idnits 2.17.1 draft-ietf-idr-bgp-optimal-route-reflection-07.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 ([RFC4456]), 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 == Line 84 has weird spacing: '...routing from...' == Line 95 has weird spacing: '... potato routi...' == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (July 31, 2014) is 3550 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'RFC2119' is defined on line 926, but no explicit reference was found in the text == Unused Reference: 'RFC4271' is defined on line 929, but no explicit reference was found in the text == Unused Reference: 'RFC4360' is defined on line 932, but no explicit reference was found in the text == Unused Reference: 'RFC5492' is defined on line 935, but no explicit reference was found in the text == Unused Reference: 'I-D.ietf-idr-add-paths' is defined on line 940, but no explicit reference was found in the text == Unused Reference: 'RFC1997' is defined on line 945, but no explicit reference was found in the text == Unused Reference: 'RFC1998' is defined on line 948, but no explicit reference was found in the text == Unused Reference: 'RFC4384' is defined on line 952, but no explicit reference was found in the text == Unused Reference: 'RFC4893' is defined on line 959, but no explicit reference was found in the text == Unused Reference: 'RFC5668' is defined on line 966, but no explicit reference was found in the text == Unused Reference: 'RFC6774' is defined on line 972, but no explicit reference was found in the text == Outdated reference: A later version (-15) exists of draft-ietf-idr-add-paths-09 -- Obsolete informational reference (is this intentional?): RFC 4893 (Obsoleted by RFC 6793) Summary: 1 error (**), 0 flaws (~~), 16 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IDR Working Group R. Raszuk 3 Internet-Draft Individual 4 Intended status: Standards Track C. Cassar 5 Expires: February 1, 2015 Cisco Systems 6 E. Aman 7 TeliaSonera 8 B. Decraene 9 S. Litkowski 10 Orange 11 July 31, 2014 13 BGP Optimal Route Reflection (BGP-ORR) 14 draft-ietf-idr-bgp-optimal-route-reflection-07 16 Abstract 18 [RFC4456] asserts that, because the Interior Gateway Protocol (IGP) 19 cost to a given point in the network will vary across routers, "the 20 route reflection approach may not yield the same route selection 21 result as that of the full IBGP mesh approach." One practical 22 implication of this assertion is that the deployment of route 23 reflection may thwart the ability to achieve hot potato routing. Hot 24 potato routing attempts to direct traffic to the closest AS egress 25 point in cases where no higher priority policy dictates otherwise. 26 As a consequence of the route reflection method, the choice of exit 27 point for a route reflector and its clients will be the egress point 28 closest to the route reflector - and not necessarily closest to the 29 RR clients. 31 Section 11 of [RFC4456] describes a deployment approach and a set of 32 constraints which, if satsified, would result in the deployment of 33 route reflection yielding the same results as the iBGP full mesh 34 approach. Such a deployment approach would make route reflection 35 compatible with the application of hot potato routing policy. 37 As networks evolved to accommodate architectural requirements of new 38 services, tunneled (LSP/IP tunneling) networks with centralized route 39 reflectors became commonplace. This is one type of common deployment 40 where it would be impractical to satisfy the constraints described in 41 Section 11 of [RFC4456]. Yet, in such an environment, hot potato 42 routing policy remains desirable. 44 This document proposes two new solutions which can be deployed to 45 facilitate the application of closest exit point policy centralized 46 route reflection deployments. 48 Status of This Memo 50 This Internet-Draft is submitted in full conformance with the 51 provisions of BCP 78 and BCP 79. 53 Internet-Drafts are working documents of the Internet Engineering 54 Task Force (IETF). Note that other groups may also distribute 55 working documents as Internet-Drafts. The list of current Internet- 56 Drafts is at http://datatracker.ietf.org/drafts/current/. 58 Internet-Drafts are draft documents valid for a maximum of six months 59 and may be updated, replaced, or obsoleted by other documents at any 60 time. It is inappropriate to use Internet-Drafts as reference 61 material or to cite them other than as "work in progress." 63 This Internet-Draft will expire on February 1, 2015. 65 Copyright Notice 67 Copyright (c) 2014 IETF Trust and the persons identified as the 68 document authors. All rights reserved. 70 This document is subject to BCP 78 and the IETF Trust's Legal 71 Provisions Relating to IETF Documents 72 (http://trustee.ietf.org/license-info) in effect on the date of 73 publication of this document. Please review these documents 74 carefully, as they describe your rights and restrictions with respect 75 to this document. Code Components extracted from this document must 76 include Simplified BSD License text as described in Section 4.e of 77 the Trust Legal Provisions and are provided without warranty as 78 described in the Simplified BSD License. 80 Table of Contents 82 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 83 2. Proposed solutions . . . . . . . . . . . . . . . . . . . . . 5 84 3. Best path selection for BGP hot potato routing from 85 customized IGP network position . . . . . . . . . . . . . . . 6 86 3.1. Client's perspective best path selection algorithm . . . 7 87 3.1.1. Flat IGP network . . . . . . . . . . . . . . . . . . 7 88 3.1.2. Hierarchical IGP network . . . . . . . . . . . . . . 8 89 3.2. Aside: Configuration-based flexible route reflector 90 placement . . . . . . . . . . . . . . . . . . . . . . . . 9 91 3.3. Route reflector client grouping . . . . . . . . . . . . . 10 92 3.3.1. Route Reflector Client Group ID . . . . . . . . . . . 10 93 3.4. Discussion . . . . . . . . . . . . . . . . . . . . . . . 11 94 3.5. Advantages . . . . . . . . . . . . . . . . . . . . . . . 12 95 4. Angular distance approximation for BGP warm potato routing . 