idnits 2.17.1 draft-jakab-lisp-deployment-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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 470 has weird spacing: '...ructure x ...' -- The document date (February 21, 2011) is 4812 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-24) exists of draft-ietf-lisp-09 == Outdated reference: A later version (-10) exists of draft-ietf-lisp-alt-05 == Outdated reference: A later version (-06) exists of draft-ietf-lisp-interworking-01 == Outdated reference: A later version (-16) exists of draft-ietf-lisp-ms-06 == Outdated reference: A later version (-09) exists of draft-lear-lisp-nerd-08 == Outdated reference: A later version (-02) exists of draft-meyer-lisp-eid-block-01 Summary: 0 errors (**), 0 flaws (~~), 8 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group L. Jakab 3 Internet-Draft A. Cabellos-Aparicio 4 Intended status: Informational F. Coras 5 Expires: August 25, 2011 J. Domingo-Pascual 6 Technical University of Catalonia 7 D. Lewis 8 Cisco Systems 9 February 21, 2011 11 LISP Network Element Deployment Considerations 12 draft-jakab-lisp-deployment-02.txt 14 Abstract 16 This document discusses the different scenarios in which the LISP 17 protocol may be deployed. Changes or extensions to other protocols 18 needed by some of the scenarios are also highlighted. 20 Status of this Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on August 25, 2011. 37 Copyright Notice 39 Copyright (c) 2011 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 55 2. Tunnel Routers . . . . . . . . . . . . . . . . . . . . . . . . 4 56 2.1. Customer Edge . . . . . . . . . . . . . . . . . . . . . . 4 57 2.2. Provider Edge . . . . . . . . . . . . . . . . . . . . . . 5 58 2.3. Split ITR/ETR . . . . . . . . . . . . . . . . . . . . . . 6 59 2.4. Inter-Service Provider Traffic Engineering . . . . . . . . 8 60 2.5. Tunnel Routers Behind NAT . . . . . . . . . . . . . . . . 10 61 2.5.1. ITR . . . . . . . . . . . . . . . . . . . . . . . . . 10 62 2.5.2. ETR . . . . . . . . . . . . . . . . . . . . . . . . . 10 63 2.6. Summary and Feature Matrix . . . . . . . . . . . . . . . . 11 64 3. Map-Resolvers and Map-Servers . . . . . . . . . . . . . . . . 11 65 3.1. Map-Servers . . . . . . . . . . . . . . . . . . . . . . . 11 66 3.2. Map-Resolvers . . . . . . . . . . . . . . . . . . . . . . 12 67 4. Proxy Tunnel Routers . . . . . . . . . . . . . . . . . . . . . 13 68 4.1. P-ITR . . . . . . . . . . . . . . . . . . . . . . . . . . 13 69 4.1.1. LISP+BGP . . . . . . . . . . . . . . . . . . . . . . . 14 70 4.1.2. Mapping Service Provider P-ITR Service . . . . . . . . 15 71 4.1.3. Tier 1 P-ITR Service . . . . . . . . . . . . . . . . . 15 72 4.1.4. Migration Summary . . . . . . . . . . . . . . . . . . 16 73 4.1.5. Content Provider Load Balancing . . . . . . . . . . . 17 74 4.2. P-ETR . . . . . . . . . . . . . . . . . . . . . . . . . . 17 75 5. Security Considerations . . . . . . . . . . . . . . . . . . . 18 76 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 77 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 19 78 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 79 8.1. Normative References . . . . . . . . . . . . . . . . . . . 19 80 8.2. Informative References . . . . . . . . . . . . . . . . . . 19 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 20 83 1. Introduction 85 The Locator/Identifier Separation Protocol (LISP) addresses the 86 scaling issues of the global Internet routing system by separating 87 the current addressing scheme into Endpoint IDentifiers (EIDs) and 88 Routing LOCators (RLOCs). The main protocol specification 89 [I-D.ietf-lisp] describes how the separation is achieved, which new 90 network elements are introduced, and details the packet formats for 91 the data and control planes. 93 While the boundary between the core and edge is not strictly defined, 94 one widely accepted definition places it at the border routers of 95 stub autonomous systems, which may carry a partial or complete 96 default-free zone (DFZ) routing table. The initial design of LISP 97 took this location as a baseline for protocol development. However, 98 the applications of LISP go beyond of just decreasing the size of the 99 DFZ routing table, and include improved multihoming and ingress 100 traffic engineering (TE) support for edge networks, and even 101 individual hosts. Throughout the draft we will use the term LISP 102 site to refer to these networks/hosts behind a LISP Tunnel Router. 103 We formally define it as: 105 LISP site: A single host or a set of network elements in an edge 106 network under the administrative control of a single organization, 107 delimited from other networks by LISP Tunnel Router(s). 109 Since LISP is a protocol which can be used for different purposes, it 110 is important to identify possible deployment scenarios and the 111 additional requierements they may impose on the protocol 112 specification and other protocols. The main specification 113 [I-D.ietf-lisp] mentions positioning of tunnel routers, but without 114 an in-depth discussion. This document fills that gap, by exploring 115 the most common cases. While the theoretical combinations of device 116 placements are quite numerous, the more practical scenarios are given 117 preference in the following. 119 Additionally, this documents is intended as a guide for the 120 operational community for LISP deployments in their networks. It is 121 expected to evolve as LISP deployment progresses, and the described 122 scenarios are better understood or new scenarios are discovered. 