindings received from all the remote nodes. 1064 Furthermore, based on these bindings, a local MAC-VRF can now provide 1065 Proxy-ARP/ND functionality for all ARP requests and ND solicitations 1066 directed to the IP address pool learned through BGP. Therefore, the 1067 amount of unnecessary L2 flooding, ARP/ND requests/solicitations in 1068 this case, can be further reduced by the introduction of Proxy-ARP/ND 1069 functionality across all EVI MAC-VRFs. 1071 8.1.3. Unknown Unicast flooding suppression 1073 Given that all locally learned MAC addresses are advertised through 1074 BGP to all remote PEs, suppressing flooding of any Unknown Unicast 1075 traffic towards the remote PEs is a feasible network optimization. 1077 The assumption in the use case is made that any network device that 1078 appears on a remote MAC-VRF will somehow signal its presence to the 1079 network. This signaling can be done through e.g. gratuitous ARPs. 1080 Once the remote PE acknowledges the presence of the node in the MAC- 1081 VRF, it will do two things: install its MAC address in its local FIB 1082 and advertise this MAC address to all other BGP speakers via EVPN 1083 NLRI. Therefore, we can assume that any active MAC address is 1084 propagated and learnt through the entire EVI. Given that MAC 1085 addresses become pre-populated - once nodes are alive on the network 1086 - there is no need to flood any unknown unicast towards the remote 1087 PEs. If the owner of a given destination MAC is active, the BGP route 1088 will be present in the local RIB and FIB, assuming that the BGP 1089 import policies are successfully applied; otherwise, the owner of 1090 such destination MAC is not present on the network. 1092 It is worth noting that unless: a) control or management plane 1093 learning is performed through the entire EVI or b) all the EVI- 1094 attached devices signal their presence when they come up (GARPs or 1095 similar), unknown unicast flooding MUST be enabled. 1097 8.1.4. Optimization of Inter-subnet forwarding 1099 In a scenario in which both L2 and L3 services are needed over the 1100 same physical topology, some interaction between EVPN and IP-VPN is 1101 required. A common way of stitching the two service planes is through 1102 the use of an IRB interface, which allows for traffic to be either 1103 routed or bridged depending on its destination MAC address. If the 1104 destination MAC address is the one of the IRB interface, traffic 1105 needs to be passed through a routing module and potentially be either 1106 routed to a remote PE or forwarded to a local subnet. If the 1107 destination MAC address is not the one of the IRB, the MAC-VRF 1108 follows standard bridging procedures. 1110 A typical example of EVPN inter-subnet forwarding would be a scenario 1111 in which multiple IP subnets are part of a single or multiple EVIs, 1112 and they all belong to a single IP-VPN. In such topologies, it is 1113 desired that inter-subnet traffic can be efficiently routed without 1114 any tromboning effects in the network. Due to the overlapping 1115 physical and service topology in such scenarios, all inter-subnet 1116 connectivity will be locally routed trough the IRB interface. 1118 In addition to optimizing the traffic patterns in the network, local 1119 inter-subnet forwarding also optimizes greatly the amount of 1120 processing needed to cross the subnets. Through EVPN MAC 1121 advertisements, the local PE learns the real destination MAC address 1122 associated with the remote IP address and the inter-subnet forwarding 1123 can happen locally. When the packet is received at the egress PE, it 1124 is directly mapped to an egress MAC-VRF, bypassing any egress IP-VPN 1125 processing. 1127 Please refer to [EVPN-INTERSUBNET] for more information about the IP 1128 inter-subnet forwarding procedures in EVPN. 1130 8.2. Packet Walkthrough Examples 1131 Assuming that the services are setup according to figure 1 in section 1132 2, the following flow optimization processes will take place in terms 1133 of creating, receiving and forwarding packets across the network. 1135 8.2.1. Proxy-ARP example for CE2 to CE3 traffic 1137 Using figure 1 in section 2, consider EVI 400 residing on PE1, PE2 1138 and PE3 connecting CE2 and CE3 networks. Also, consider that PE1 and 1139 PE2 are part of the all-active multi-homing ES for CE2, and that PE2 1140 is elected designated-forwarder for EVI400. We assume that all the 1141 PEs implement the proxy-ARP functionality in the MAC-VRF-400 context. 1143 In this scenario, PE3 will not only advertise the MAC addresses 1144 through the EVPN MAC Advertisement Route but also IP addresses of 1145 individual hosts, i.e. /32 prefixes, behind CE3. Upon receiving the 1146 EVPN routes, PE1 and PE2 will install the MAC addresses in the MAC- 1147 VRF-400 FIB and based on the associated received IP addresses, PE1 1148 and PE2 can now build a proxy-ARP table within the context of MAC- 1149 VRF-400. 