13 96 4.1. Problem statement . . . . . . . . . . . . . . . . . . . . 13 97 4.2. Proposed solution . . . . . . . . . . . . . . . . . . . . 14 98 4.3. Centralized vs distributed route reflectors . . . . . . . 15 99 5. Client's perspective policy based best path selection . . . . 16 100 5.1. Proposal . . . . . . . . . . . . . . . . . . . . . . . . 17 101 5.2. Example . . . . . . . . . . . . . . . . . . . . . . . . . 17 102 5.3. Avoiding routing loops . . . . . . . . . . . . . . . . . 18 103 6. Deployment considerations . . . . . . . . . . . . . . . . . . 19 104 7. Security considerations . . . . . . . . . . . . . . . . . . . 20 105 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 106 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 20 107 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 108 10.1. Normative References . . . . . . . . . . . . . . . . . . 20 109 10.2. Informative References . . . . . . . . . . . . . . . . . 21 110 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 21 112 1. Introduction 114 There are three types of BGP deployments within Autonomous Systems 115 today: full mesh, confederations and route reflection. 117 BGP route reflection is the most popular way to distribute BGP routes 118 between BGP speakers belonging to the same administrative domain. 119 Traditionally route reflectors have been deployed in the forwarding 120 path and carefully placed on the POP to core boundaries. That model 121 of BGP route reflector placement has started to evolve. The 122 placement of route reflectors outside the forwarding path was 123 triggered by applications which required traffic to be tunneled from 124 AS ingress PE to egress PE: for example L3VPN. 126 This evolving model of intra-domain network design has enabled 127 deployments of centralized route reflectors. Initially this model 128 was only employed for new address families e.g. L3VPNs, L2VPNs etc 130 With edge to edge MPLS or IP encapsulation also being used to carry 131 internet traffic, this model has been gradually extended to other BGP 132 address families including IPv4 and IPv6 Internet routing. This is 133 also applicable to new services achieved with BGP as control plane 134 for example 6PE. 136 Such centralized route reflectors can be placed on the POP to core 137 boundaries, but they are often placed in arbitrary locations in the 138 core of large networks. 140 Such deployments suffer from a critical drawback in the context of 141 best path selection. A route reflector with knowledge of multiple 142 paths for a given prefix will pick the best path and only advertise 143 that best path to the the route reflector clients. If the best path 144 for a prefix is selected on the basis of an IGP tie break, the best 145 path advertised from the route reflector to its clients will be the 146 exit point closest to the route reflector. But route reflector 147 clients will be in a place in the network toplogy which is different 148 from the route reflector. In networks with centralized route 149 reflectors, this difference will be even more acute. It follows that 150 the best path chosen by the route reflector is not necessarily the 151 same as the path which would have been chosen by the client if the 152 client considered the same set of candidate paths as the route 153 reflector. Furthermore, the path chosen by the client might have 154 been a better path from that chosen by the route reflector for 155 traffic entering the network at the client. The path chosen by the 156 client would have guaranteed the lowest cost and delay trajectory 157 through the network. 159 Route reflector clients switch packets using routing information 160 learnt from route reflectors which are not on the forwarding path of 161 the packet through the network even in the absence of end-to-end 162 encapsulation. In those cases the path chosen as best and propagated 163 to the clients will often not be the optimal path chosen by the 164 client given all available paths. 166 Eliminating the IGP distance to the BGP nexthop as a tie breaker on 167 centralized route reflectors does not address the issue. Ignoring 168 IGP distance to the BGP next hop results in the tie breaking 169 procedure contributing the best path by differentiating between paths 170 using attributes otherwise considered less important than IGP cost to 171 the BGP nexthop. 173 One possible valid solution or workaround to this problem requires 174 sending all domain external paths from the RR to all its clients. 175 This approach suffers the significant drawback of pushing a large 176 amount of BGP state to all the edge routers. In many networks, the 177 number of EBGP peers over which full Internet routing information is 178 received would correlate directly to the number of paths present in 179 each ASBR. This could easily result in tens of paths for each 180 prefix. 182 Notwithstanding this drawback, there are a number of reasons for 183 sending more than just the single best path to the clients. Improved 184 path diversity at the edge is a requirement for fast connectivity 185 restoration, and a requirement for effective BGP level load 186 balancing. 188 In practical terms, add/diverse path deployments are expected to 189 result in the distribution of 2, 3 or n (where n is a small number) 190 'good' paths rather than all domain external paths. While the route 191 reflector chooses one set of n paths and distributes those same n 192 paths to all its route reflector clients, those n paths may not be 193 the right n paths for all clients. In the context of the problem 194 described above, those n paths will not necessarily include the 195 closest egress point out of the network for each route reflector 196 client. The mechanisms proposed in this document are likely to be 197 complementary to mechanisms aimed at improving path diversity. 199 2. Proposed solutions 201 This document proposes two simple solutions to the problem described 202 above. Both of these solutions make it possible for route reflector 203 clients to direct traffic to their closest exit point in hot potato 204 routing deployments, without requiring further state to be pushed out 205 to the edge. These solutions are primarily applicable in deployments 206 using centralized route reflectors, which are typically implemented 207 in devices without a capable forwarding plane. 