124 Each subsection considers an element type, discussing the impact of 125 deployment scenarios on the protocol specification. For definition 126 of terms, please refer to the appropriate documents (as cited in the 127 respective sections). 129 Comments and discussions about this memo should be directed to the 130 LISP working group mailing list: lisp@ietf.org. 132 2. Tunnel Routers 134 LISP is a map-and-encap protocol, with the main goal of improving 135 global routing scalability. To achieve its goal, it introduces 136 several new network elements, each performing specific functions 137 necessary to separate the edge from the core. The device that is the 138 gateway between the edge and the core is called Tunnel Router (xTR), 139 performing one or both of two separate functions: 141 1. Encapsulating packets originating from an end host to be 142 transported over intermediary (transit) networks towards the 143 other end-point of the communication 145 2. Decapsulating packets entering from intermediary (transit) 146 networks, originated at a remote end host. 148 The first function is performed by an Ingress Tunnel Router (ITR), 149 the second by an Egress Tunnel Router (ETR). 151 Section 8 of the main LISP specification [I-D.ietf-lisp] has a short 152 discussion of where Tunnel Routers can be deployed and some of the 153 associated advantages and disadvantages. This section adds more 154 detail to the scenarios presented there, and provides additional 155 scenarios as well. 157 2.1. Customer Edge 159 LISP was designed with deployment at the core-edge boundary in mind, 160 which can be approximated as the set of DFZ routers belonging to non- 161 transit ASes. For the purposes of this document, we will consider 162 this boundary to be consisting of the routers connecting LISP sites 163 to their upstreams. As such, this is the most common expected 164 scenario for xTRs, and this document considers it the reference 165 location, comparing the other scenarios to this one. 167 ISP1 ISP2 168 | | 169 | | 170 +----+ +----+ 171 +--|xTR1|--|xTR2|--+ 172 | +----+ +----+ | 173 | | 174 | Customer | 175 +------------------+ 177 Figure 1: xTRs at the customer edge 179 From the LISP site perspective the main advantage of this type of 180 deployment (compared to the one described in the next section) is 181 having direct control over its ingress traffic engineering. This 182 makes it is easy to set up and maintain active/active, active/backup, 183 or more complex TE policies, without involving third parties. 185 Being under the same administrative control, reachability information 186 of all ETRs is easier to synchronize, because the necessary control 187 traffic can be allowed between the locators of the ETRs. A correct 188 synchronous global view of the reachability status is thus available, 189 and the Loc-Status-Bits can be set correctly in the LISP data header 190 of outgoing packets. 192 By placing the tunnel router at the edge of the site, existing 193 internal network configuration does not need to be modified. 194 Firewall rules, router configurations and address assignments inside 195 the LISP site remain unchanged. This helps with incremental 196 deployment and allows a quick upgrade path to LISP. For larger sites 197 with many external connections, distributed in geographically diverse 198 PoPs, and complex internal topology, it may however make more sense 199 to both encapsulate and decapsulate as soon as possible, to benefit 200 from the information in the IGP to choose the best path (see 201 Section 2.3 for a discussion of this scenario). 203 Another thing to consider when placing tunnel routers are MTU issues. 204 Since encapsulating packets increases overhead, the MTU of the end- 205 to-end path may decrease, when encapsulated packets need to travel 206 over segments having close to minimum MTU. Some transit networks are 207 known to provide larger MTU than the typical value of 1500 bytes of 208 popular access technologies used at end hosts (e.g., IEEE 802.3 and 209 802.11). However, placing the LISP router connecting to such a 210 network at the customer edge could possibly bring up MTU issues, 211 depending on the link type to the provider as opposed to the 212 following scenario. 214 2.2. Provider Edge 216 The other location at the core-edge boundary for deploying LISP 217 routers is at the internet service provider edge. The main incentive 218 for this case is that the customer does not have to upgrade the CE 219 router(s), or change the configuration of any equipment. 220 Encapsulation/decapsulation happens in the provider's network, which 221 may be able to serve several customers with a single device. For 222 large ISPs with many residential/business customers asking for LISP 223 this can lead to important savings, since there is no need to upgrade 224 the software (or hardware, if it's the case) at each client's 225 location. Instead, they can upgrade the software (or hardware) on a 226 few PE routers serving the customers. This scenario is depicted in 227 Figure 2. 229 +----------+ +------------------+ 230 | ISP1 | | ISP2 | 231 | | | | 232 | +----+ | | +----+ +----+ | 233 +--|xTR1|--+ +--|xTR2|--|xTR3|--+ 234 +----+ +----+ +----+ 235 | | | 236 | | | 237 +--<[Customer]>----+-------+ 239 Figure 2: xTR at the PE 241 While this approach can make transition easy for customers and may be 242 cheaper for providers, the LISP site looses one of the main benefits 243 of LISP: ingress traffic engineering. Since the provider controls 244 the ETRs, additional complexity would be needed to allow customers to 245 modify their mapping entries. 