idnits 2.17.1 draft-ietf-bess-evpn-usage-02.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 19, 2016) is 2960 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC7432' is mentioned on line 1276, but not defined == Missing Reference: 'RFC4761' is mentioned on line 240, but not defined == Missing Reference: 'RFC4762' is mentioned on line 240, but not defined == Missing Reference: 'RFC6074' is mentioned on line 240, but not defined == Missing Reference: 'RFC7209' is mentioned on line 253, but not defined == Missing Reference: 'PE-IP' is mentioned on line 355, but not defined == Missing Reference: 'AS' is mentioned on line 361, but not defined == Missing Reference: 'RFC2119' is mentioned on line 1269, but not defined == Outdated reference: A later version (-15) exists of draft-ietf-bess-evpn-inter-subnet-forwarding-01 Summary: 1 error (**), 0 flaws (~~), 10 warnings (==), 1 comment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 K. Patel 8 AT&T A. Sajassi 9 Cisco 11 T. Boyes A. Isaac 12 Bloomberg Juniper 14 Expires: September 20, 2016 March 19, 2016 16 Usage and applicability of BGP MPLS based Ethernet VPN 17 draft-ietf-bess-evpn-usage-02 19 Abstract 21 This document discusses the usage and applicability of BGP MPLS based 22 Ethernet VPN (EVPN) in a simple and fairly common deployment 23 scenario. The different EVPN procedures will be explained on the 24 example scenario, analyzing the benefits and trade-offs of each 25 option. Along with [RFC7432], this document is intended to provide a 26 simplified guide for the deployment of EVPN in Service Provider 27 networks. 29 Status of this Memo 31 This Internet-Draft is submitted in full conformance with the 32 provisions of BCP 78 and BCP 79. 34 Internet-Drafts are working documents of the Internet Engineering 35 Task Force (IETF), its areas, and its working groups. Note that 36 other groups may also distribute working documents as Internet- 37 Drafts. 39 Internet-Drafts are draft documents valid for a maximum of six months 40 and may be updated, replaced, or obsoleted by other documents at any 41 time. It is inappropriate to use Internet-Drafts as reference 42 material or to cite them other than as "work in progress." 43 The list of current Internet-Drafts can be accessed at 44 http://www.ietf.org/ietf/1id-abstracts.txt 46 The list of Internet-Draft Shadow Directories can be accessed at 47 http://www.ietf.org/shadow.html 49 This Internet-Draft will expire on September 20, 2016. 51 Copyright Notice 53 Copyright (c) 2016 IETF Trust and the persons identified as the 54 document authors. All rights reserved. 56 This document is subject to BCP 78 and the IETF Trust's Legal 57 Provisions Relating to IETF Documents 58 (http://trustee.ietf.org/license-info) in effect on the date of 59 publication of this document. Please review these documents 60 carefully, as they describe your rights and restrictions with respect 61 to this document. Code Components extracted from this document must 62 include Simplified BSD License text as described in Section 4.e of 63 the Trust Legal Provisions and are provided without warranty as 64 described in the Simplified BSD License. 66 Table of Contents 68 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 69 2. Use-case scenario description . . . . . . . . . . . . . . . . . 4 70 3. Provisioning Model . . . . . . . . . . . . . . . . . . . . . . 6 71 3.1. Common provisioning tasks . . . . . . . . . . . . . . . . . 6 72 3.1.1. Non-service specific parameters . . . . . . . . . . . . 7 73 3.1.2. Service specific parameters . . . . . . . . . . . . . . 7 74 3.2. Service interface dependent provisioning tasks . . . . . . 8 75 3.2.1. VLAN-based service interface EVI . . . . . . . . . . . 8 76 3.2.2. VLAN-bundle service interface EVI . . . . . . . . . . . 9 77 3.2.3. VLAN-aware bundling service interface EVI . . . . . . . 9 78 4. BGP EVPN NLRI usage . . . . . . . . . . . . . . . . . . . . . . 9 79 5. MAC-based forwarding model use-case . . . . . . . . . . . . . . 10 80 5.1. EVPN Network Startup procedures . . . . . . . . . . . . . . 10 81 5.2. VLAN-based service procedures . . . . . . . . . . . . . . . 11 82 5.2.1. Service startup procedures . . . . . . . . . . . . . . 11 83 5.2.2. Packet walkthrough . . . . . . . . . . . . . . . . . . 12 84 5.3. VLAN-bundle service procedures . . . . . . . . . . . . . . 15 85 5.3.1. Service startup procedures . . . . . . . . . . . . . . 15 86 5.3.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 16 87 5.4. VLAN-aware bundling service procedures . . . . . . . . . . 16 88 5.4.1. Service startup procedures . . . . . . . . . . . . . . 16 89 5.4.2. Packet Walkthrough . . . . . . . . . . . . . . . . . . 17 90 6. MPLS-based forwarding model use-case . . . . . . . . . . . . . 18 91 6.1. Impact of MPLS-based forwarding on the EVPN network 92 startup . . . . . . . . . . . . . . . . . . . . . . . . . . 19 93 6.2. Impact of MPLS-based forwarding on the VLAN-based service 94 procedures . . . . . . . . . . . . . . . . . . . . . . . . 19 95 6.3. Impact of MPLS-based forwarding on the VLAN-bundle 96 service procedures . . . . . . . . . . . . . . . . . . . . 19 97 6.4. Impact of MPLS-based forwarding on the VLAN-aware service 98 procedures . . . . . . . . . . . . . . . . . . . . . . . . 20 99 7. Comparison between MAC-based and MPLS-based forwarding models . 21 100 8. Traffic flow optimization . . . . . . . . . . . . . . . . . . . 22 101 8.1. Control Plane Procedures . . . . . . . . . . . . . . . . . 22 102 8.1.1. MAC learning options . . . . . . . . . . . . . . . . . 22 103 8.1.2. Proxy-ARP/ND . . . . . . . . . . . . . . . . . . . . . 23 104 8.1.3. Unknown Unicast flooding suppression . . . . . . . . . 23 105 8.1.4. Optimization of Inter-subnet forwarding . . . . . . . . 24 106 8.2. Packet Walkthrough Examples . . . . . . . . . . . . . . . . 24 107 8.2.1. Proxy-ARP example for CE2 to CE3 traffic . . . . . . . 25 108 8.2.2. Flood suppression example for CE1 to CE3 traffic . . . 25 109 8.2.3. Optimization of inter-subnet forwarding example for 110 CE3 to CE2 traffic . . . . . . . . . . . . . . . . . . 26 111 9. Conventions used in this document . . . . . . . . . . . . . . . 27 112 10. Security Considerations . . . . . . . . . . . . . . . . . . . 28 113 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 114 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 28 115 12.1. Normative References . . . . . . . . . . . . . . . . . . . 28 116 12.2. Informative References . . . . . . . . . . . . . . . . . . 29 117 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 29 118 14. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 29 119 14. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 29 121 1. Introduction 123 This document complements [RFC7432] by discussing the applicability 124 of the technology in a simple and fairly common deployment scenario, 125 which is described in section 2. 127 After describing the topology of the use-case scenario and the 128 characteristics of the service to be deployed, section 3 will 129 describe the provisioning model, comparing the EVPN procedures with 130 the provisioning tasks required for other VPN technologies, such as 131 VPLS or IP-VPN. 133 Once the provisioning model is analyzed, sections 4, 5 and 6 will 134 describe the control plane and data plane procedures in the example 135 scenario, for the two potential disposition/forwarding models: 136 MAC-based and MPLS-based models. While both models can interoperate 137 in the same network, each one has different trade-offs that are 138 analyzed in section 7. 140 Finally, EVPN provides some potential traffic flow optimization tools 141 that are also described in section 8, in the context of the example 142 scenario. 144 2. Use-case scenario description 146 The following figure depicts the scenario that will be referenced 147 throughout the rest of the document. 