209 The two alternatives are: 211 "Best path selection for BGP hot potato routing from client's IGP 212 network position" 214 "Angular distance approximation for BGP warm potato routing" 216 Both solutions rely upon all route reflectors learning all paths 217 which are eligible for consideration for hot potato routing. In 218 order to satisfy this requirement, path diversity enhancing 219 mechanisms such as add paths/diverse paths may need to be deployed 220 between route reflectors. 222 In both of these solutions the route reflector selects and 223 distributes a route to each client based on what would be optimal 224 from the client's perspective. By optimal we refer in this document 225 to the decision made during best path selection at the IGP metric to 226 BGP next hop comparison step. Clearly the overall path selection 227 preference may be chosen based other policy step and provisions as 228 defined in this document would not apply. 230 In the respective solutions the choice is made either factoring in 231 IGP costs or the configured angular distance to the next hop. The 232 route reflector makes different decisions for different clients only 233 in the case where the tie breaker for path selection would have been 234 the IGP distance to the BGP nexthop (as in hot potato routing). 236 A significant advantage of this approach is that the RR clients do 237 not need to run new software or hardware. 239 Besides these solutions to manage hot potato routing, there are 240 deployment scenarios where service providers want to have more 241 control of traffic exiting the AS by assigning per client preference 242 to gateways. 244 This document proposes to introduce a solution to perform a policy 245 based route-reflection to address those scenarios. This solution has 246 the same requirements (regarding path diversity) and advantages than 247 the two IGP metric based solutions. 249 3. Best path selection for BGP hot potato routing from customized IGP 250 network position 252 This section describes a method for calculating the order of 253 preference of BGP paths from the point of view of each separate route 254 reflector client. More specifically, the route reflector will 255 compute the IGP metric to the BGP nexthop from the position of the 256 client to which the resulting path will be distributed, if the IGP 257 metric is the tie breaker applied to a set of possible paths. In the 258 subsequent model authors will propose virtual reflector placement at 259 operator's selected IGP location. 261 In the case of a hierarchical IGP deployment where the client is in a 262 different level in the hierarchy to the route reflector, the route 263 reflector will compute IGP distance to the BGP nexthop from the Area 264 Border Routers (ABR) leading to the client in lieu of the route 265 reflector client itself, and use the shortest distance from these 266 ABRs to the nexthop. This provides an approximation to the desired 267 functionality. Rather than a client picking the closest path, the 268 client would be picking the exit point closest to the client region 269 as defined by area or level. In cases where one or more nexthops are 270 in the same region as the client, one of those nexthops would be 271 preferred, with tie breaking within those nexthops performed from the 272 route reflector's position in the network. 274 It is assumed that reachability through a set of ABRs is always 275 advertised through identical prefixes from those ABRs. If a nexthop 276 is reachable through multiple ABRs but the ABRs advertise 277 reachability through prefixes of different length, then only the ABR 278 advertising the longest prefix will be considered as a viable path to 279 the nexthop. 281 BGP best path selection and its distribution has a natural 282 consequence of limiting the amount of state in the network. That is 283 not in itself a drawback. BGP speakers will rarely need to receive 284 all available BGP paths. In network deployments with multiple 285 upstream peerings or with very dense peering schemes, the number of 286 available BGP paths for a given BGP prefix can be high. Real network 287 deployments with the number of paths for a prefix ranging from 10s to 288 100s have been observed. It would be wasteful to propagate all of 289 those paths to all clients, such that each client can select paths 290 according to the position of the nexthop relative to the client. 292 Whenever a BGP route reflector would need to decide what path or 293 paths need to be selected for advertisement to one of its clients, 294 the route reflector would need to virtually position itself in its 295 client IGP network location in order to choose the right set of paths 296 based on the IGP metric to the next hops from the client's 297 perspective. 299 This technique applies in deployments with or without diverse paths 300 or the various path selection modes contemplated in add-paths. 302 In the network architectures consisting of more then single pair of 303 route reflectors it is required that all reflectors are fully meshed 304 and have ability to learn and maintain all external BGP paths. In 305 the event of constructing a hierarchy of reflectors to relax the full 306 RR mesh requirements ORR should not be run between such route 307 reflectors. 309 3.1. Client's perspective best path selection algorithm 311 For each centralized route reflector the proposal assumes that the 312 route reflector participates in a common IGP with its clients. There 313 are two scenarios to consider - flat versus hierarchical IGP network. 315 3.1.1. Flat IGP network 317 Reflectors run SPF from the client IGP node point of view such 318 that the cost of BGP nexthops from the client can be determined if 319 necessary. For the purpose of BGP path selection the interesting 320 product of this calculation is the ability to determine the IGP 321 distance from a client to a BGP next hop. This distance to a 322 nexthop would be interesting in cases where that next hop is for a 323 path which is contending with otherwise equally preferred paths. 