247 The problem is aggravated when the LISP site is multihomed. Consider 248 the scenario in Figure 2: whenever a change to TE policies is 249 required, the customer contacts both ISP1 and ISP2 to make the 250 necessary changes on the routers (if they provide this possibility). 251 It is however unlikely, that both ISPs will apply changes 252 simultanously, which may lead to unconsistent state for the mappings 253 of the LISP site (e.g., weights for the same priority don't sum 100). 254 Since the different upstream ISPs are usually competing business 255 entities, the ETRs may even be configured to compete, either to 256 attract all the traffic or to get no traffic. The former will happen 257 if the customer pays per volume, the latter if the connectivity has a 258 fixed price. A solution could be to have the mappings in the Map- 259 Server(s), and have their operator give control over the entries to 260 customer, much like in today's DNS. 262 Additionally, since xTR1, xTR2, and xTR3 are in different 263 administrative domains, locator reachability information is unlikely 264 to be exchanged among them, making it difficult to set Loc-Status- 265 Bits correctly on encapsulated packets. 267 Compared to the customer edge scenario, deploying LISP at the 268 provider edge might have the advantage of diminishing potential MTU 269 issues, because the tunnel router is closer to the core, where links 270 typically have higher MTUs than edge network links. 272 2.3. Split ITR/ETR 274 In a simple LISP deployment, xTRs are located at the border of the 275 LISP site (see Section 2.1). In this scenario packets are routed 276 inside the domain according to the EID. However, more complex 277 networks may want to route packets according to the destination RLOC. 278 This would enable them to choose the best egress point. 280 The LISP specification separates the ITR and ETR functionality and 281 considers that both entities can be deployed in separated network 282 equipment. ITRs can be deployed closer to the host (i.e., access 283 routers). This way packets are encapsulated as soon as possible, and 284 packets exit the network through the best egress point in terms of 285 BGP policy. In turn, ETRs can be deployed at the border routers of 286 the network, and packets are decapsulated as soon as possible. 287 Again, once decapsulated packets are routed according to the EID, and 288 can follow the best path according to internal routing policy. 290 In the following figure we can see an example. The Source (S) 291 transmits packets using its EID and in this particular case packets 292 are encapsulated at ITR_1. The encapsulated packets are routed 293 inside the domain according to the destination RLOC, and can egress 294 the network through the best point (i.e., closer to the RLOC's AS). 295 On the other hand, inbound packets are received by ETR_1 which 296 decapsulates them. Then packets are routed towards S according to 297 the EID, again following the best path. 299 +---------------------------------------+ 300 | | 301 | +-------+ +-------+ +-------+ 302 | | ITR_1 |---------+ | ETR_1 |-RLOC_A--| ISP_A | 303 | +-------+ | +-------+ +-------+ 304 | +-+ | | | 305 | |S| | IGP | | 306 | +-+ | | | 307 | +-------+ | +-------+ +-------+ 308 | | ITR_2 |---------+ | ETR_2 |-RLOC_B--| ISP_B | 309 | +-------+ +-------+ +-------+ 310 | | 311 +---------------------------------------+ 313 Figure 3: Split ITR/ETR Scenario 315 This scenario has a set of implications: 317 o The site must carry at least partial BGP routes in order to choose 318 the best egress point, increasing the complexity of the network. 319 However, this is usually already the case for LISP sites that 320 would benefit from this scenario. 322 o If the site is multihomed to different ISPs and any of the 323 upstream ISPs is doing uRPF filtering, this scenario may become 324 impractical. ITRs need to determine the exit ETR, for setting the 325 correct source RLOC in the encapsulation header. This adds 326 complexity and reliability concerns. 328 o In LISP, ITRs set the reachability bits when encapsulating data 329 packets. Hence, ITRs need a mechanism to be aware of the liveness 330 of ETRs. 332 o ITRs encapsulate packets and in order to achieve efficient 333 communications, the MTU of the site must be large enough to 334 accommodate this extra header. 336 o In this scenario, each ITR is serving fewer hosts than in the case 337 when it is deployed at the border of the network. It has been 338 shown that cache hit ratio grows logarithmically with the amount 339 of users [cache]. Taking this into account, when ITRs are 340 deployed closer to the host the effectiveness of the mapping cache 341 may be lower (i.e., the miss ratio is higher). Another 342 consequence of this is that the site will transmit a higher amount 343 of Map-Requests, increasing the load on the distributed mapping 344 database. 346 2.4. Inter-Service Provider Traffic Engineering 348 With LISP, two LISP sites can route packets among them and control 349 their ingress TE policies. Typically, LISP is seen as applicable to 350 stub networks, however the LISP protocol can also be applied to 351 transit networks recursively. 