1151 From the forwarding perspective, when a node behind CE2 sends a frame 1152 destined to a node behind CE3, it will first send an ARP request to 1153 e.g. PE2 (based on the result of the CE2 hashing). Assuming that PE2 1154 has populated its proxy-ARP table for all active nodes behind the 1155 CE3, and that the IP address in the ARP message matches the entry in 1156 the table, PE2 will respond to the ARP request with the actual MAC 1157 address on behalf of the node behind CE3. 1159 Once the nodes behind CE2 learn the actual MAC address of the nodes 1160 behind CE3, all the MAC-to-MAC communications between the two 1161 networks will be unicast. 1163 8.2.2. Flood suppression example for CE1 to CE3 traffic 1165 Using figure 1 in section 2, consider EVI 500 residing on PE1 and PE3 1166 connecting CE1 and CE3 networks. Consider that both PE1 and PE3 have 1167 disabled unknown unicast flooding for this specific EVI context. Once 1168 the network devices behind CE3 come online they will learn their MAC 1169 addresses and create local FIB entries for these devices. Note that 1170 local FIB entries could also be created through either a control or 1171 management plane between PE and CE as well. Consequently, PE3 will 1172 automatically create EVPN Type 2 MAC Advertisement Routes and 1173 advertise all locally learned MAC addresses. The routes will also 1174 include the corresponding MPLS label. 1176 Given that PE1 automatically learns and installs all MAC addresses 1177 behind CE3, its MAC-VRF FIB will already be pre-populated with the 1178 respective next-hops and label assignments associated with the MAC 1179 addresses behind CE3. As such, as soon as the traffic sent by CE1 to 1180 nodes behind CE3 is received into the context of EVI 500, PE1 will 1181 push the MPLS Label(s) onto the original Ethernet frame and send the 1182 packet to the MPLS network. As usual, once PE3 receives this packet, 1183 and depending on the forwarding model, PE3 will either do a next-hop 1184 lookup in the EVI 500 context, or will just forward the traffic 1185 directly to the CE3. In the case that PE1 MAC-VRF-500 does not have a 1186 MAC entry for a specific destination that CE1 is trying to reach, PE1 1187 will drop the frame since unknown unicast flooding is disabled. 1189 Based on the assumption that all the MAC entries behind the CEs are 1190 pre-populated through gratuitous-ARP and/or DHCP requests, if one 1191 specific MAC entry is not present in the MAC-VRF-500 FIB on PE1, the 1192 owner of that MAC is not alive on the network behind the CE3, hence 1193 the traffic can be dropped at PE1 instead of be flooded and consume 1194 network bandwidth. 1196 8.2.3. Optimization of inter-subnet forwarding example for CE3 to CE2 1197 traffic 1199 Using figure 1 in section 2 consider that there is an IP-VPN 666 1200 context residing on PE1, PE2 and PE3 which connects CE1, CE2 and CE3 1201 into a single IP-VPN domain. Also consider that there are two EVIs 1202 present on the PEs, EVI 600 and EVI 60. Each IP subnet is associated 1203 to a different MAC-VRF context. Thus there is a single subnet, subnet 1204 600, between CE1 and CE3 that is established through EVI 600. 1205 Similarly, there is another subnet, subnet 60, between CE2 and CE3 1206 that is established through EVI 60. Since both subnets are part of 1207 the same IP VPN, there is a mapping of each EVI (or individual 1208 subnet) to a local IRB interface on the three PEs. 1210 If a node behind CE2 wants to communicate with a node on the same 1211 subnet seating behind CE3, the communication flow will follow the 1212 standard EVPN procedures, i.e. FIB lookup within the PE1 (or PE2) 1213 after adding the corresponding EVPN label to the MPLS label stack 1214 (downstream label allocation from PE3 for EVI 60). 1216 When it comes to crossing the subnet boundaries, the ingress PE 1217 implements local inter-subnet forwarding. For example, when a node 1218 behind CE2 (EVI 60) sends a packet to a node behind CE1 (EVI 600) the 1219 destination IP address will be in the subnet 600, but the destination 1220 MAC address will be the address of source node's default gateway, 1221 which in this case will be an IRB interface on PE1 (connecting EVI 60 1222 to IP-VPN 666). Once PE1 sees the traffic destined to its own MAC 1223 address, it will route the packet to EVI 600, i.e. it will change the 1224 source MAC address to the one of the IRB interface in EVI 600 and 1225 change the destination MAC address to the address belonging to the 1226 node behind CE1, which is already populated in the MAC-VRF-600 FIB, 1227 either through data or control plane learning. 1229 An important optimization to be noted is the local inter-subnet 1230 forwarding in lieu of IP VPN routing. If the node from subnet 60 1231 (behind CE2) is sending a packet to the remote end node on subnet 600 1232 (behind CE3), the mechanism in place still honors the local inter- 1233 subnet (inter-EVI) forwarding. 