324 This approach works in tunneled as well as conventional hop-by-hop 325 IP forwarding cores. 327 When the path selection tie breaker for a prefix is the IGP metric 328 to the BGP nexthops of the contending paths, then the route 329 reflector will determine the order of preference of the contending 330 paths by considering the distance from the client to the path 331 nexthops in order to decide what path/s to advertise to a client 332 (or group of clients where feasible). It should be noted that an 333 operator may wish to provide a distance tolerance value, such that 334 beyond a certain granularity, differences between IGP metric are 335 invisible to the path selection algorithm. This will allow a 336 route reflector some leeway in selecting between paths such that 337 rather than pick one path over another on the basis of a 338 difference in distance which is operationally irrelevant, the 339 route reflector can choose to optimize for update generation 340 grouping. Furthermore, this tolerance will reduce the likelihood 341 of generation of BGP updates when the IGP topology changes in a 342 way which is not operationally relevant. In the case that a path 343 is selected from a set for a given prefix while ignoring 344 differences in distance within the tolerance figure, then that 345 same path must always be preferred for all clients where the paths 346 are within the tolerance figure 348 3.1.2. Hierarchical IGP network 350 Hierarchy introduces two challenges: 352 The first challenge is that the RR IGP view may differ from a 353 client IGP view by virtue of one or the other having a summarized 354 view versus the other. Summarization, by its nature, loses 355 information. Consider the example where a client within a PoP 356 sees two prefixes with two metrics for two egress points within 357 the PoP, but where the RR only sees a single summary covering 358 reachability to both nexthops as injected by the ABR. For 359 clarification purposes in the case of ISIS by ABR we refer to L1/ 360 L2 node. However it needs to be observed that inter area networks 361 running LDP are required to disable summarisation of all FEC 362 advertised in LDP (typically all loopbacks) unless [RFC5283] is 363 deployed. Such deployments are not likely to suffer summarization 364 difficulties. 366 The second challenge is that in cases where the client is in a 367 different level of hierarchy from the RR, the RR can not build a 368 Shortest Path First (SPF) tree with the client node as root, 369 simply because the topology derived by the IGP will not include 370 the client node. It will instead only include reachability to the 371 client from one or more ABRs. In order to overcome this problem, 372 the RR could compute an SPF tree from the ABRs in the area. The 373 RR would then determine the shortest distance from a client which 374 lives behind the ABRs, to a nexthop, by adding the advertised 375 distances from an ABR to the client and the distance from the ABR 376 to a nexthop, for each ABR, and picking the minimum. This assumes 377 that IGP metrics on links are symmetric; i.e. that the distance 378 from the ABR to the client or nexthop is equal to the distance 379 from the client or nexthop to the ABR. 381 There are cases where the above approach does not help. If RR is 382 trying to arbitrate amongst a set of paths for a client which is 383 in the same hierarchy as some of those paths, and in a different 384 hierarchy to the RR, the opaqueness of the region containing the 385 client at the RR defeats the selection process. It is impossible 386 to determine the relative position of the RR client and the paths 387 within the client region. 389 The solution for hierarchical IGP networks also assumes that if 390 RRs are present and are responsible for calculation of BGP best 391 path to clients they are either placed in each local area 392 coinciding with area containing clients or they are placed in the 393 core (area 0/level 2) of the network. 395 3.2. Aside: Configuration-based flexible route reflector placement 397 The ability to exploit topology information available in the IGP in 398 ways described above can also be used to virtually place the RR at 399 different points in the network for purposes other than hot potato 400 routing. 402 A route reflector can be globally configured to "pretend" its logical 403 location is one of any of the other nodes within a given IGP area/ 404 level flooding scope regardless of its physical connectivity. 406 Such flexibility provides a useful tool for reflector virtualization, 407 and supports moving or replacing physical route reflectors without 408 any effect on routing. Such a change can be permanent or it could be 409 performed during network maintenance in order to minimize network 410 impact. 412 A possible variation would allow the virtual placement of RR to be 413 effected on a per-AF or AF plus update/peer group granularity. It 414 should be noted that this approach provides for splitting one 415 centralized route reflector such that it is virtually positioned at 416 various network locations, with the network location depending upon 417 of address family or address family plus update/peer group. 419 Virtual slicing of a centralized route reflector relaxes the need to 420 propagate all BGP paths between RRs in a alternative conventional 421 distributed RR deployment. It is expected that such RRs would be 422 deployed in redundant sets, and that those RRs would not need to be 423 physically collocated, while still benefiting from the possibility of 424 being logically collocated, and therefore not compromising any of the 425 best path selection symmetry. 