353 Consider the scenario depicted in Figure 4. Packets originating from 354 the LISP site Stub1, client of ISP_A, with destination Stub4, client 355 of ISP_B, are LISP encapsulated at their entry point into the ISP_A's 356 network. The external IP header now has as the source RLOC an IP 357 from ISP_A's address space and destination RLOC from ISP_B's address 358 space. One or more ASes separate ISP_A from ISP_B. With a single 359 level of LISP encapsulation, Stub4 has control over its ingress 360 traffic. However, ISP_B only has the current tools (such as BGP 361 prefix deaggregation) to control on which of his own upstream or 362 peering links should packets enter. This is either not feasible (if 363 fine-grained per-customer control is required, the very specific 364 prefixes may not be propagated) or increases DFZ table size. 366 _.--. 367 Stub1 ... +-------+ ,-'' `--. +-------+ ... Stub3 368 \ | R_A1|----,' `. ---|R_B1 | / 369 --| R_A2|---( Transit ) | |-- 370 Stub2 .../ | R_A3|-----. ,' ---|R_B2 | \... Stub4 371 +-------+ `--. _.-' +-------+ 372 ... ISP_A `--'' ISP_B ... 374 Figure 4: Inter-Service provider TE scenario 376 A solution for this is to apply LISP recursively. ISP_A and ISP_B 377 may reach a bilateral agreement to deploy their own private mapping 378 system. ISP_A then encapsulates packets destined for the prefixes of 379 ISP_B, which are listed in the shared mapping system. Note that in 380 this case the packet is double-encapsulated. ISP_B's ETR removes the 381 outer, second layer of LISP encapsulation from the incoming packet, 382 and routes it towards the original RLOC, the ETR of Stub4, which does 383 the final decapsulation. 385 If ISP_A and ISP_B agree to share a private distributed mapping 386 database, both can control their ingress TE without the need of 387 disaggregating prefixes. In this scenario the private database 388 contains RLOC-to-RLOC bindings. The convergence time on the TE 389 policies updates is fast, since ISPs only have to update/query a 390 mapping to/from the database. 392 This deployment scenario includes two important recommendations. 393 First, it is intended to be deployed only between two ISPs (ISP_A and 394 ISP_B in Figure 4). If more than two ISPs use this approach, then 395 the xTRs deployed at the participating ISPs must either query 396 multiple mapping systems, or the ISPs must agree on a common shared 397 mapping system. Second, the scenario is only recommended for ISPs 398 providing connectivity to LISP sites, such that source RLOCs of 399 packets to be reencapsulated belong to said ISP. Otherwise the 400 participating ISPs must register prefixes they do not own in the 401 above mentioned private mapping system. Failure to follow these 402 recommendations may lead to operational and security issues when 403 deploying this scenario. 405 Besides these recommendations, the main disadvantages of this 406 deployment case are: 408 o Extra LISP header is needed. This increases the packet size and, 409 for efficient communications, it requires that the MTU between 410 both ISPs can accomodate double-encapsulated packets. 412 o The ISP ITR must encapsulate packets and therefore must know the 413 RLOC-to-RLOC binding. These bindings are stored in a mapping 414 database and may be cached in the ITR's mapping cache. Cache 415 misses lead to an extra lookup latency, unless NERD 416 [I-D.lear-lisp-nerd] is used for the lookups. 418 o The operational overhead of maintaining the shared mapping 419 database. 421 2.5. Tunnel Routers Behind NAT 423 NAT in this section refers to IPv4 network address and port 424 translation. 426 2.5.1. ITR 428 Packets encapsulated by an ITR are just UDP packets from a NAT 429 device's point of view, and they are handled like any UDP packet, 430 there are no additional requirements for LISP data packets. 432 Map-Requests sent by an ITR, which create the state in the NAT table 433 have a different 5-tuple in the IP header than the Map-Reply 434 generated by the authoritative ETR. Since the source address of this 435 packet is different from the destination address of the request 436 packet, no state will be matched in the NAT table and the packet will 437 be dropped. To avoid this, the NAT device has to do the following: 439 o Send all UDP packets with source port 4342, regardless of the 440 destination port, to the RLOC of the ITR. The most simple way to 441 achieve this is configuring 1:1 NAT mode from the external RLOC of 442 the NAT device to the ITR's RLOC (Called "DMZ" mode in consumer 443 broadband routers). 445 o Rewrite the ITR-AFI and "Originating ITR RLOC Address" fields in 446 the payload. 448 This setup supports a single ITR behind the NAT device. 450 2.5.2. ETR 452 An ETR placed behind NAT is reachable from the outside by the 453 Internet-facing locator of the NAT device. It needs to know this 454 locator (and configure a loopback interface with it), so that it can 455 use it in the Map-Replies. Thus support for dynamic locators for the 456 mapping database is needed in LISP equipment. 458 Again, only one ETR behind the NAT device is supported. 460 An implication of the issues described above is that LISP sites with 461 xTRs can not be behind carrier based NATs, since two different sites 462 would collide on the port forwarding. 