149 +--------------+ 150 | | 151 +----+ +----+ | | +----+ +----+ 152 | CE1|-----| | | | | |---| CE3| 153 +----+ /| PE1| | IP/MPLS | | PE3| +----+ 154 / +----+ | Network | +----+ 155 / | | 156 / +----+ | | 157 +----+/ | | | | 158 | CE2|-----| PE2| | | 159 +----+ +----+ | | 160 +--------------+ 162 Figure 1 EVPN use-case scenario 164 There are three PEs and three CEs considered in this example: PE1, 165 PE2, PE3, as well as CE1, CE2 and CE3. Layer-2 traffic must be 166 extended among the three CEs. The following service requirements are 167 assumed in this scenario: 169 o Redundancy requirements: CE1 and CE3 are single-homed to PE1 and 170 PE3 respectively. CE2 requires multi-homing connectivity to PE1 and 171 PE2, not only for redundancy purposes, but also for adding more 172 upstream/downstream connectivity bandwidth to/from the network. If 173 CE2 has a single CE-VID (or a few CE-VIDs) the current VPLS 174 multi-homing solutions (based on load-balancing per CE-VID or 175 service) do not provide the optimized link utilization required in 176 this example. Another redundancy requirement that must be met is 177 fast convergence. E.g.: if the link between CE2 and PE1 goes down, 178 a fast convergence mechanism must be supported so that PE3 can 179 immediately send the traffic to PE2, irrespectively of the number 180 of affected services and MAC addresses. EVPN provides the 181 flow-based load-balancing multi-homing solution required in this 182 scenario to optimize the upstream/downstream link utilization 183 between CE2 and PE1-PE2. EVPN also provides a fast convergence 184 solution so that PE3 can immediately send the traffic to PE2 upon 185 failure on the link between CE2 and PE1. 187 o Service interface requirements: service definition must be flexible 188 in terms of CE-VID-to-broadcast-domain assignment and service 189 contexts in the core. The following three services are required in 190 this example: 192 EVI100 - It will use VLAN-based service interfaces in the three CEs 193 with a 1:1 mapping (VLAN-to-EVI). The CE-VIDs at the three CEs can 194 be the same, e.g.: VID 100, or different at each CE, e.g.: VID 101 195 in CE1, VID 102 in CE2 and VID 103 in CE3. A single broadcast 196 domain needs to be created for EVI100 in any case; therefore CE- 197 VIDs will require translation at the egress PEs if they are not 198 consistent across the three CEs. The case when the same CE-VID is 199 used across the three CEs for EVI100 is referred in [RFC7432] as 200 the "Unique VLAN" EVPN case. This term will be used throughout this 201 document too. 203 EVI200 - It will use VLAN-bundle service interfaces in CE1, CE2 and 204 CE3, based on an N:1 VLAN-to-EVI mapping. In this case, the service 205 provider just needs to assign a pre-configured number of CE-VIDs on 206 the ingress PE to EVI200, and send the customer frames with the 207 original CE-VIDs. The Service Provider will build a single 208 broadcast domain for the customer. The customer will be responsible 209 for the CE-VID handling. 211 EVI300 - It will use VLAN-aware bundling service interfaces in CE1, 212 CE2 and CE3. At the ingress PE, an N:1 VLAN-to-EVI mapping will be 213 done, however and as opposed to EVI200, a separate core broadcast 214 domain is required per CE-VID. In addition to that, the CE-VIDs can 215 be different (hence CE-VID translation is required). Note that, 216 while the requirements stated for EVI100 and EVI200 might be met 217 with the current VPLS solutions, the VLAN-aware bundling service 218 interfaces required by EVI300 are not supported by the current VPLS 219 tools. 221 NOTE: in section 3.2.1, only EVI100 is used as an example of 222 VLAN-based service provisioning. In sections 5.2 and 6.2, 4k 223 VLAN-based EVIs (EVI1 to EVI4k) are used so that the impact of MAC 224 vs. MPLS disposition models in the control plane can be evaluated. In 225 the same way, EVI200 and EVI300 will be described with a 4k:1 mapping 226 (CE-VIDs-to-EVI mapping) in sections 5.3-4 and 6.3-4. 228 o BUM (Broadcast, Unknown unicast, Multicast) optimization 229 requirements: The solution must be able to support ingress 230 replication and P2MP MPLS LSPs and the user must be able to decide 231 what kind of provider tree will be used by each EVI service. For 232 example, if we assume that EVI100 and EVI200 will not carry much 233 BUM traffic, we can use ingress replication for those service 234 instances. The benefit is that the core will not need to maintain 235 any states for the multicast trees associated to EVI100 and EVI200. 236 On the contrary, if EVI300 is presumably carrying a significant 237 amount of multicast traffic, P2MP MPLS LSPs can be used for this 238 service. 240 The current VPLS solutions, based on [RFC4761][RFC4762][RFC6074], 241 cannot meet all the above set of requirements and therefore a new 242 solution is needed. The rest of the document will describe how EVPN 243 can be used to meet those service requirements and even optimize the 244 network further by: 246 o Providing the user with an option to reduce (and even suppress) the 247 ARP-flooding. 249 o Supporting ARP termination for inter-subnet forwarding 251 3. Provisioning Model 253 One of the requirements stated in [RFC7209] is the ease of 254 provisioning. BGP parameters and service context parameters should be 255 auto-provisioned so that the addition of a new MAC-VRF to the EVI 256 requires a minimum number of single-sided provisioning touches. 257 However this is only possible in a limited number of cases. This 258 section describes the provisioning tasks required for the services 259 described in section 2, i.e. EVI100 (VLAN-based service interfaces), 260 EVI200 (VLAN-bundle service interfaces) and EVI300 (VLAN-aware 261 bundling service interfaces). 263 3.1. Common provisioning tasks 264 Regardless of the service interface type (VLAN-based, VLAN-bundle or 265 VLAN-aware), the following sub-sections describe the parameters to be 266 provisioned in the three PEs. 268 3.1.1. Non-service specific parameters 270 The multi-homing function in EVPN requires the provisioning of 271 certain parameters which are not service-specific and that are shared 272 by all the MAC-VRFs in the node using the multi-homing capabilities. 273 In our use-case, these parameters are only provisioned in PE1 and 274 PE2, and are listed below: 276 o Ethernet Segment Identifier (ESI): only the ESI associated to CE2 277 needs to be considered in our example. Single-homed CEs such as CE1 278 and CE3 do not require the provisioning of an ESI (the ESI will be 279 coded as zero in the BGP NLRIs). In our example, a LAG is used 280 between CE2 and PE1-PE2 (since all-active multi-homing is a 281 requirement) therefore the ESI can be auto-derived from the LACP 282 information as described in [RFC7432]. Note that the ESI MUST be 283 unique across all the PEs in the network, therefore the 284 auto-provisioning of the ESI is only recommended in case the CEs 285 are managed by the Service Provider. Otherwise the ESI should be 286 manually provisioned (type 0 as in [RFC7432]) in order to avoid 287 potential conflicts. 289 o ES-Import Route Target (ES-Import RT): this is the RT that will be 290 sent by PE1 and PE2, along with the ES route. Regardless of how the 291 ESI is provisioned in PE1 and PE2, the ES-Import RT must always be 292 auto-derived from the 6-byte MAC address portion of the ESI value. 294 o Ethernet Segment Route Distinguisher (ES RD): this is the RD to be 295 encoded in the ES route and Ethernet Auto-Discovery (A-D) route to 296 be sent by PE1 and PE2 for the CE2 ESI. This RD should always be 297 auto-derived from the PE IP address, as described in [RFC7432]. 299 o Multi-homing type: the user must be able to provision the 300 multi-homing type to be used in the network. In our use-case, the 301 multi-homing type will be set to all-active for the CE2 ESI. This 302 piece of information is encoded in the ESI Label extended community 303 flags and sent by PE1 and PE2 along with the Ethernet A-D route for 304 the CE2 ESI. 306 In our use-case, besides the above parameters, the same LACP 307 parameters will be configured in PE1 and PE2 for the ESI, so that CE2 308 can send different flows to PE1 and PE2 for the same CE-VID as though 309 they were forming a single system from the CE2 perspective. 311 3.1.2. Service specific parameters 312 The following parameters must be provisioned in PE1, PE2 and PE3 per 313 EVI service: 315 o EVI identifier: global identifier per EVI that is shared by all the 316 PEs part of the EVI, i.e. PE1, PE2 and PE3 will be provisioned with 317 EVI100, 200 and 300. The EVI identifier can be associated to (or be 318 the same value as) the EVI default Ethernet Tag (4-byte default 319 broadcast domain identifier for the EVI). The Ethernet Tag is 320 different from zero in the EVPN BGP routes only if the service 321 interface type (of the source PE) is VLAN-aware. 323 o EVI Route Distinguisher (EVI RD): This RD is a unique value across 324 all the MAC-VRFs in a PE. Auto-derivation of this RD might be 325 possible depending on the service interface type being used in the 326 EVI. Next section discusses the specifics of each service interface 327 type. 329 o EVI Route Target(s) (EVI RT): one or more RTs can be provisioned 330 per MAC-VRF. The RT(s) imported and exported can be equal or 331 different, just as the RT(s) in IP-VPNs. Auto-derivation of this 332 RT(s) might be possible depending on the service interface type 333 being used in the EVI. Next section discusses the specifics of each 334 service interface type. 336 o CE-VID and port/LAG binding to EVI identifier or Ethernet Tag: see 337 section 3.2. 