427 3.3. Route reflector client grouping 429 It may be appropriate to allow the operator, or the route reflector 430 itself, to group clients together using IGP distance between clients 431 to determine grouping. All the operation discussed above which 432 relied upon computing best path for each client, and measuring 433 distances from each client to different nexthops, would instead be 434 performed for each group of clients. Configurable thresholds can be 435 used to determine which IGP metric changes should be visible to BGP, 436 and trigger best paths recomputation. The latter would be beneficial 437 in existing BGP RR code too. 439 Alternatively route reflector client grouping could be accomplished 440 statically by the operator by coloring clients belonging to a common 441 group (for example being part of the same POP). In order to 442 accomplish such marking it is proposed that BGP OPEN message be 443 augmented with an optional parameter indicating the Group ID given 444 peer belongs to. 446 3.3.1. Route Reflector Client Group ID 448 This is an Optional Parameter in BGP OPEN message that is used by a 449 BGP speaker to convey to its route reflectors the Group ID value. 450 Such value will allow automatic and predictable peer grouping on the 451 route reflectors as deemed necessary from operator's network 452 architecture. 454 The parameter contains precisely one set of [Group_ID Code, Group_ID 455 Length, Group_ID Value] encoded as shown below: 457 +----------------------------+ 458 | Group ID Code (1 octet) | 459 +----------------------------+ 460 | Group ID Length (1 octet) | 461 +----------------------------+ 462 | Group ID Value (4 octets) | 463 +----------------------------+ 465 The use and meaning of these fields are as follows: 467 Group ID Code: 469 Group ID Code is a one octet field that identifies Group ID 470 optional parameter of BGP OPEN message. Value TBD by IANA 471 Recommended value: 3. 473 Group ID Length: 475 Group ID Length is a one octet field that contains the length 476 of the Group ID Value field in octets. It is fixed and equals 477 to 4. 479 Group ID Value: 481 Group ID Value is a fixed length field of size equal to 482 four octets that contains the numerical value of group given 483 BGP speaker should be part of on the route reflector. 485 Two special values are reserved: 487 0x00000000 - No grouping preference 488 0xFFFFFFFF - Do not group this BGP speaker 490 An implementation may allow automatic population of 491 GROUP_ID value using IGP area identifier. 493 Route reflectors or EBGP speakers receiving such Group IDs from their 494 respective BGP peers as part of the BGP OPEN procedure MAY use them 495 when constructing update or peer groups in addition to any of the 496 existing grouping mechanism already available. An implementation may 497 allow operator to explicitly allow or disallow honoring such grouping 498 or provide means for manual overwrite via explicit configuration. 500 3.4. Discussion 502 This is not the first instance where a router participating in an IGP 503 is required to build the SPF tree using a root other than itself. 504 Determination of loop free alternate paths as described in [RFC5714] 505 is one such example. 507 Determining the shortest path and associated cost between any two 508 arbitrary points in a network based on the IGP topology learned by a 509 router is expected to add some extra cost in terms of CPU resource. 510 However SPF tree generation code is now implemented efficiently in a 511 number of implementations, and therefor this is not expected to be a 512 major drawback. The number of SPTs computed in the general non- 513 hierarchical case is expected to be of the order of the number of 514 clients of an RR whenever a topology change is detected. Advanced 515 optimizations like partial and incremental SPF may also be exploited. 516 By the nature of route reflection, the number of clients can be split 517 arbitrarily by the deployment of more route reflectors for a given 518 number of clients. While this is not expected to be necessary in 519 existing networks with best in class route reflectors available 520 today, this avenue to scaling up the route reflection infrastructure 521 would be available. If we consider the overall network wide cost/ 522 benefit factor, the only alternative to achieve the same level of 523 optimality would require significantly increasing state on the edges 524 of the network, which, in turn, will consume CPU and memory resources 525 on all BGP speakers in the network. Building this client perspective 526 into the route reflectors seems appropriate. 528 3.5. Advantages 530 The solution described provides a model for integrating the client 531 perspective into the best path computation for RRs. More 532 specifically, the choice or BGP path factors in the IGP metric 533 between the client and the nexthop, rather than the distance from the 534 RR to the nexthop. The documented method does not require any BGP or 535 IGP protocol changes as required changes are contained within the RR 536 implementation. 538 This solution can be deployed in traditional hop-by-hop forwarding 539 networks as well as in end-to-end tunneled environments. In the 540 networks where there are multiple route reflectors and hop-by-hop 541 forwarding without encapsulation, such optimizations should be 542 enabled on all route reflectors. Otherwise clients may receive an 543 inconsistent view of the network and in turn lead to intra-domain 544 forwarding loops. 546 With this approach, an ISP can effect a hot potato routing policy 547 even if route reflection has been moved from the forwarding plane to 548 the core and hop-by-hop switching has been replaced by end to end 549 MPLS or IP encapsulation. 551 As per above, the approach reduces the amount of state which needs to 552 be pushed to the edge in order to perform hot potato routing. The 553 memory and CPU resource required at the edge to provide hot potato 554 routing using this approach is lower than what would be required in 555 order to achieve the same level of optimality by pushing and 556 retaining all available paths (potentially 10s) per each prefix at 557 the edge. 559 The proposal allows for a fast and safe transition to BGP control 560 plane route reflection without compromising an operator's closest 561 exit operational principle. Hot potato routing is important to most 562 ISPs. The inability to perform hot potato routing effectively stops 563 migrations to centralized route reflection and edge-to-edge LSP/IP 564 encapsulation for traffic to IPv4 and IPv6 prefixes. 