1235 In our use-case, therefore, when node from subnet 60 behind CE2 sends 1236 traffic to the node on subnet 600 behind CE3, the destination MAC 1237 address is the PE1 MAC-VRF-60 IRB MAC address. However, once the 1238 traffic locally crosses EVIs, to EVI 600, via the IRB interface on 1239 PE1, the source MAC address is changed to that of the IRB interface 1240 and the destination MAC address is changed to the one advertised by 1241 PE3 via EVPN and already installed in MAC-VRF-600. The rest of the 1242 forwarding through PE1 is using the MAC-VRF-600 forwarding context 1243 and label space. 1245 Another very relevant optimization is due to the fact that traffic 1246 between PEs is forwarded through EVPN, rather than through IP-VPN. In 1247 the example described above for traffic from EVI 60 on CE2 to EVI 600 1248 on CE3, there is no need for IP-VPN processing on the egress PE3. 1249 Traffic is forwarded either to the EVI 600 context in PE3 for further 1250 MAC lookup and next-hop processing, or directly to the node behind 1251 CE3, depending on the egress forwarding model being used. 1253 9. Conventions used in this document 1255 In the examples, the following conventions are used: 1257 o CE-VIDs refer to the VLAN tag identifiers being used at CE1, CE2 1258 and CE3 to tag customer traffic sent to the Service Provider E- 1259 VPN network 1261 o CE1-MAC, CE2-MAC and CE3-MAC refer to source MAC addresses "behind" 1262 each CE respectively. Those MAC addresses can belong to the CEs 1263 themselves or to devices connected to the CEs. 1265 o CE1-IP, CE2-IP and CE3-IP refer to IP addresses associated to the 1266 above MAC addresses. 1268 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 1269 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 1270 document are to be interpreted as described in RFC-2119 [RFC2119]. 1272 In this document, these words will appear with that interpretation 1273 only when in ALL CAPS. Lower case uses of these words are not to be 1274 interpreted as carrying RFC-2119 significance. 1276 10. Security Considerations 1278 Please refer to the "Security Considerations" section in [RFC7432]. 1280 11. IANA Considerations 1282 No new IANA considerations are needed. 1284 12. References 1286 12.1. Normative References 1288 [RFC4761]Kompella, K., Ed., and Y. Rekhter, Ed., "Virtual Private LAN 1289 Service (VPLS) Using BGP for Auto-Discovery and Signaling", RFC 4761, 1290 DOI 10.17487/RFC4761, January 2007, . 1293 [RFC4762]Lasserre, M., Ed., and V. Kompella, Ed., "Virtual Private 1294 LAN Service (VPLS) Using Label Distribution Protocol (LDP) 1295 Signaling", RFC 4762, DOI 10.17487/RFC4762, January 2007, 1296 . 1298 [RFC6074]Rosen, E., Davie, B., Radoaca, V., and W. Luo, 1299 "Provisioning, Auto-Discovery, and Signaling in Layer 2 Virtual 1300 Private Networks (L2VPNs)", RFC 6074, DOI 10.17487/RFC6074, January 1301 2011, . 1303 [RFC4364]Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1304 Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February 2006, 1305 . 1307 [RFC7209]Sajassi, A., Aggarwal, R., Uttaro, J., Bitar, N., 1308 Henderickx, W., and A. Isaac, "Requirements for Ethernet VPN (EVPN)", 1309 RFC 7209, DOI 10.17487/RFC7209, May 2014, . 1312 [RFC7117]Aggarwal, R., Ed., Kamite, Y., Fang, L., Rekhter, Y., and C. 1313 Kodeboniya, "Multicast in Virtual Private LAN Service (VPLS)", 1314 RFC 7117, DOI 10.17487/RFC7117, February 2014, . 1317 [RFC7432]Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 1318 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based Ethernet 1319 VPN", RFC 7432, DOI 10.17487/RFC7432, February 2015, . 1322 12.2. Informative References 1324 [EVPN-INTERSUBNET] Sajassi et al., "IP Inter-subnet forwarding in 1325 EVPN", draft-ietf-bess-evpn-inter-subnet-forwarding-03.txt 1327 13. Acknowledgments 1329 The authors want to thank Giles Heron for his detailed review of the 1330 document. We also thank Stefan Plug, and Eric Wunan for their 1331 comments. 1333 14. Contributors 1334 In addition to the authors listed on the front page, the following 1335 co-authors have also contributed to this document: 1337 Florin Balus 1339 14. Authors' Addresses 1341 Jorge Rabadan 1342 Nokia 1343 777 E. Middlefield Road 1344 Mountain View, CA 94043 USA 1345 Email: jorge.rabadan@nokia.com 1347 Senad Palislamovic 1348 Nokia 1349 Email: senad.palislamovic@nokia.com 1351 Wim Henderickx 1352 Nokia 1353 Email: wim.henderickx@nokia.com 1355 Keyur Patel 1356 Arrcus 1357 Email: keyur@arrcus.com 1359 Ali Sajassi 1360 Cisco 1361 Email: sajassi@cisco.com 1363 James Uttaro 1364 AT&T 1365 Email: uttaro@att.com 1367 Aldrin Isaac 1368 Juniper Networks 1369 Email: aisaac@juniper.net 1371 Truman Boyes 1372 Bloomberg 1373 Email: tboyes@bloomberg.net