339 3.2. Service interface dependent provisioning tasks 341 Depending on the service interface type being used in the EVI, a 342 specific CE-VID binding provisioning must be specified. 344 3.2.1. VLAN-based service interface EVI 346 In our use-case, EVI100 is a VLAN-based service interface EVI. 348 EVI100 can be a "unique-VLAN" EVPN if the CE-VID being used for this 349 service in CE1, CE2 and CE3 is equal, e.g. VID 100. In that case, the 350 VID 100 binding must be provisioned in PE1, PE2 and PE3 for EVI100 351 and the associated port or LAG. The MAC-VRF RD and RT can be auto- 352 derived from the CE-VID: 354 o The auto-derived MAC-VRF RD will be a Type 1 RD, as recommended in 355 [RFC7432], and it will be comprised of [PE-IP]:[zero-padded-VID]; 356 where PE-IP is the IP address of the PE (a loopback address) and 357 [zero-padded-VID] is a 2-byte value where the low order 12 bits are 358 the VID (VID 100 in our example) and the high order 4 bits are 359 zero. 361 o The auto-derived MAC-VRF RT will be composed of [AS]:[zero-padded- 362 VID]; where AS is the Autonomous System that the PE belongs to and 363 [zero-padded-VID] is a 2 or 4-byte value where the low order 12 364 bits are the VID (VID 100 in our example) and the high order bits 365 are zero. Note that auto-deriving the RT implies supporting a basic 366 any-to-any topology in the EVI and using the same import and export 367 RT in the EVI. 369 If EVI100 is not a "unique-VLAN" EVPN, each individual CE-VID must be 370 configured in each PE, and MAC-VRF RDs and RTs cannot be auto- 371 derived, hence they must be provisioned by the user. 373 3.2.2. VLAN-bundle service interface EVI 375 Assuming EVI200 is a VLAN-bundle service interface EVI, and VIDs 376 200-250 are assigned to EVI200, the CE-VID bundle 200-250 must be 377 provisioned on PE1, PE2 and PE3. Note that this model does not allow 378 CE-VID translation and the CEs must use the same CE-VIDs for EVI200. 379 No auto-derived EVI RDs or EVI RTs are possible. 381 3.2.3. VLAN-aware bundling service interface EVI 383 If EVI300 is a VLAN-aware bundling service interface EVI, CE-VID 384 binding to EVI300 does not have to match on the three PEs (only on 385 PE1 and PE2, since they are part of the same ES). E.g.: PE1 and PE2 386 CE-VID binding to EVI300 can be set to the range 300-310 and PE3 to 387 321-330. Note that each individual CE-VID will be assigned to a core 388 broadcast domain, i.e. Ethernet Tag, which will be encoded in the BGP 389 EVPN routes. 391 Therefore, besides the CE-VID bundle range bound to EVI300 in each 392 PE, associations between each individual CE-VID and the EVPN Ethernet 393 Tag must be provisioned by the user. No auto-derived EVI RDs/RTs are 394 possible. 396 4. BGP EVPN NLRI usage 398 [RFC7432] defines four different types of routes and four different 399 extended communities advertised along with the different routes. 400 However not all the PEs in a network must generate and process all 401 the different routes and extended communities. The following table 402 shows the routes that must be exported and imported in the use-case 403 described in this document. "Export", in this context, means that the 404 PE must be capable of generating and exporting a given route, 405 assuming there are no BGP policies to prevent it. In the same way, 406 "Import" means the PE must be capable of importing and processing a 407 given route, assuming the right RTs and policies. "N/A" means neither 408 import nor export actions are required. 410 +-------------------+---------------+---------------+ 411 | BGP EVPN routes | PE1-PE2 | PE3 | 412 +-------------------+---------------+---------------+ 413 | ES | Export/import | N/A | 414 | A-D per ESI | Export/import | Import | 415 | A-D per EVI | Export/import | Import | 416 | MAC | Export/import | Export/import | 417 | Inclusive mcast | Export/import | Export/import | 418 +-------------------+---------------+---------------+ 420 PE3 is only required to export MAC and Inclusive multicast routes and 421 be able to import and process A-D routes, as well as MAC and 422 Inclusive multicast routes. If PE3 did not support importing and 423 processing A-D routes per ESI and per EVI, fast convergence and 424 aliasing functions (respectively) would not be possible in this 425 use-case. 427 5. MAC-based forwarding model use-case 429 This section describes how the BGP EVPN routes are exported and 430 imported by the PEs in our use-case, as well as how traffic is 431 forwarded assuming that PE1, PE2 and PE3 support a MAC-based 432 forwarding model. In order to compare the control and data plane 433 impact in the two forwarding models (MAC-based and MPLS-based) and 434 different service types, we will assume that CE1, CE2 and CE3 need to 435 exchange traffic for up to 4k CE-VIDs. 437 5.1. EVPN Network Startup procedures 439 Before any EVI is provisioned in the network, the following 440 procedures are required: 442 o Infrastructure setup: the proper MPLS infrastructure must be setup 443 among PE1, PE2 and PE3 so that the EVPN services can make use of 444 P2P and P2MP LSPs. In addition to the MPLS transport, PE1 and PE2 445 must be properly configured with the same LACP configuration to 446 CE2. Details are provided in [RFC7432]. Once the LAG is properly 447 setup, the ESI for the CE2 Ethernet Segment, e.g. ESI12, can be 448 auto-generated by PE1 and PE2 from the LACP information exchanged 449 with CE2 (ESI type 1), as discussed in section 3.1. Alternatively, 450 the ESI can also be manually provisioned on PE1 and PE2 (ESI type 451 0). PE1 and PE2 will auto-configure a BGP policy that will import 452 any ES route matching the auto-derived ES-import RT for ESI12. 454 o Ethernet Segment route exchange and DF election: PE1 and PE2 will 455 advertise a BGP Ethernet Segment route for ESI12, where the ESI RD 456 and ES-Import RT will be auto-generated as discussed in section 457 3.1.1. PE1 and PE2 will import the ES routes of each other and will 458 run the DF election algorithm for any existing EVI (if any, at this 459 point). PE3 will simply discard the route. Note that the DF 460 election algorithm can support service carving, so that the 461 downstream BUM traffic from the network to CE2 can be load-balanced 462 across PE1 and PE2 on a per-service basis. 464 At the end of this process, the network infrastructure is ready to 465 start deploying EVPN services. PE1 and PE2 are aware of the existence 466 of a shared Ethernet Segment, i.e. ESI12. 468 5.2. VLAN-based service procedures 470 Assuming that the EVPN network must carry traffic among CE1, CE2 and 471 CE3 for up to 4k CE-VIDs, the Service Provider can decide to 472 implement VLAN-based service interface EVIs to accomplish it. In this 473 case, each CE-VID will be individually mapped to a different EVI. 474 While this means a total number of 4k MAC-VRFs is required per PE, 475 the advantages of this approach are the auto-provisioning of most of 476 the service parameters if no VLAN translation is needed (see section 477 3.2.1) and great control over each individual customer broadcast 478 domain. We assume in this section that the range of EVIs from 1 to 4k 479 is provisioned in the network. 481 5.2.1. Service startup procedures 483 As soon as the EVIs are created in PE1, PE2 and PE3, the following 484 control plane actions are carried out: 486 o Flooding tree setup per EVI (4k routes): Each PE will send one 487 Inclusive Multicast Ethernet Tag route per EVI (up to 4k routes per 488 PE) so that the flooding tree per EVI can be setup. Note that 489 ingress replication or P2MP LSPs can optionally be signaled in the 490 PMSI Tunnel attribute and the corresponding tree be created. 492 o Ethernet A-D routes per ESI (a set of routes for ESI12): A set of 493 A-D routes with a list of 4k RTs (one per EVI) for ESI12 will be 494 issued from PE1 and PE2 (it has to be a set of routes so that the 495 total number of RTs can be conveyed). As per [RFC7432], each 496 Ethernet A-D route per ESI is differentiated from the other routes 497 in the set by a different Route Distinguisher (ES RD). This set 498 will also include ESI Label extended communities with the active- 499 standby flag set to zero (all-active multi-homing type) and an ESI 500 Label different from zero (used for split-horizon functions). These 501 routes will be imported by the three PEs, since the RTs match the 502 EVI RTs locally configured. The A-D routes per ESI will be used for 503 fast convergence and split-horizon functions, as discussed in 504 [RFC7432]. 