566 4. Angular distance approximation for BGP warm potato routing 568 This section describes an alternative solution to the use of IGP 569 topology information to virtually position the RR at the client 570 location in the network. This solution involves modeling the network 571 topology as a set of elements (regions, PoPs or routers) arranged in 572 a circle. Route reflector clients and inter-domain exit points would 573 then be statically assigned to those elements such that one can 574 compute the angular distance between route-reflector clients and the 575 various exit points in order to infer the distance between any two 576 elements. This measure of distance can be used as an effective 577 alternative to the IGP distance as a tie breaker in the path 578 selection algorithm if necessary. 580 4.1. Problem statement 582 This solution addresses the problem described in earlier sections, 583 while attempting to minimize computational overhead. The aim of the 584 proposed solution is to enable a route reflector to provide a route 585 reflector client with an exit point for a prefix which is 'closest' 586 to the client rather than the route-reflector, without having to 587 distribute all paths to that client, or having to derive each 588 client's view of the network topology. The measure of closest is 589 based on a simplistic description of network topology provided by the 590 operator. 592 Consider the following example of an ISP network topology drawn to 593 reflect the location of the nodes and POPs: 595 N4 POP4 597 CLIENT B 598 POP4 POP1 N1 600 CORE 601 RR(s) POP2 N2 603 N5 POP3 POP2 N3 605 CLIENT A 606 POP3 608 N - represents the different exit points for a given prefix. POP2 is 609 a geographically large PoP with two paths; N2 and N3. 611 In a deployment where the centralized RRs tie break on the basis of 612 their IGP-based view of the network, N1 above would be advertised to 613 all clients on the basis that it is closest to the RR. Path N4 would 614 be a more appropriate choice for client B. Similarly, N5 would be 615 more appropriate for client A since path N5 is closer to client A 616 then path N1. 618 4.2. Proposed solution 620 The proposed solution revolves around the operator establishing the 621 angular position of the route-reflector clients and inter-domain exit 622 points in the network. The route reflector then picks the path to 623 advertise to a client based on the client's angular position versus 624 the angular position of the inter-domain exit points originating the 625 paths. The operator can choose the granularity of angular position 626 appropriate to the desired goals. On one hand, the coarseness of the 627 angular position will effect the operator overhead; versus the 628 optimality of routing on the other. The finest granularity possible 629 will be the relative position of originating clients. 631 Note that this solution has nothing to do with actual IGP link 632 metrics and resulting topology in the network. 634 It can be shown that for each network topology, elements such as AS 635 exit points can be mapped on to a circle. By putting POPs, Regions 636 or individual clients onto the hypothetical circle we can identify an 637 angular location for each element relative to some fixed direction; 638 for example defining the angular north of the circle at 0 degrees. 640 The angular position of elements in the network can be conveyed to a 641 route reflector in a number of ways: 643 Assignment of angular position of each RR client through 644 configuration on the route reflector itself; per client 645 configuration on RR 647 Assignment of angular position of an RR client at each client, 648 then propagating it to RRs. 650 The proposed angular distance approximation is compatible with both 651 flat and hierarchical IGP deployments. 653 In the example illustrated above the route reflector might learn or 654 be configured with the following set of paths and corresponding 655 angular positions: 657 Prefix X/Y N1 N2 N3 N4 N5 659 Location 660 in degrees 60 85 120 290 260 662 If the absolute angular position of clients A and B were as follows: 664 Client A: 260 degrees 666 Client B: 290 degrees 668 Then the corresponding angular distances for those clients versus the 669 exit points can be calculated as follows: 671 Prefix X/Y N1 N2 N3 N4 N5 673 Client A 200 175 140 30 0 675 Client B 230 205 170 0 30 677 With an RR running the BGP best path algorithm modified to use the 678 angular distance from the client to the nexthops, rather than its IGP 679 distance to the nexthops as tie breaker, each client is provided with 680 its closest path with the measure of closeness reflecting the angular 681 position as configured by the operator. 683 The model used by the operator in order to determine the angular 684 position of a client or exit point, might involve grouping elements 685 together by region or PoP, or might involve no grouping at all. 686 Implementations should allow the operator to pick the appropriate 687 granularity. 689 4.3. Centralized vs distributed route reflectors 691 In an environment where the RR clusters are distributed (yet 692 centralized enough to make hot potato routing hard), and each RR 693 cluster serves a subset of clients, it becomes necessary to propagate 694 the angular position of the clients between route reflectors. This 695 can be achieved as follows: 697 Deploy add-paths between route reflectors in order to maximize 698 path diversity within the cluster. 700 A non AS transitive BGP community of type (TBA by IANA) can be 701 used to encode and propagate angular position between 0 and 359 of 702 a client. This community is only relevant to the route reflectors 703 of a given BGP domain and should be stripped either at the ASBR 704 boundary or when propagating updates to BGP peers which are not 705 route reflectors. 707 The angular position marking could also be added by clients and 708 advertised to the route reflector. This would require some 709 configuration effort. 