464 2.6. Summary and Feature Matrix 466 Feature CE PE Split Rec. 467 -------------------------------------------------------- 468 Control of ingress TE x - x x 469 No modifications to existing 470 int. network infrastructure x x - - 471 Loc-Status-Bits sync x - x x 472 MTU/PMTUD issues minimized - x - x 474 3. Map-Resolvers and Map-Servers 476 3.1. Map-Servers 478 The Map-Server learns EID-to-RLOC mapping entries from an 479 authoritative source and publishes them in the distributed mapping 480 database. These entries are learned through authenticated Map- 481 Register messages sent by authoritative ETRs. Also, upon reception 482 of a Map-Request, the Map-Server verifies that the destination EID 483 matches an EID-prefix for which it is responsible for, and then re- 484 encapsulates and forwards it to a matching ETR. Map-Server 485 functionality is described in detail in [I-D.ietf-lisp-ms]. 487 The Map-Server is provided by a Mapping Service Provider (MSP). A 488 MSP can be any of the following: 490 o EID registrar. Since the IPv4 address space is nearing 491 exhaustion, IPv4 EIDs will come from already allocated Provider 492 Independent (PI) space. The registrars in this case remain the 493 current five Regional Internet Registries (RIRs). In the case of 494 IPv6, the possibility of reserving a /16 block as EID space is 495 currently under consideration [I-D.meyer-lisp-eid-block]. If 496 granted by IANA, the community will have to determine the body 497 resposible for allocations from this block, and the associated 498 policies. For already allocated IPv6 prefixes the principles from 499 IPv4 should be applied. 501 o Third parties. Participating in the LISP mapping system is 502 similar to participating in global routing or DNS: as long as 503 there is at least another already participating entity willing to 504 forward the newcomer's traffic, there is no barrier to entry. 505 Still, just like routing and DNS, LISP mappings have the issue of 506 trust, with efforts underway to make the published information 507 verifiable. When these mechanisms will be deployed in the LISP 508 mapping system, the burden of providing and verifying trust should 509 be kept away from MSPs, which will simply host the secured 510 mappings. This will keep the low barrier of entry to become an 511 MSP for third parties. 513 In all cases, the MSP configures its Map-Server(s) to publish the 514 prefixes of its clients in the distributed mapping database and start 515 encapsulating and forwarding Map-Requests to the ETRs of the AS. 516 These ETRs register their prefix(es) with the Map-Server(s) through 517 periodic authenticated Map-Register messages. In this context, for 518 some LISP end sites, there is a need for mechanisms to: 520 o Automatically distribute EID prefix(es) shared keys between the 521 ETRs and the EID-registrar Map-Server. 523 o Dynamically obtain the address of the Map-Server in the ETR of the 524 AS. 526 The Map-Server plays a key role in the reachability of the EID- 527 prefixes it is serving. On the one hand it is publishing these 528 prefixes into the distributed mapping database and on the other hand 529 it is encapsulating and forwarding Map-Requests to the authoritative 530 ETRs of these prefixes. ITRs encapsulating towards EIDs under the 531 responsibility of a failed Map-Server will be unable to look up any 532 of their covering prefixes. The only exception are the ITRs that 533 already contain the mappings in their local cache. In this case ITRs 534 can reach ETRs until the entry expires (typically 24 hours). For 535 this reason, redundant Map-Server deployments are desirable. A set 536 of Map-Servers providing high-availability service to the same set of 537 prefixes is called a redundancy group. ETRs are configured to send 538 Map-Register messages to all Map-Servers in the redundancy group. To 539 achieve fail-over (or load-balancing, if desired), current known BGP 540 practices can be used on the LISP+ALT BGP overlay network. 542 Additionally, if a Map-Server has no reachability for any ETR serving 543 a given EID block, it should not originate that block into the 544 mapping system. 546 3.2. Map-Resolvers 548 A Map-Resolver a is a network infrastructure component which accepts 549 LISP encapsulated Map-Requests, typically from an ITR, and finds the 550 appropriate EID-to-RLOC mapping by either consulting its local cache 551 or by consulting the distributed mapping database. Map-Resolver 552 functionality is described in detail in [I-D.ietf-lisp-ms]. 554 Anyone with access to the distributed mapping database can set up a 555 Map-Resolver and provide EID-to-RLOC mapping lookup service. In the 556 case of the LISP+ALT mapping system, the Map-Resolver needs to become 557 part of the ALT overlay so that it can forward packets to the 558 appropriate Map-Servers. For more detail on how the ALT overlay 559 works, see [I-D.ietf-lisp-alt] 561 For performance reasons, it is recommended that LISP sites use Map- 562 Resolvers that are topologically close to their ITRs. ISPs 563 supporting LISP will provide this service to their customers, 564 possibly restricting access to their user base. LISP sites not in 565 this position can use open access Map-Resolvers, if avaiable. 566 However, regardless of the availability of open access resolvers, the 567 MSP providing the Map-Server(s) for a LISP site should also make 568 available Map-Resolver(s) for the use of that site. 570 In medium to large-size ASes, ITRs must be configured with the RLOC 571 of a Map-Resolver, operation which can be done manually. However, in 572 Small Office Home Office (SOHO) scenarios a mechanism for 573 autoconfiguration should be provided. 