506 o Ethernet A-D routes per EVI (4k routes): An A-D route per EVI will 507 be sent by PE1 and PE2 for ESI12. Each individual route includes 508 the corresponding EVI RT and an MPLS label to be used by PE3 for 509 the aliasing function. These routes will be imported by the three 510 PEs. 512 5.2.2. Packet walkthrough 514 Once the services are setup, the traffic can start flowing. Assuming 515 there are no MAC addresses learnt yet and that MAC learning at the 516 access is performed in the data plane in our use-case, this is the 517 process followed upon receiving frames from each CE (example for 518 EVI1). 520 (1) BUM frame example from CE1: 522 a) An ARP-request with CE-VID=1 is issued from source MAC CE1-MAC 523 (MAC address coming from CE1 or from a device connected to CE1) to 524 find the MAC address of CE3-IP. 526 b) Based on the CE-VID, the frame is identified to be forwarded in 527 the MAC-VRF-1 (EVI1) context. A source MAC lookup is done in the 528 MAC FIB and the sender's CE1-IP in the proxy-ARP table within the 529 MAC-VRF-1 (EVI1) context. If CE1-MAC/CE1-IP are unknown in both 530 tables, three actions are carried out (assuming the source MAC is 531 accepted by PE1): (1) a forwarding state is added for CE1-MAC 532 associated to the corresponding port and CE-VID, (2) the ARP- 533 request is snooped and the tuple CE1-MAC/CE1-IP is added to the 534 proxy-ARP table and (3) a BGP MAC advertisement route is triggered 535 from PE1 containing the EVI1 RD and RT, ESI=0, Ethernet-Tag=0 and 536 CE1-MAC/CE1-IP along with an MPLS label assigned to MAC-VRF-1 from 537 the PE1 label space. Note that depending on the implementation, 538 the MAC FIB and proxy-ARP learning processes can independently 539 send two BGP MAC advertisements instead of one (one containing 540 only the CE1-MAC and another one containing CE1-MAC/CE1-IP). 542 Since we assume a MAC forwarding model, a label per MAC-VRF is 543 normally allocated and signaled by the three PEs for MAC 544 advertisement routes. Based on the RT, the route is imported by 545 PE2 and PE3 and the forwarding state plus ARP entry are added to 546 their MAC-VRF-1 context. From this moment on, any ARP request from 547 CE2 or CE3 destined to CE1-IP, can be directly replied by PE1, PE2 548 or PE3 and ARP flooding for CE1-IP is not needed in the core. 550 c) Since the ARP frame is a broadcast frame, it is forwarded by PE1 551 using the Inclusive multicast tree for EVI1 (CE-VID=1 should be 552 kept if translation is required). Depending on the type of tree, 553 the label stack may vary. E.g. assuming ingress replication, the 554 packet is replicated to PE2 and PE3 with the downstream allocated 555 labels and the P2P LSP transport labels. No other labels are added 556 to the stack. 558 d) Assuming PE1 is the DF for EVI1 on ESI12, the frame is locally 559 replicated to CE2. 561 e) The MPLS-encapsulated frame gets to PE2 and PE3. Since PE2 is non- 562 DF for EVI1 on ESI12, and there is no other CE connected to PE2, 563 the frame is discarded. At PE3, the frame is de-encapsulated, CE- 564 VID translated if needed and replicated to CE3. 566 Any other type of BUM frame from CE1 would follow the same 567 procedures. BUM frames from CE3 would follow the same procedures too. 569 (2) BUM frame example from CE2: 571 a) An ARP-request with CE-VID=1 is issued from source MAC CE2-MAC to 572 find the MAC address of CE3-IP. 574 b) CE2 will hash the frame and will forward it to e.g. PE2. Based on 575 the CE-VID, the frame is identified to be forwarded in the EVI1 576 context. A source MAC lookup is done in the MAC FIB and the 577 sender's CE2-IP in the proxy-ARP table within the MAC-VRF-1 578 context. If both are unknown, three actions are carried out 579 (assuming the source MAC is accepted by PE2): (1) a forwarding 580 state is added for CE2-MAC associated to the corresponding LAG/ESI 581 and CE-VID, (2) the ARP-request is snooped and the tuple CE2- 582 MAC/CE2-IP is added to the proxy-ARP table and (3) a BGP MAC 583 advertisement route is triggered from PE2 containing the EVI1 RD 584 and RT, ESI=12, Ethernet-Tag=0 and CE2-MAC/CE2-IP along with an 585 MPLS label assigned from the PE2 label space (one label per MAC- 586 VRF). Again, depending on the implementation, the MAC FIB and 587 proxy-ARP learning processes can independently send two BGP MAC 588 advertisements instead of one. 590 Note that, since PE3 is not part of ESI12, it will install a 591 forwarding state for CE2-MAC as long as the A-D routes for ESI12 592 are also active on PE3. On the contrary, PE1 is part of ESI12, 593 therefore PE1 will not modify the forwarding state for CE2-MAC if 594 it has previously learnt CE2-MAC locally attached to ESI12. 595 Otherwise it will add forwarding state for CE2-MAC associated to 596 the local ESI12 port. 598 c) Assuming PE2 does not have the ARP information for CE3-IP yet, and 599 since the ARP is a broadcast frame and PE2 the non-DF for EVI1 on 600 ESI12, the frame is forwarded by PE2 in the Inclusive multicast 601 tree for EVI1, adding the ESI label for ESI12 at the bottom of the 602 stack. The ESI label has been previously allocated and signaled by 603 the A-D routes for ESI12. Note that, as per [RFC7432], if the 604 result of the CE2 hashing is different and the frame sent to PE1, 605 PE1 SHOULD add the ESI label too (PE1 is the DF for EVI1 on 606 ESI12). 608 d) The MPLS-encapsulated frame gets to PE1 and PE3. PE1 609 de-encapsulate the Inclusive multicast tree label(s) and based on 610 the ESI label at the bottom of the stack, it decides to not 611 forward the frame to the ESI12. It will pop the ESI label and will 612 replicate it to CE1 though, since CE1 is not part of the ESI 613 identified by the ESI label. At PE3, the Inclusive multicast tree 614 label is popped and the frame forwarded to CE3. If a P2MP LSP is 615 used as Inclusive multicast tree for EVI1, PE3 will find an ESI 616 label after popping the P2MP LSP label. The ESI label will simply 617 be ignored and popped, since CE3 is not part of ESI12. 619 (3) Unicast frame example from CE3 to CE1: 621 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 622 and destination MAC CE1-MAC (we assume PE3 has previously resolved 623 an ARP request from CE3 to find the MAC of CE1-IP, and has added 624 CE3-MAC/CE3-IP to its proxy-ARP table). 626 b) Based on the CE-VID, the frame is identified to be forwarded in 627 the EVI1 context. A source MAC lookup is done in the MAC FIB 628 within the MAC-VRF-1 context and this time, since we assume CE3- 629 MAC is known, no further actions are carried out as a result of 630 the source lookup. A destination MAC lookup is performed next and 631 the label stack associated to the MAC CE1-MAC is found (including 632 the label associated to MAC-VRF-1 in PE1 and the P2P LSP label to 633 get to PE1). The unicast frame is then encapsulated and forwarded 634 to PE1. 636 c) At PE1, the packet is identified to be part of EVI1 and a 637 destination MAC lookup is performed in the MAC-VRF-1 context. The 638 labels are popped and the frame forwarded to CE1 with CE-VID=1. 640 Unicast frames from CE1 to CE3 or from CE2 to CE3 follow the same 641 procedures described above. 643 (4) Unicast frame example from CE3 to CE2: 645 a) A unicast frame with CE-VID=1 is issued from source MAC CE3-MAC 646 and destination MAC CE2-MAC (we assume PE3 has previously resolved 647 an ARP request from CE3 to find the MAC of CE2-IP). 649 b) Based on the CE-VID, the frame is identified to be forwarded in 650 the MAC-VRF-1 context. We assume CE3-MAC is known. A destination 651 MAC lookup is performed next and PE3 finds CE2-MAC associated to 652 PE2 on ESI12, an Ethernet Segment for which PE3 has two active A-D 653 routes per ESI (from PE1 and PE2) and two active A-D routes for 654 EVI1 (from PE1 and PE2). Based on a hashing function for the 655 frame, PE3 may decide to forward the frame using the label stack 656 associated to PE2 (label received from the MAC advertisement 657 route) or the label stack associated to PE1 (label received from 658 the A-D route per EVI for EVI1). Either way, the frame is 659 encapsulated and sent to the remote PE. 661 c) At PE2 (or PE1), the packet is identified to be part of EVI1 based 662 on the bottom label, and a destination MAC lookup is performed. At 663 either PE (PE2 or PE1), the FIB lookup yields a local ESI12 port 664 to which the frame is sent. 666 Unicast frames from CE1 to CE2 follow the same procedures. Aliasing 667 is possible in this case too, since ESI12 is local to PE1 and load 668 balancing through PE1 and PE2 may happen. 