711 5. Client's perspective policy based best path selection 713 There is some deployment scenarios where a service provider wants to 714 achieve a stronger control on traffic exiting the AS (for capacity 715 planning) rather than using hot potato routing based on IGP metric. 717 | | | | 718 | | | | 719 GW1 GW2 GW3 GW4 721 RR1 RR2 723 R1 R2 R3 725 Considering the figure above, all gateways have iBGP sessions to RR1 726 and RR2, and R1 R2 R3 have iBGP sessions as well to RR1 and RR2. 727 Gateway routers are meshed to an external network (for example, a 728 transit service provider). 730 We would like to achieve a strong control on the gateway used 731 (primary and backup) for each router (or each set of routers) in the 732 network (taking into account that routers do not support ADD PATHs). 733 For example, R1 using GW1 as primary and GW2 as backup; R2 using GW2 734 as primary and GW3 as backup; R3 using GW3 as primary and GW4 as 735 backup. 737 Basically, today a prefix P1 is received on each gateway from the 738 external network. Each gateway will send the prefix to both route 739 reflectors. Each route-reflector will receive four paths for P1 and 740 choose the best one based on his own decision process. Note that RR1 741 and RR2 may choose a different path as best. Each route-reflector 742 sends his best path towards R1, R2 and R3. Each router will receive 743 the same paths from the route-reflectors for P1 (at max, only two 744 gateways are visible from Rx routers). So default behavior does not 745 fit our requirements in term of traffic flows. 747 Using current BGP mechanisms available, we could achieve our 748 requirements using two solutions : 750 o Modify the BGP meshing: for example, R1 meshed directly to GW1 and 751 GW2 and apply inbound policies on R1; R2 meshed directly to GW2 752 and GW3 and apply inbound policies on R2 ... 754 o Adding more route-reflectors (one RR per gateway used as primary) 755 and applying inbound policies on RRs to make each RR choosing a 756 different primary gateway and apply policies on routers to select 757 his own primary gateway. 759 These solutions have many drawbacks: first one is not flexible (re- 760 meshing needed when we want to change gateway of a router), second 761 one requires a lot of CAPEX. 763 We would like to introduce a solution where a single currently 764 deployed route-reflector chassis may take a different best path 765 decision for different set of clients based on preferences. 767 It should be noted that in simple scenarios (example: two RRs and two 768 gateways), RFC6774 would be able to fulfill service provider needs. 769 The solution proposed here would permit to handle more complex 770 scenarios and fine gateway choice per client or groups of clients. 772 5.1. Proposal 774 Our proposal is to reuse the concept introduced in [I.D.ietf-idr-ix- 775 bgp-route-server] in an iBGP context. To perform per client best 776 path selection, the router should maintain a per client BGP local-RIB 777 (or Adj-RIB-Out) associated with inbound policies implemented between 778 Adj-RIB-In and client LOC-RIB. 780 It would not be very scalable to use a per client policy (considering 781 hundreds of peers on a route-reflector), therefor our proposal is to 782 group clients sharing common policies inside a client group to 783 minimize computation/memory overhead. Client grouping could be done 784 statically (by configuration) or dynamically using the solution 785 described in section 3.3.1 of this document. Client grouping would 786 be performed with a per AFI/SAFI granularity as gateway/client 787 mapping may change in each AFI/SAFI context. A route-reflector 788 should be able to implement multiple client groups (with associated 789 inbound policies) as well as a default client group for clients that 790 does not require any specific policy decision: in this case, the 791 overall BGP best path computation would be used. 793 5.2. Example 794 GW1 GW2 GW3 795 \ | / 796 \ | / 797 RR1 798 / | \ 799 R1 R2 R3 801 In the above figure GW1, GW2, GW3 and R3 are standard ibgp route- 802 reflector clients. R1 and R2 want to use a special gateway 803 combination (primary GW3, backup GW2, last resort GW1). R1 and R2 804 are configured in a specific client group CG1 on the route-reflector 805 while other peers are in the default client group. CG1 is associated 806 with a policy achieving the expected GW preference for R1 and R2, and 807 letting other paths without any change. 809 All routes received by RR1 (ebgp, ibgp, ibgp rr client, ibgp rr 810 client routing context) must be evaluated using overall BGP best path 811 computation as well as in client group, the client group policy will 812 accept or not the route to be evaluated by the local decision 813 process. 815 o Paths from GW1, GW2, GW3 are compared within default client group 816 leading to one GW (for example GW1) to be selected as best and 817 installed in global LOC-RIB. GW1 path will be advertised to GW2, 818 GW3 and R3 as they are in default CG. In CG1, preference of GW 819 paths has been modified, leading to GW3 being the best path and 820 installed in client group LOC-RIB. GW3 path will be advertised to 821 R1 and R2, as R1 and R2 are part of CG1. 823 o Paths from R3 are compared within default client group and 824 advertised to GW1, GW2, GW3. Those paths are also compared within 825 CG1 (as accepted by policy) and advertised to R1 and R2. 827 o Paths from R1 are compared within default client group and 828 advertised to GW1, GW2, GW3 and R3. Those paths are also compared 829 within GG1 (as accepted by policy) and advertised to R2. 831 o Paths from R2 are compared within default client group and 832 advertised to GW1, GW2, GW3 and R3. Those paths are also compared 833 within CG1 (as accepted by policy) and advertised to R1. 835 5.3. Avoiding routing loops 837 Compared to the IGP approaches described in this document, the policy 838 based route-reflection should be limited to end-to-end encapsulation 839 environments to avoid intra-domain forwarding loops. Using end-to- 840 end encapsulation permit Edge routers to transport the traffic to the 841 targeted/preferred ASBR without any loop in the core. 843 To avoid a potential rerouting of the ASBR into the core (and 844 possible loop between Edges and ASBR), we must enforce forwarding at 845 the ASBR to the eBGP peer. This could be done by : 847 o implementing policies on ASBR to prefer eBGP path and install it 848 in FIB. 850 o implementing tunneling of traffic until the outside interface 851 (ASBR action to switch to outside interface). 