575 One solution to avoid manual configuration in LISP sites of any size 576 is the use of anycast RLOCs for Map-Resolvers similar to the DNS root 577 server infrastructure. Since LISP uses UDP encapsulation, the use of 578 anycast would not affect reliability. LISP routers are then shipped 579 with a preconfigured list of well know Map-Resolver RLOCs, which can 580 be edited by the network administrator, if needed. 582 The use of anycast also helps improving mapping lookup performance. 583 Large MSPs can increase the number and geographical diversity of 584 their Map-Resolver infrastructure, using a single anycasted RLOC. 585 Once LISP deployment is advanced enough, very large content providers 586 may also be interested running this kind of setup, to ensure minimal 587 connection setup latency for those connecting to their network from 588 LISP sites. 590 While Map-Servers and Map-Resolvers implement different 591 functionalities withing the LISP mapping system, they can coexist on 592 the same device. For example, MSPs offering both services, can 593 deploy a single Map-Resolver/Map-Server in each PoP where they have a 594 presence, as long as load on the two functions is comparable. 596 4. Proxy Tunnel Routers 598 4.1. P-ITR 600 Proxy Ingress Tunnel Routers (P-ITRs) are part of the non-LISP/LISP 601 transition mechanism, allowing non-LISP sites to reach LISP sites. 602 They announce via BGP certain EID prefixes (aggregated, whenever 603 possible) to attract traffic from non-LISP sites towards EIDs in the 604 covered range. They do the mapping system lookup, and encapsulate 605 received packets towards the appropriate ETR. Note that for the 606 reverse path LISP sites can reach non-LISP sites simply by not 607 encapsulating traffic. See [I-D.ietf-lisp-interworking] for a 608 detailed description of P-ITR functionality. 610 The success of new protocols depends greatly on their ability to 611 maintain backwards compatibility and interoperate with the 612 protocol(s) they intend to enhance or replace, and on the incentives 613 to deploy the necessary new software or equipment. A LISP site needs 614 an interworking mechanism to be reachable from non-LISP sites. A 615 P-ITR can fulfill this role, enabling early adopters to see the 616 benefits of LISP, similar to tunnel brokers helping the transition 617 from IPv4 to IPv6. A site benefits from new LISP functionality 618 (proportionally with existing global LISP deployment) when going 619 LISP, so it has the incentives to deploy the necessary tunnel 620 routers. In order to be reachable from non-LISP sites it has two 621 options: keep announcing its prefix(es) with BGP (see next 622 subsection), or have a P-ITR announce prefix(es) covering them. 624 If the goal of reducing the DFZ routing table size is to be reached, 625 the second option is preferred. Moreover, the second option allows 626 LISP-based ingress traffic engineering from all sites. However, the 627 placement of P-ITRs greatly influences performance and deployment 628 incentives. The following subsections present the LISP+BGP 629 transition strategy and then possible P-ITR deployment scenarios. 630 They use the loosely defined terms of "early transition phase" and 631 "late transition phase", which refer to time periods when LISP sites 632 are a minority and a majority respectively. 634 4.1.1. LISP+BGP 636 For sites wishing to go LISP with their PI prefix the least 637 disruptive way is to upgrade their border routers to support LISP, 638 register the prefix into the LISP mapping system, but keep announcing 639 it with BGP as well. This way LISP sites will reach them over LISP, 640 while legacy sites will be unaffected by the change. The main 641 disadvantage of this approach is that no decrease in the DFZ routing 642 table size is achieved. Still, just increasing the number of LISP 643 sites is an important gain, as an increasing LISP/non-LISP site ratio 644 will slowly decrease the need for BGP-based traffic engineering that 645 leads to prefix deaggregation. That, in turn, may lead to a decrease 646 in the DFZ size in the late transition phase. 648 This scenario is not limited to sites that already have their 649 prefixes announced with BGP. Newly allocated EID blocks could follow 650 this strategy as well during the early LISP deployment phase, if the 651 costs of setting up BGP routing are lower than using P-ITR services, 652 or the expected performance is better. Since this leads to an 653 increase in the DFZ size, one of the following scenarios should be 654 preferred for new allocations. 656 4.1.2. Mapping Service Provider P-ITR Service 658 In addition to publishing their clients' registered prefixes in the 659 mapping system, MSPs with enough transit capacity can offer them 660 P-ITR service as a separate service. This service is especially 661 useful for new PI allocations, to sites without existing BGP 662 infrastructure, that wish to avoid BGP altogether. The MSP announces 663 the prefix into the DFZ, and the client benefits from ingress traffic 664 engineering without prefix deaggregation. The downside of this 665 scenario is path strech, which is greater than 1. 667 Routing all non-LISP ingress traffic through a third party which is 668 not one of its ISPs is only feasible for sites with modest amounts of 669 traffic (like those using the IPv6 tunnel broker services today), 670 especially in the first stage of the transition to LISP, with 671 sigficant numbers of legacy sites. When the LISP/non-LISP site ratio 672 becomes high enough, this approach can prove increasingly attractive. 