670 5.3. VLAN-bundle service procedures 672 Instead of using VLAN-based interfaces, the Service Provider can 673 choose to implement VLAN-bundle interfaces to carry the traffic for 674 the 4k CE-VIDs among CE1, CE2 and CE3. If that is the case, the 4k 675 CE-VIDs can be mapped to the same EVI, e.g. EVI200, at each PE. The 676 main advantage of this approach is the low control plane overhead 677 (reduced number of routes and labels) and easiness of provisioning, 678 at the expense of no control over the customer broadcast domains, 679 i.e. a single inclusive multicast tree for all the CE-VIDs and no CE- 680 VID translation in the Provider network. 682 5.3.1. Service startup procedures 684 As soon as the EVI200 is created in PE1, PE2 and PE3, the following 685 control plane actions are carried out: 687 o Flooding tree setup per EVI (one route): Each PE will send one 688 Inclusive Multicast Ethernet Tag route per EVI (hence only one 689 route per PE) so that the flooding tree per EVI can be setup. Note 690 that ingress replication or P2MP LSPs can optionally be signaled 691 in the PMSI Tunnel attribute and the corresponding tree be 692 created. 694 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 695 route for ESI12 will be issued from PE1 and PE2. This route will 696 include a single RT (RT for EVI200), an ESI Label extended 697 community with the active-standby flag set to zero (all-active 698 multi-homing type) and an ESI Label different from zero (used by 699 the non-DF for split-horizon functions). This route will be 700 imported by the three PEs, since the RT matches the EVI200 RT 701 locally configured. The A-D routes per ESI will be used for fast 702 convergence and split-horizon functions, as described in 703 [RFC7432]. 705 o Ethernet A-D routes per EVI (one route): An A-D route (EVI200) will 706 be sent by PE1 and PE2 for ESI12. This route includes the EVI200 707 RT and an MPLS label to be used by PE3 for the aliasing function. 708 This route will be imported by the three PEs. 710 5.3.2. Packet Walkthrough 712 The packet walkthrough for the VLAN-bundle case is similar to the one 713 described for EVI1 in the VLAN-based case except for the way the 714 CE-VID is handled by the ingress PE and the egress PE: 716 o No VLAN translation is allowed and the CE-VIDs are kept untouched 717 from CE to CE, i.e. the ingress CE-VID MUST be kept at the 718 imposition PE and at the disposition PE. 720 o The frame is identified to be forwarded in the MAC-VRF-200 context 721 as long as its CE-VID belongs to the VLAN-bundle defined in the 722 PE1/PE2/PE3 port to CE1/CE2/CE3. Our example is a special VLAN- 723 bundle case, since the entire CE-VID range is defined in the 724 ports, therefore any CE-VID would be part of EVI200. 726 Please refer to section 5.2.2 for more information about the control 727 plane and forwarding plane interaction for BUM and unicast traffic 728 from the different CEs. 730 5.4. VLAN-aware bundling service procedures 732 The last potential service type analyzed in this document is 733 VLAN-aware bundling. When this type of service interface is used to 734 carry the 4k CE-VIDs among CE1, CE2 and CE3, all the CE-VIDs will be 735 mapped to the same EVI, e.g. EVI300. The difference, compared to the 736 VLAN-bundle service type in the previous section, is that each 737 incoming CE-VID will also be mapped to a different "normalized" 738 Ethernet-Tag in addition to EVI300. If no translation is required, 739 the Ethernet-tag will match the CE-VID. Otherwise a translation 740 between CE-VID and Ethernet-tag will be needed at the imposition PE 741 and at the disposition PE. The main advantage of this approach is the 742 ability to control customer broadcast domains while providing a 743 single EVI to the customer. 745 5.4.1. Service startup procedures 746 As soon as the EVI300 is created in PE1, PE2 and PE3, the following 747 control plane actions are carried out: 749 o Flooding tree setup per EVI per Ethernet-Tag (4k routes): Each PE 750 will send one Inclusive Multicast Ethernet Tag route per EVI and 751 per Ethernet-Tag (hence 4k routes per PE) so that the flooding 752 tree per customer broadcast domain can be setup. Note that ingress 753 replication or P2MP LSPs can optionally be signaled in the PMSI 754 Tunnel attribute and the corresponding tree be created. In the 755 described use-case, since all the CE-VIDs and Ethernet-Tags are 756 defined on the three PEs, multicast tree aggregation might make 757 sense in order to save forwarding states. 759 o Ethernet A-D routes per ESI (one route for ESI12): A single A-D 760 route for ESI12 will be issued from PE1 and PE2. This route will 761 include a single RT (RT for EVI300), an ESI Label extended 762 community with the active-standby flag set to zero (all-active 763 multi-homing type) and an ESI Label different than zero (used by 764 the non-DF for split-horizon functions). This route will be 765 imported by the three PEs, since the RT matches the EVI300 RT 766 locally configured. The A-D routes per ESI will be used for fast 767 convergence and split-horizon functions, as described in 768 [RFC7432]. 770 o Ethernet A-D routes per EVI (one route): An A-D route (EVI300) will 771 be sent by PE1 and PE2 for ESI12. This route includes the EVI300 772 RT and an MPLS label to be used by PE3 for the aliasing function. 773 This route will be imported by the three PEs. 775 5.4.2. Packet Walkthrough 777 The packet walkthrough for the VLAN-aware case is similar to the one 778 described before. Compared to the other two cases, VLAN-aware 779 services allow for CE-VID translation and for an N:1 CE-VID to EVI 780 mapping. Both things are not supported at once in either of the two 781 other service interfaces. Note that this model requires qualified 782 learning on the MAC FIBs. Some differences compared to the packet 783 walkthrough described in section 5.2.2 are: 785 o At the ingress PE, the frames are identified to be forwarded in the 786 EVI300 context as long as their CE-VID belong to the range defined 787 in the PE port to the CE. In addition to it, CE-VID=x is mapped to 788 a "normalized" Ethernet-Tag=y at the MAC-VRF-300 (where x and y 789 might be equal if no translation is needed). Qualified learning is 790 now required (a different FIB space is allocated within MAC-VRF- 791 300 for each Ethernet-Tag). Potentially the same MAC could be 792 learnt in two different Ethernet-Tag bridge domains of the same 793 MAC-VRF. 795 o Any new locally learnt MAC on the MAC-VRF-300/Ethernet-Tag=y 796 interface is advertised by the ingress PE in a MAC advertisement 797 route, using now the Ethernet-Tag field (Ethernet-Tag=y) so that 798 the remote PE learns the MAC associated to the MAC-VRF- 799 300/Ethernet-Tag=y FIB. Note that the Ethernet-Tag field is not 800 used in advertisements of MACs learnt on VLAN-based or VLAN-bundle 801 service interfaces. 803 o At the ingress PE, BUM frames are sent to the corresponding 804 flooding tree for the particular Ethernet-Tag they are mapped to. 805 Each individual Ethernet-Tag can have a different flooding tree 806 within the same EVI300. For instance, Ethernet-Tag=y can use 807 ingress replication to get to the remote PEs whereas Ethernet- 808 Tag=z can use a p2mp LSP. 810 o At the egress PE, Ethernet-Tag=y, for a given broadcast domain 811 within MAC-VRF-300, can be translated to egress CE-VID=x. That is 812 not possible for VLAN-bundle interfaces. It is possible for VLAN- 813 based interfaces, but it requires a separate EVI per CE-VID. 815 6. MPLS-based forwarding model use-case 817 EVPN supports an alternative forwarding model, usually referred to as 818 MPLS-based forwarding or disposition model as opposed to the 819 MAC-based forwarding or disposition model described in section 5. 820 Using MPLS-based forwarding model instead of MAC-based model might 821 have an impact on: 823 o The number of forwarding states required 825 o The FIB where the forwarding states are handled: MAC FIB or MPLS 826 LFIB. 828 The MPLS-based forwarding model avoids the destination MAC lookup at 829 the egress PE MAC FIB, at the expense of increasing the number of 830 next-hop forwarding states at the egress MPLS LFIB. This also has an 831 impact on the control plane and the label allocation model, since an 832 MPLS-based disposition PE MUST send as many routes and labels as 833 required next-hops in the egress MAC-VRF. This concept is equivalent 834 to the forwarding models supported in IP-VPNs at the egress PE, where 835 an IP lookup in the IP-VPN FIB might be necessary or not depending on 836 the available next-hop forwarding states in the LFIB. 838 The following sub-sections highlight the impact on the control and 839 data plane procedures described in section 5 when and MPLS-based 840 forwarding model is used. 842 Note that both forwarding models are compatible and interoperable in 843 the same network. The implementation of either model in each PE is a 844 local decision to the PE node. 846 6.1. Impact of MPLS-based forwarding on the EVPN network startup 848 The MPLS-based forwarding model has no impact on the procedures 849 explained in section 5.1. 