853 The exact choice of encapsulation and techniques to prevent transport 854 loops (including potential loops at gateways) is left to the operator 855 choice and its specification is outside of the scope of this 856 document. 858 6. Deployment considerations 860 The solutions are primarily intended for end-to-end tunneled 861 environments, i.e. where traffic is label switched or IP tunneled 862 across the core. If unencapsulated hop-by-hop forwarding is used, 863 either misconfigurations or conflicts between these optimizations and 864 classical BGP path selection rules could lead to intra-domain 865 forwarding loops. Under certain circumstances the solutions can also 866 be deployable without end-to-end tunneling. In particular the best 867 path selection based on the client's IGP best-path selection is 868 guaranteed not to cause any forwarding loops (other than micro loops 869 associated with reconvergence) when deployed in a flat IGP area 870 provided that no distance tolerance value is used so that the path 871 choice is truly made on a per-client basis. 873 Regarding potential intra-domain forwarding loops at ASBR level, this 874 could be solved by enforcing external route preference or by 875 performing tunnel to external interface switching action on ASBRs. 877 Regarding client's IGP best-path selection, it should be self evident 878 that this solution does not interfere with policies enforced above 879 IGP tie breaking in the BGP best path algorithm. 881 The solution applies to NLRIs of all address families which can be 882 route reflected. 884 It should be noted that customized per-client or group of clients 885 best path selection is already in use today in the context of 886 Internet Exchange Point (IXP) route servers. In an IXP route server 887 the client best path is selected as a result of different policies 888 rather than IGP metric distance to BGP next hop. 890 A possible scalability impact of optimizing path selection to take 891 account of the RR client position or operator's policy based 892 preference is that different RR clients receive different paths, and 893 therefore update/peer group efficiency diminishes. This cost is 894 imposed by the requirement to optimize the egress path from the 895 client's perspective. It is also likely that groups of clients will 896 end up receiving the same best path/s, in which case, inefficiency of 897 update generation will be minimized. It should be noted that in the 898 cases described under flexible router placement where placement is 899 determined on a per update/peer group basis or per route reflector, 900 the scale benefits of peer groupings are retained. 902 7. Security considerations 904 No new security issues are introduced to the BGP protocol by this 905 specification. 907 8. IANA Considerations 909 IANA is requested to allocate a type code for the Standard BGP 910 Community to be used for inter cluster propagation of angular 911 position of the clients. 913 IANA is requested to allocate a new type code from BGP OPEN Optional 914 Parameter Types registry to be used for Group_ID propagation. 916 9. Acknowledgments 918 Authors would like to thank Eric Rosen, Clarence Filsfils, Uli 919 Bornhauser Russ White, Jakob Heitz, Mike Shand and Jon Mitchell for 920 their valuable input. 922 10. References 924 10.1. Normative References 926 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 927 Requirement Levels", BCP 14, RFC 2119, March 1997. 929 [RFC4271] Rekhter, Y., Li, T., and S. Hares, "A Border Gateway 930 Protocol 4 (BGP-4)", RFC 4271, January 2006. 932 [RFC4360] Sangli, S., Tappan, D., and Y. Rekhter, "BGP Extended 933 Communities Attribute", RFC 4360, February 2006. 935 [RFC5492] Scudder, J. and R. Chandra, "Capabilities Advertisement 936 with BGP-4", RFC 5492, February 2009. 938 10.2. Informative References 940 [I-D.ietf-idr-add-paths] 941 Walton, D., Retana, A., Chen, E., and J. Scudder, 942 "Advertisement of Multiple Paths in BGP", draft-ietf-idr- 943 add-paths-09 (work in progress), October 2013. 945 [RFC1997] Chandrasekeran, R., Traina, P., and T. Li, "BGP 946 Communities Attribute", RFC 1997, August 1996. 948 [RFC1998] Chen, E. and T. Bates, "An Application of the BGP 949 Community Attribute in Multi-home Routing", RFC 1998, 950 August 1996. 952 [RFC4384] Meyer, D., "BGP Communities for Data Collection", BCP 114, 953 RFC 4384, February 2006. 955 [RFC4456] Bates, T., Chen, E., and R. Chandra, "BGP Route 956 Reflection: An Alternative to Full Mesh Internal BGP 957 (IBGP)", RFC 4456, April 2006. 959 [RFC4893] Vohra, Q. and E. Chen, "BGP Support for Four-octet AS 960 Number Space", RFC 4893, May 2007. 962 [RFC5283] Decraene, B., Le Roux, JL., and I. Minei, "LDP Extension 963 for Inter-Area Label Switched Paths (LSPs)", RFC 5283, 964 July 2008. 966 [RFC5668] Rekhter, Y., Sangli, S., and D. Tappan, "4-Octet AS 967 Specific BGP Extended Community", RFC 5668, October 2009. 969 [RFC5714] Shand, M. and S. Bryant, "IP Fast Reroute Framework", RFC 970 5714, January 2010. 972 [RFC6774] Raszuk, R., Fernando, R., Patel, K., McPherson, D., and K. 973 Kumaki, "Distribution of Diverse BGP Paths", RFC 6774, 974 November 2012. 976 Authors' Addresses 978 Robert Raszuk 979 Individual 981 Email: robert@raszuk.net 982 Christian Cassar 983 Cisco Systems 984 10 New Square Park 985 Bedfont Lakes, FELTHAM TW14 8HA 986 UK 988 Email: ccassar@cisco.com 990 Erik Aman 991 TeliaSonera 992 Marbackagatan 11 993 Farsta SE-123 86 994 Sweden 996 Email: erik.aman@teliasonera.com 998 Bruno Decraene 999 Orange 1000 38-40 rue du General Leclerc 1001 Issy les Moulineaux cedex 9 92794 1002 France 1004 Email: bruno.decraene@orange.com 1006 Stephane Litkowski 1007 Orange 1008 9 rue du chene germain 1009 Cesson Sevigne 35512 1010 France 1012 Email: stephane.litkowski@orange.com