674 Compared to LISP+BGP, this approach avoids DFZ bloat caused by prefix 675 deaggregation for traffic engineering purposes, resulting in slower 676 routing table increase in the case of new allocations and potential 677 decrease for existing ones. Moreover, MSPs serving different clients 678 with adjacent aggregable prefixes may lead to additional decrease, 679 but quantifying this decrease is subject to future research study. 681 4.1.3. Tier 1 P-ITR Service 683 The ideal location for a P-ITR is on the traffic path, as close to 684 non-LISP site as possible, to minimize or completely eliminate path 685 strech. However, this location is far away from the networks that 686 most benefit from the P-ITR services (i.e., LISP sites, destinations 687 of encapsulated traffic) and have the most incentives to deploy them. 688 But the biggest challenge having P-ITRs close to the traffic source 689 is the large number of devices and their wide geographical diversity 690 required to have a good coverage, in addition to considerable transit 691 capacity. Tier 1 service providers fulfill these requirements and 692 have clear incentives to deploy P-ITRs: to attract more traffic from 693 their customers. Since a large fraction is multihomed to different 694 providers with more than one active link, they compete with the other 695 providers for traffic. 697 To operate the P-ITR service, the ISP announces an aggregate of all 698 known EID prefixes (a mechanism will be needed to obtain this list) 699 downstream to their customers with BGP. First, the performance 700 concerns of the MSP P-ITR service described in the previous section 701 are now addressed, as P-ITRs are on-path, eliminating path strech 702 (except when combined with LISP+BGP, see below). Second, thanks to 703 the direction of the announcements, the DFZ routing table size is not 704 affected. 706 The main downside of this approach is non-global coverage for the 707 announced prefixes, caused by the dowstream direction of the 708 announcements. As a result, a LISP site will be only reachable from 709 customers of service providers running P-ITRs, unless one of the 710 previous approaches is used as well. Due to this issue, it is 711 unlikely that existing BGP speakers migrating to LISP will withdraw 712 their announcements to the DFZ, resulting in a combination of this 713 approach with LISP+BGP. At the same time, smaller new LISP sites 714 still depend on MSP for global reachability. The early transition 715 phase thus will keep the status quo in the DFZ routing table size, 716 but offers the benefits of increasingly better ingress traffic 717 engineering to early adopters. 719 As the number of LISP destinations increases, traffic levels from 720 those non-LISP, large multihomed clients who rely on BGP path length 721 for provider selection (such as national/regional ISPs), start to 722 shift towards the Tier 1 providing P-ITRs. The competition is then 723 incentivised to deploy their own service, thus improving global P-ITR 724 coverage. If all Tier 1 providers have P-ITR service, the LISP+BGP 725 and MSP alternatives are not required for global reachability of LISP 726 sites. Still, LISP+BGP user may still want to keep announcing their 727 prefixes for security reasons (i.e., preventing hijacking). DFZ size 728 evolution in this phase depends on that choice, and the aggragability 729 of all LISP prefixes. As a result, it may decrease or stay at the 730 same level. 732 For performance reasons, and to simplify P-ITR management, it is 733 desirable to minimize the number of non-aggregable EID prefixes. In 734 IPv6 this can be easily achieved if a large prefix block is reserved 735 as LISP EID space [I-D.meyer-lisp-eid-block]. If the EID space is 736 not fragmented, new LISP sites will not cause increase in the DFZ 737 size, unless they do LISP+BGP. 739 4.1.4. Migration Summary 741 The following table presents the expected effects of the different 742 transition scenarios during a certain phase on the DFZ routing table 743 size: 745 Phase | LISP+BGP | MSP | Tier-1 746 -----------------+--------------+-------------------+------------- 747 Early transition | no change | slowdown increase | no change 748 Late transition | may decrease | slowdown increase | may decrease 749 LISP Internet | considerable decrease 751 It is expected that a combination of these scenarios will exist 752 during the migration period, in particular existing sites chosing 753 LISP+BGP, new small sites choosing MSP, and competition between Tier 754 1 providers bringing optimized service. If all Tier 1 ISPs have 755 P-ITR service in place, the other scenarios can be deprecated, 756 greatly reducing DFZ size. 758 4.1.5. Content Provider Load Balancing 760 By deploying P-ITRs in strategic locations, traffic engineering could 761 be improved beyond what is currently offered by DNS, by adjusting 762 percentages of traffic flow to certain data centers, depending on 763 their load. This can be achieved by setting the appropriate 764 priorities, weights and loc-status-bits in mappings. And since the 765 P-ITRs are controlled by the content provider, changes can take place 766 instantaneously. 