851 6.2. Impact of MPLS-based forwarding on the VLAN-based service 852 procedures 854 Compared to the MAC-based forwarding model, the MPLS-based forwarding 855 model has no impact in terms of number of routes, when all the 856 service interfaces are VLAN-based. The differences for the use-case 857 described in this document are summarized in the following list: 859 o Flooding tree setup per EVI (4k routes per PE): no impact compared 860 to the MAC-based model. 862 o Ethernet A-D routes per ESI (one set of routes for ESI12 per PE): 863 no impact compared to the MAC-based model. 865 o Ethernet A-D routes per EVI (4k routes per PE/ESI): no impact 866 compared to the MAC-based model. 868 o MAC-advertisement routes: instead of allocating and advertising the 869 same MPLS label for all the new MACs locally learnt on the same 870 MAC-VRF, a different label MUST be advertised per CE next-hop or 871 MAC so that no MAC FIB lookup is needed at the egress PE. In 872 general, this means that a different label at least per CE must be 873 advertised, although the PE can decide to implement a label per 874 MAC if more granularity (hence less scalability) is required in 875 terms of forwarding states. E.g. if CE2 sends traffic from two 876 different MACs to PE1, CE2-MAC1 and CE2-MAC2, the same MPLS 877 label=x can be re-used for both MAC advertisements since they both 878 share the same source ESI12. It is up to the PE1 implementation to 879 use a different label per individual MAC within the same ES 880 Segment (even if only one label per ESI is enough). 882 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 883 learning new local CE MAC addresses on the data plane, but will 884 rather add forwarding states to the MPLS LFIB. 886 6.3. Impact of MPLS-based forwarding on the VLAN-bundle service 887 procedures 889 Compared to the MAC-based forwarding model, the MPLS-based forwarding 890 model has no impact in terms of number of routes when all the service 891 interfaces are VLAN-bundle type. The differences for the use-case 892 described in this document are summarized in the following list: 894 o Flooding tree setup per EVI (one route): no impact compared to the 895 MAC-based model. 897 o Ethernet A-D routes per ESI (one route for ESI12 per PE): no impact 898 compared to the MAC-based model. 900 o Ethernet A-D routes per EVI (one route per PE/ESI): no impact 901 compared to the MAC-based model since no VLAN translation is 902 required. 904 o MAC-advertisement routes: instead of allocating and advertising the 905 same MPLS label for all the new MACs locally learnt on the same 906 MAC-VRF, a different label MUST be advertised per CE next-hop or 907 MAC so that no MAC FIB lookup is needed at the egress PE. In 908 general, this means that a different label at least per CE must be 909 advertised, although the PE can decide to implement a label per 910 MAC if more granularity (hence less scalability) is required in 911 terms of forwarding states. It is up to the PE1 implementation to 912 use a different label per individual MAC within the same ES 913 Segment (even if only one label per ESI is enough). 915 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 916 learning new local CE MAC addresses on the data plane, but will 917 rather add forwarding states to the MPLS LFIB. 919 6.4. Impact of MPLS-based forwarding on the VLAN-aware service 920 procedures 922 Compared to the MAC-based forwarding model, the MPLS-based forwarding 923 model has definitively an impact in terms of number of A-D routes 924 when all the service interfaces are VLAN-aware bundle type. The 925 differences for the use-case described in this document are 926 summarized in the following list: 928 o Flooding tree setup per EVI (4k routes per PE): no impact compared 929 to the MAC-based model. 931 o Ethernet A-D routes per ESI (one route for ESI12 per PE): no impact 932 compared to the MAC-based model. 934 o Ethernet A-D routes per EVI (4k routes per PE/ESI): PE1 and PE2 935 will send 4k routes for EVI300, one per 936 tuple. This will allow the egress PE to find out all the 937 forwarding information in the MPLS LFIB and even support Ethernet- 938 Tag to CE-VID translation at the egress. The MAC-based forwarding 939 model would allow the PEs to send a single route per PE/ESI for 940 EVI300, since the packet with the embedded Ethernet-Tag would be 941 used to perform a MAC lookup and find out the egress CE-VID. 943 o MAC-advertisement routes: instead of allocating and advertising the 944 same MPLS label for all the new MACs locally learnt on the same 945 MAC-VRF, a different label MUST be advertised per CE next-hop or 946 MAC so that no MAC FIB lookup is needed at the egress PE. In 947 general, this means that a different label at least per CE must be 948 advertised, although the PE can decide to implement a label per 949 MAC if more granularity (hence less scalability) is required in 950 terms of forwarding states. It is up to the PE1 implementation to 951 use a different label per individual MAC within the same ES 952 Segment. Note that, in this model, the Ethernet-Tag will be set to 953 a non-zero value for the MAC-advertisement routes. The same MAC 954 address can be announced with different Ethernet-Tag value. This 955 will make the advertising PE install two different forwarding 956 states in the MPLS LFIB. 958 o PE1, PE2 and PE3 will not add forwarding states to the MAC FIB upon 959 learning new local CE MAC addresses on the data plane, but will 960 rather add forwarding states to the MPLS LFIB. 962 7. Comparison between MAC-based and MPLS-based forwarding models 964 Both forwarding models are possible in a network deployment and each 965 one has its own trade-offs. 967 The MAC-based forwarding model can save A-D routes per EVI when VLAN- 968 aware bundling services are deployed and therefore reduce the control 969 plane overhead. This model also saves a significant amount of MPLS 970 labels compared to the MPLS-based forwarding model. All the MACs and 971 A-D routes for the same EVI can signal the same MPLS label, saving 972 labels from the local PE space. A MAC FIB lookup at the egress PE is 973 required in order to do so. 975 The MPLS-based forwarding model can save forwarding states at the 976 egress PEs if labels per next hop CE (as opposed to per MAC) are 977 implemented. No egress MAC lookup is required. An A-D route per is required for VLAN-aware services, as opposed to an 979 A-D route per EVI. Also, a different label per next-hop CE per MAC- 980 VRF is consumed, as opposed to a single label per MAC-VRF. 982 The following table summarizes the implementation details of both 983 models for the VLAN-aware bundling service type. 985 +-----------------------------+----------------+----------------+ 986 | 4k CE-VID VLANs | MAC-based | MPLS-based | 987 | | Model | Model | 988 +-----------------------------+----------------+----------------+ 989 | A-D routes/EVI | 1 per ESI/EVI | 4k per ESI/EVI | 990 | MPLS labels consumed | 1 per MAC-VRF | 1 per CE/EVI | 991 | Egress PE Forwarding states | 1 per MAC | 1 per next-hop | 992 | Egress PE Lookups | 2 (MPLS+MAC) | 1 (MPLS) | 993 +-----------------------------+----------------+----------------+ 995 The egress forwarding model is an implementation local to the egress 996 PE and is independent of the model supported on the rest of the PEs, 997 i.e. in our use-case, PE1, PE2 and PE3 could have either egress 998 forwarding model without any dependencies. 1000 8. Traffic flow optimization 1002 In addition to the procedures described across sections 1 through 7, 1003 EVPN [RFC7432] procedures allow for optimized traffic handling in 1004 order to minimize unnecessary flooding across the entire 1005 infrastructure. Optimization is provided through specific ARP 1006 termination and the ability to block unknown unicast flooding. 1007 Additionally, EVPN procedures allow for intelligent, close to the 1008 source, inter-subnet forwarding and solves the commonly known sub- 1009 optimal routing problem. Besides the traffic efficiency, ingress 1010 based inter-subnet forwarding also optimizes packet forwarding rules 1011 and implementation at the egress nodes as well. Details of these 1012 procedures are outlined in sections 8.1 and 8.2. 1014 8.1. Control Plane Procedures 1016 8.1.1. MAC learning options 1018 The fundamental premise of [RFC7432] is the notion of a different 1019 approach to MAC address learning compared to traditional IEEE 802.1 1020 bridge learning methods; specifically EVPN differentiates between 1021 data and control plane driven learning mechanisms. 1023 Data driven learning implies that there is no separate communication 1024 channel used to advertise and propagate MAC addresses. Rather, MAC 1025 addresses are learned through IEEE defined bridge-learning procedures 1026 as well as by snooping on DHCP and ARP requests. As different MAC 1027 addresses show up on different ports, the L2 FIB is populated with 1028 the appropriate MAC addresses. 1030 Control plane driven learning implies a communication channel that 1031 could be either a control-plane protocol or a management-plane 1032 mechanism. In the context of EVPN, two different learning procedures 1033 are defined, i.e. local and remote procedures: 1035 o Local learning defines the procedures used for learning the MAC 1036 addresses of network elements locally connected to a MAC-VRF. 1037 Local learning could be implemented through all three learning 1038 procedures: control plane, management plane as well as data plane. 1039 However, the expectation is that for most of the use cases, local 1040 learning through data plane should be sufficient. 1042 o Remote learning defines the procedures used for learning MAC 1043 addresses of network elements remotely connected to a MAC-VRF, 1044 i.e. far-end PEs. Remote learning procedures defined in [RFC7432] 1045 advocate using only control plane learning; specifically BGP. 1046 Through the use of BGP EVPN NLRIs, the remote PE has the 1047 capability of advertising all the MAC addresses present in its 1048 local FIB. 1050 8.1.2. Proxy-ARP/ND 1052 In EVPN, MAC addresses are advertised via the MAC/IP Advertisement 1053 Route, as discussed in [RFC7432]. Optionally an IP address can be 1054 advertised along with the MAC address announcement. However, there 1055 are certain rules put in place in terms of IP address usage: if the 1056 MAC Advertisement Route contains an IP address, and the IP Address 1057 Length is 32 bits (or 128 in the IPv6 case), this particular IP 1058 address correlates directly with the advertised MAC address. Such 1059 advertisement allows us to build a proxy-ARP/ND table populated with 1060 the IP<->MAC bindings received from all the remote nodes. 1062 Furthermore, based on these bindings, a local MAC-VRF can now provide 1063 Proxy-ARP/ND functionality for all ARP requests and ND solicitations 1064 directed to the IP address pool learned through BGP. Therefore, the 1065 amount of unnecessary L2 flooding, ARP/ND requests/solicitations in 1066 this case, can be further reduced by the introduction of Proxy-ARP/ND 1067 functionality across all EVI MAC-VRFs. 1069 8.1.3. Unknown Unicast flooding suppression 1071 Given that all locally learned MAC addresses are advertised through 1072 BGP to all remote PEs, suppressing flooding of any Unknown Unicast 1073 traffic towards the remote PEs is a feasible network optimization. 1075 The assumption in the use case is made that any network device that 1076 appears on a remote MAC-VRF will somehow signal its presence to the 1077 network. This signaling can be done through e.g. gratuitous ARPs. 1078 Once the remote PE acknowledges the presence of the node in the MAC- 1079 VRF, it will do two things: install its MAC address in its local FIB 1080 and advertise this MAC address to all other BGP speakers via EVPN 1081 NLRI. Therefore, we can assume that any active MAC address is 1082 propagated and learnt through the entire EVI. Given that MAC 1083 addresses become pre-populated - once nodes are alive on the network 1084 - there is no need to flood any unknown unicast towards the remote 1085 PEs. If the owner of a given destination MAC is active, the BGP route 1086 will be present in the local RIB and FIB, assuming that the BGP 1087 import policies are successfully applied; otherwise, the owner of 1088 such destination MAC is not present on the network. 1090 It is worth noting that unless: a) control or management plane 1091 learning is performed through the entire EVI or b) all the EVI- 1092 attached devices signal their presence when they come up (GARPs or 1093 similar), unknown unicast flooding MUST be enabled. 1095 8.1.4. Optimization of Inter-subnet forwarding 1097 In a scenario in which both L2 and L3 services are needed over the 1098 same physical topology, some interaction between EVPN and IP-VPN is 1099 required. A common way of stitching the two service planes is through 1100 the use of an IRB interface, which allows for traffic to be either 1101 routed or bridged depending on its destination MAC address. If the 1102 destination MAC address is the one of the IRB interface, traffic 1103 needs to be passed through a routing module and potentially be either 1104 routed to a remote PE or forwarded to a local subnet. If the 1105 destination MAC address is not the one of the IRB, the MAC-VRF 1106 follows standard bridging procedures. 1108 A typical example of EVPN inter-subnet forwarding would be a scenario 1109 in which multiple IP subnets are part of a single or multiple EVIs, 1110 and they all belong to a single IP-VPN. In such topologies, it is 1111 desired that inter-subnet traffic can be efficiently routed without 1112 any tromboning effects in the network. Due to the overlapping 1113 physical and service topology in such scenarios, all inter-subnet 1114 connectivity will be locally routed trough the IRB interface. 1116 In addition to optimizing the traffic patterns in the network, local 1117 inter-subnet forwarding also optimizes greatly the amount of 1118 processing needed to cross the subnets. Through EVPN MAC 1119 advertisements, the local PE learns the real destination MAC address 1120 associated with the remote IP address and the inter-subnet forwarding 1121 can happen locally. When the packet is received at the egress PE, it 1122 is directly mapped to an egress MAC-VRF, bypassing any egress IP-VPN 1123 processing. 1125 Please refer to [EVPN-INTERSUBNET] for more information about the IP 1126 inter-subnet forwarding procedures in EVPN. 1128 8.2. Packet Walkthrough Examples 1129 Assuming that the services are setup according to figure 1 in section 1130 2, the following flow optimization processes will take place in terms 1131 of creating, receiving and forwarding packets across the network. 1133 8.2.1. Proxy-ARP example for CE2 to CE3 traffic 1135 Using figure 1 in section 2, consider EVI 400 residing on PE1, PE2 1136 and PE3 connecting CE2 and CE3 networks. Also, consider that PE1 and 1137 PE2 are part of the all-active multi-homing ES for CE2, and that PE2 1138 is elected designated-forwarder for EVI400. We assume that all the 1139 PEs implement the proxy-ARP functionality in the MAC-VRF-400 context. 1141 In this scenario, PE3 will not only advertise the MAC addresses 1142 through the EVPN MAC Advertisement Route but also IP addresses of 1143 individual hosts, i.e. /32 prefixes, behind CE3. Upon receiving the 1144 EVPN routes, PE1 and PE2 will install the MAC addresses in the MAC- 1145 VRF-400 FIB and based on the associated received IP addresses, PE1 1146 and PE2 can now build a proxy-ARP table within the context of MAC- 1147 VRF-400. 1149 From the forwarding perspective, when a node behind CE2 sends a frame 1150 destined to a node behind CE3, it will first send an ARP request to 1151 e.g. PE2 (based on the result of the CE2 hashing). Assuming that PE2 1152 has populated its proxy-ARP table for all active nodes behind the 1153 CE3, and that the IP address in the ARP message matches the entry in 1154 the table, PE2 will respond to the ARP request with the actual MAC 1155 address on behalf of the node behind CE3. 1157 Once the nodes behind CE2 learn the actual MAC address of the nodes 1158 behind CE3, all the MAC-to-MAC communications between the two 1159 networks will be unicast. 1161 8.2.2. Flood suppression example for CE1 to CE3 traffic 1163 Using figure 1 in section 2, consider EVI 500 residing on PE1 and PE3 1164 connecting CE1 and CE3 networks. Consider that both PE1 and PE3 have 1165 disabled unknown unicast flooding for this specific EVI context. Once 1166 the network devices behind CE3 come online they will learn their MAC 1167 addresses and create local FIB entries for these devices. Note that 1168 local FIB entries could also be created through either a control or 1169 management plane between PE and CE as well. Consequently, PE3 will 1170 automatically create EVPN Type 2 MAC Advertisement Routes and 1171 advertise all locally learned MAC addresses. The routes will also 1172 include the corresponding MPLS label. 