768 4.2. P-ETR 770 In contrast to P-ITRs, P-ETRs are not required for the correct 771 functioning of all LISP sites. There are two cases, where they can 772 be of great help: 774 o LISP sites with unicast reverse path forwarding (uRPF) 775 restrictions, and 777 o LISP sites without native IPv6 communicating with LISP nodes with 778 IPv6-only locators. 780 In the first case, uRPF filtering is applied at their upstream PE 781 router. When forwarding traffic to non-LISP sites, an ITR does not 782 encapsulate packets, leaving the original IP headers intact. As a 783 result, packets will have EIDs in their source address. Since we are 784 discussing the transition period, we can assume that a prefix 785 covering the EIDs belonging to the LISP site is advertized to the 786 global routing tables by a P-ITR, and the PE router has a route 787 towards it. However, the next hop will not be on the interface 788 towards the CE router, so non-encapsulated packets will fail uRPF 789 checks. 791 To avoid this filtering, the affected ITR encapsulates packets 792 towards the locator of the P-ETR for non-LISP destinations. Now the 793 source address of the packets, as seen by the PE router is the ITR's 794 locator, which will not fail the uRPF check. The P-ETR then 795 decapsulates and forwards the packets. 797 The second use case is IPv4-to-IPv6 transition. Service providers 798 using older access network hardware, which only supports IPv4 can 799 still offer IPv6 to their clients, by providing a CPE device running 800 LISP, and P-ETR(s) for accessing IPv6-only non-LISP sites and LISP 801 sites, with IPv6-only locators. Packets originating from the client 802 LISP site for these destinations would be encapsulated towards the 803 P-ETR's IPv4 locator. The P-ETR is in a native IPv6 network, 804 decapsulating and forwarding packets. For non-LISP destination, the 805 packet travels natively from the P-ETR. For LISP destinations with 806 IPv6-only locators, the packet will go through a P-ITR, in order to 807 reach its destination. 809 For more details on P-ETRs see the [I-D.ietf-lisp-interworking] 810 draft. 812 P-ETRs can be deployed by ISPs wishing to offer value-added services 813 to their customers. As is the case with P-ITRs, P-ETRs too may 814 introduce path stretch. Because of this the ISP needs to consider 815 the tradeoff of using several devices, close to the customers, to 816 minimize it, or few devices, farther away from the customers, 817 minimizing cost instead. 819 Since the deployment incentives for P-ITRs and P-ETRs are different, 820 it is likely they will be deployed in separate devices, except for 821 the CDN case, which may deploy both in a single device. 823 In all cases, the existance of a P-ETR involves another step in the 824 configuration of a LISP router. CPE routers, which are typically 825 configured by DHCP, stand to benefit most from P-ETRs. To enable 826 autoconfiguration of the P-ETR locator, a DHCP option would be 827 required. 829 As a security measure, access to P-ETRs should be limited to 830 legitimate users by enforcing ACLs. 832 5. Security Considerations 834 Security implications of LISP deployments are to be discussed in a 835 separate document. 837 6. IANA Considerations 839 This memo includes no request to IANA. 841 7. Acknowledgements 843 Many thanks to Margaret Wasserman for her contribution to the IETF76 844 presentation that kickstarted this work. The authors would also like 845 to thank Damien Saucez, Luigi Iannone, Joel Halpern, Vince Fuller, 846 Dino Farinacci, Terry Manderson, Noel Chiappa, and everyone else who 847 provided input. 849 8. References 851 8.1. Normative References 853 [I-D.ietf-lisp] 854 Farinacci, D., Fuller, V., Meyer, D., and D. Lewis, 855 "Locator/ID Separation Protocol (LISP)", 856 draft-ietf-lisp-09 (work in progress), October 2010. 858 [I-D.ietf-lisp-alt] 859 Fuller, V., Farinacci, D., Meyer, D., and D. Lewis, "LISP 860 Alternative Topology (LISP+ALT)", draft-ietf-lisp-alt-05 861 (work in progress), October 2010. 863 [I-D.ietf-lisp-interworking] 864 Lewis, D., Meyer, D., Farinacci, D., and V. Fuller, 865 "Interworking LISP with IPv4 and IPv6", 866 draft-ietf-lisp-interworking-01 (work in progress), 867 August 2010. 869 [I-D.ietf-lisp-ms] 870 Fuller, V. and D. Farinacci, "LISP Map Server", 871 draft-ietf-lisp-ms-06 (work in progress), October 2010. 873 8.2. Informative References 875 [I-D.lear-lisp-nerd] 876 Lear, E., "NERD: A Not-so-novel EID to RLOC Database", 877 draft-lear-lisp-nerd-08 (work in progress), March 2010. 879 [I-D.meyer-lisp-eid-block] 880 Lewis, D., Meyer, D., and V. Fuller, "LISP EID Block", 881 draft-meyer-lisp-eid-block-01 (work in progress), 882 May 2008. 884 [cache] Jung, J., Sit, E., Balakrishnan, H., and R. Morris, "DNS 885 performance and the effectiveness of caching", 2002. 887 Authors' Addresses 889 Lorand Jakab 890 Technical University of Catalonia 891 C/Jordi Girona, s/n 892 BARCELONA 08034 893 Spain 895 Email: ljakab@ac.upc.edu 897 Albert Cabellos-Aparicio 898 Technical University of Catalonia 899 C/Jordi Girona, s/n 900 BARCELONA 08034 901 Spain 903 Email: acabello@ac.upc.edu 905 Florin Coras 906 Technical University of Catalonia 907 C/Jordi Girona, s/n 908 BARCELONA 08034 909 Spain 911 Email: fcoras@ac.upc.edu 913 Jordi Domingo-Pascual 914 Technical University of Catalonia 915 C/Jordi Girona, s/n 916 BARCELONA 08034 917 Spain 919 Email: jordi.domingo@ac.upc.edu 920 Darrel Lewis 921 Cisco Systems 922 170 Tasman Drive 923 San Jose, CA 95134 924 USA 926 Email: darlewis@cisco.com