1174 Given that PE1 automatically learns and installs all MAC addresses 1175 behind CE3, its MAC-VRF FIB will already be pre-populated with the 1176 respective next-hops and label assignments associated with the MAC 1177 addresses behind CE3. As such, as soon as the traffic sent by CE1 to 1178 nodes behind CE3 is received into the context of EVI 500, PE1 will 1179 push the MPLS Label(s) onto the original Ethernet frame and send the 1180 packet to the MPLS network. As usual, once PE3 receives this packet, 1181 and depending on the forwarding model, PE3 will either do a next-hop 1182 lookup in the EVI 500 context, or will just forward the traffic 1183 directly to the CE3. In the case that PE1 MAC-VRF-500 does not have a 1184 MAC entry for a specific destination that CE1 is trying to reach, PE1 1185 will drop the frame since unknown unicast flooding is disabled. 1187 Based on the assumption that all the MAC entries behind the CEs are 1188 pre-populated through gratuitous-ARP and/or DHCP requests, if one 1189 specific MAC entry is not present in the MAC-VRF-500 FIB on PE1, the 1190 owner of that MAC is not alive on the network behind the CE3, hence 1191 the traffic can be dropped at PE1 instead of be flooded and consume 1192 network bandwidth. 1194 8.2.3. Optimization of inter-subnet forwarding example for CE3 to CE2 1195 traffic 1197 Using figure 1 in section 2 consider that there is an IP-VPN 666 1198 context residing on PE1, PE2 and PE3 which connects CE1, CE2 and CE3 1199 into a single IP-VPN domain. Also consider that there are two EVIs 1200 present on the PEs, EVI 600 and EVI 60. Each IP subnet is associated 1201 to a different MAC-VRF context. Thus there is a single subnet, subnet 1202 600, between CE1 and CE3 that is established through EVI 600. 1203 Similarly, there is another subnet, subnet 60, between CE2 and CE3 1204 that is established through EVI 60. Since both subnets are part of 1205 the same IP VPN, there is a mapping of each EVI (or individual 1206 subnet) to a local IRB interface on the three PEs. 1208 If a node behind CE2 wants to communicate with a node on the same 1209 subnet seating behind CE3, the communication flow will follow the 1210 standard EVPN procedures, i.e. FIB lookup within the PE1 (or PE2) 1211 after adding the corresponding EVPN label to the MPLS label stack 1212 (downstream label allocation from PE3 for EVI 60). 1214 When it comes to crossing the subnet boundaries, the ingress PE 1215 implements local inter-subnet forwarding. For example, when a node 1216 behind CE2 (EVI 60) sends a packet to a node behind CE1 (EVI 600) the 1217 destination IP address will be in the subnet 600, but the destination 1218 MAC address will be the address of source node's default gateway, 1219 which in this case will be an IRB interface on PE1 (connecting EVI 60 1220 to IP-VPN 666). Once PE1 sees the traffic destined to its own MAC 1221 address, it will route the packet to EVI 600, i.e. it will change the 1222 source MAC address to the one of the IRB interface in EVI 600 and 1223 change the destination MAC address to the address belonging to the 1224 node behind CE1, which is already populated in the MAC-VRF-600 FIB, 1225 either through data or control plane learning. 1227 An important optimization to be noted is the local inter-subnet 1228 forwarding in lieu of IP VPN routing. If the node from subnet 60 1229 (behind CE2) is sending a packet to the remote end node on subnet 600 1230 (behind CE3), the mechanism in place still honors the local inter- 1231 subnet (inter-EVI) forwarding. 1233 In our use-case, therefore, when node from subnet 60 behind CE2 sends 1234 traffic to the node on subnet 600 behind CE3, the destination MAC 1235 address is the PE1 MAC-VRF-60 IRB MAC address. However, once the 1236 traffic locally crosses EVIs, to EVI 600, via the IRB interface on 1237 PE1, the source MAC address is changed to that of the IRB interface 1238 and the destination MAC address is changed to the one advertised by 1239 PE3 via EVPN and already installed in MAC-VRF-600. The rest of the 1240 forwarding through PE1 is using the MAC-VRF-600 forwarding context 1241 and label space. 1243 Another very relevant optimization is due to the fact that traffic 1244 between PEs is forwarded through EVPN, rather than through IP-VPN. In 1245 the example described above for traffic from EVI 60 on CE2 to EVI 600 1246 on CE3, there is no need for IP-VPN processing on the egress PE3. 1247 Traffic is forwarded either to the EVI 600 context in PE3 for further 1248 MAC lookup and next-hop processing, or directly to the node behind 1249 CE3, depending on the egress forwarding model being used. 1251 9. Conventions used in this document 1253 In the examples, the following conventions are used: 1255 o CE-VIDs refer to the VLAN tag identifiers being used at CE1, CE2 1256 and CE3 to tag customer traffic sent to the Service Provider E- 1257 VPN network 1259 o CE1-MAC, CE2-MAC and CE3-MAC refer to source MAC addresses "behind" 1260 each CE respectively. Those MAC addresses can belong to the CEs 1261 themselves or to devices connected to the CEs. 1263 o CE1-IP, CE2-IP and CE3-IP refer to IP addresses associated to the 1264 above MAC addresses. 1266 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 1267 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 1268 document are to be interpreted as described in RFC-2119 [RFC2119]. 1270 In this document, these words will appear with that interpretation 1271 only when in ALL CAPS. Lower case uses of these words are not to be 1272 interpreted as carrying RFC-2119 significance. 1274 10. Security Considerations 1276 Please refer to the "Security Considerations" section in [RFC7432]. 1278 11. IANA Considerations 1280 No new IANA considerations are needed. 1282 12. References 1284 12.1. Normative References 1286 [RFC4761]Kompella, K., Ed., and Y. Rekhter, Ed., "Virtual Private LAN 1287 Service (VPLS) Using BGP for Auto-Discovery and Signaling", RFC 4761, 1288 DOI 10.17487/RFC4761, January 2007, . 1291 [RFC4762]Lasserre, M., Ed., and V. Kompella, Ed., "Virtual Private 1292 LAN Service (VPLS) Using Label Distribution Protocol (LDP) 1293 Signaling", RFC 4762, DOI 10.17487/RFC4762, January 2007, 1294 . 1296 [RFC6074]Rosen, E., Davie, B., Radoaca, V., and W. Luo, 1297 "Provisioning, Auto-Discovery, and Signaling in Layer 2 Virtual 1298 Private Networks (L2VPNs)", RFC 6074, DOI 10.17487/RFC6074, January 1299 2011, . 1301 [RFC4364]Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1302 Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February 2006, 1303 . 1305 [RFC7209]Sajassi, A., Aggarwal, R., Uttaro, J., Bitar, N., 1306 Henderickx, W., and A. Isaac, "Requirements for Ethernet VPN (EVPN)", 1307 RFC 7209, DOI 10.17487/RFC7209, May 2014, . 1310 [RFC7117]Aggarwal, R., Ed., Kamite, Y., Fang, L., Rekhter, Y., and C. 1311 Kodeboniya, "Multicast in Virtual Private LAN Service (VPLS)", 1312 RFC 7117, DOI 10.17487/RFC7117, February 2014, . 1315 [RFC7432]Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 1316 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based Ethernet 1317 VPN", RFC 7432, DOI 10.17487/RFC7432, February 2015, . 1320 12.2. Informative References 1322 [EVPN-INTERSUBNET] Sajassi et al., "IP Inter-subnet forwarding in 1323 EVPN", draft-ietf-bess-evpn-inter-subnet-forwarding-01.txt 1325 13. Acknowledgments 1327 The authors want to thank Giles Heron for his detailed review of the 1328 document. We also thank Stefan Plug, and Eric Wunan for their 1329 comments. 1331 14. Contributors 1332 In addition to the authors listed on the front page, the following 1333 co-authors have also contributed to this document: 1335 Florin Balus 1337 14. Authors' Addresses 1339 Jorge Rabadan 1340 Nokia 1341 777 E. Middlefield Road 1342 Mountain View, CA 94043 USA 1343 Email: jorge.rabadan@nokia.com 1345 Senad Palislamovic 1346 Nokia 1347 Email: senad.palislamovic@nokia.com 1349 Wim Henderickx 1350 Nokia 1351 Email: wim.henderickx@nokia.com 1353 Keyur Patel 1354 Cisco 1355 Email: keyupate@cisco.com 1357 Ali Sajassi 1358 Cisco 1359 Email: sajassi@cisco.com 1361 James Uttaro 1362 AT&T 1363 Email: uttaro@att.com 1365 Aldrin Isaac 1366 Juniper Networks 1367 Email: aisaac@juniper.net 1369 Truman Boyes 1370 Bloomberg 1371 Email: tboyes@bloomberg.net