idnits 2.17.1 draft-ietf-rtgwg-cl-requirement-03.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 504 has weird spacing: '... packet trans...' == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: FR#5 Any automatic LSP routing and/or load balancing solutions MUST not oscillate such that performance observed by users changes such that an NPO is violated. Since oscillation may cause reordering, there MUST be means to control the frequency of changing the component link over which a flow is placed. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: DR#7 When a worst case failure scenario occurs, the number of RSVP-TE LSPs to be resignaled will cause a period of unavailability as perceived by users. The resignaling time of the solution MUST meet the NPO objective for the duration of unavailability. The resignaling time of the solution MUST not increase significantly as compared with current methods. -- The document date (January 10, 2011) is 4855 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'I-D.ietf-pwe3-fat-pw' is mentioned on line 717, but not defined == Missing Reference: 'IEEE-802.1AX' is mentioned on line 754, but not defined == Missing Reference: 'ITU-T.Y.1540' is mentioned on line 597, but not defined == Missing Reference: 'ITU-T.Y.1541' is mentioned on line 555, but not defined == Missing Reference: 'RFC1717' is mentioned on line 758, but not defined ** Obsolete undefined reference: RFC 1717 (Obsoleted by RFC 1990) == Missing Reference: 'RFC2475' is mentioned on line 701, but not defined == Missing Reference: 'RFC2615' is mentioned on line 760, but not defined == Missing Reference: 'RFC2991' is mentioned on line 716, but not defined == Missing Reference: 'RFC2992' is mentioned on line 716, but not defined == Missing Reference: 'RFC3260' is mentioned on line 702, but not defined == Missing Reference: 'RFC4201' is mentioned on line 782, but not defined == Missing Reference: 'RFC4301' is mentioned on line 621, but not defined == Missing Reference: 'RFC4385' is mentioned on line 714, but not defined == Missing Reference: 'RFC4928' is mentioned on line 715, but not defined == Unused Reference: 'RFC2702' is defined on line 442, but no explicit reference was found in the text == Unused Reference: 'RFC4665' is defined on line 467, but no explicit reference was found in the text == Unused Reference: 'RFC5254' is defined on line 484, but no explicit reference was found in the text == Outdated reference: A later version (-05) exists of draft-ietf-l2vpn-vpms-frmwk-requirements-03 Summary: 1 error (**), 0 flaws (~~), 22 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RTGWG C. Villamizar, Ed. 3 Internet-Draft Infinera Corporation 4 Intended status: Informational D. McDysan, Ed. 5 Expires: July 14, 2011 S. Ning 6 A. Malis 7 Verizon 8 L. Yong 9 Huawei USA 10 January 10, 2011 12 Requirements for MPLS Over a Composite Link 13 draft-ietf-rtgwg-cl-requirement-03 15 Abstract 17 There is often a need to provide large aggregates of bandwidth that 18 are best provided using parallel links between routers or MPLS LSR. 19 In core networks there is often no alternative since the aggregate 20 capacities of core networks today far exceed the capacity of a single 21 physical link or single packet processing element. 23 The presence of parallel links, with each link potentially comprised 24 of multiple layers has resulted in additional requirements. Certain 25 services may benefit from being restricted to a subset of the 26 component links or a specific component link, where component link 27 characteristics, such as latency, differ. Certain services require 28 that an LSP be treated as atomic and avoid reordering. Other 29 services will continue to require only that reordering not occur 30 within a microflow as is current practice. 32 Current practice related to multipath is described briefly in an 33 appendix. 35 Status of this Memo 37 This Internet-Draft is submitted in full conformance with the 38 provisions of BCP 78 and BCP 79. 40 Internet-Drafts are working documents of the Internet Engineering 41 Task Force (IETF). Note that other groups may also distribute 42 working documents as Internet-Drafts. The list of current Internet- 43 Drafts is at http://datatracker.ietf.org/drafts/current/. 45 Internet-Drafts are draft documents valid for a maximum of six months 46 and may be updated, replaced, or obsoleted by other documents at any 47 time. It is inappropriate to use Internet-Drafts as reference 48 material or to cite them other than as "work in progress." 49 This Internet-Draft will expire on July 14, 2011. 51 Copyright Notice 53 Copyright (c) 2011 IETF Trust and the persons identified as the 54 document authors. All rights reserved. 56 This document is subject to BCP 78 and the IETF Trust's Legal 57 Provisions Relating to IETF Documents 58 (http://trustee.ietf.org/license-info) in effect on the date of 59 publication of this document. Please review these documents 60 carefully, as they describe your rights and restrictions with respect 61 to this document. Code Components extracted from this document must 62 include Simplified BSD License text as described in Section 4.e of 63 the Trust Legal Provisions and are provided without warranty as 64 described in the Simplified BSD License. 66 Table of Contents 68 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 69 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 4 70 2. Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 4 71 3. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 4 72 4. Network Operator Functional Requirements . . . . . . . . . . . 5 73 4.1. Availability, Stability and Transient Response . . . . . . 5 74 4.2. Component Links Provided by Lower Layer Networks . . . . . 6 75 4.3. Parallel Component Links with Different Characteristics . 7 76 5. Derived Requirements . . . . . . . . . . . . . . . . . . . . . 9 77 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 10 78 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 79 8. Security Considerations . . . . . . . . . . . . . . . . . . . 10 80 9. References . . . . . . . . . . . . . . . . . . . . . . . . . . 11 81 9.1. Normative References . . . . . . . . . . . . . . . . . . . 11 82 9.2. Informative References . . . . . . . . . . . . . . . . . . 11 83 9.3. Appendix References . . . . . . . . . . . . . . . . . . . 12 84 Appendix A. More Details on Existing Network Operator 85 Practices and Protocol Usage . . . . . . . . . . . . 13 86 Appendix B. Existing Multipath Standards and Techniques . . . . . 15 87 B.1. Common Multpath Load Spliting Techniques . . . . . . . . . 16 88 B.2. Simple and Adaptive Load Balancing Multipath . . . . . . . 17 89 B.3. Traffic Split over Parallel Links . . . . . . . . . . . . 17 90 B.4. Traffic Split over Multiple Paths . . . . . . . . . . . . 18 91 Appendix C. ITU-T G.800 Composite Link Definitions and 92 Terminology . . . . . . . . . . . . . . . . . . . . . 18 93 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 19 95 1. Introduction 97 The purpose of this document is to describe why network operators 98 require certain functions in order to solve certain business problems 99 (Section 2). The intent is to first describe why things need to be 100 done in terms of functional requirements that are as independent as 101 possible of protocol specifications (Section 4). For certain 102 functional requirements this document describes a set of derived 103 protocol requirements (Section 5). Three appendices provide 104 supporting details as a summary of existing/prior operator approaches 105 (Appendix A), a summary of implementation techniques and relevant 106 protocol standards (Appendix B), and a summary of G.800 terminology 107 used to define a composite link (Appendix C). 109 1.1. Requirements Language 111 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 112 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 113 document are to be interpreted as described in RFC 2119 [RFC2119]. 115 2. Assumptions 117 The services supported include L3VPN RFC 4364 [RFC4364], RFC 4797 118 [RFC4797]L2VPN RFC 4664 [RFC4664] (VPWS, VPLS (RFC 4761 [RFC4761], 119 RFC 4762 [RFC4762]) and VPMS VPMS Framework 120 [I-D.ietf-l2vpn-vpms-frmwk-requirements]), Internet traffic 121 encapsulated by at least one MPLS label, and dynamically signaled 122 MPLS or MPLS-TP LSPs and pseudowires. The MPLS LSPs supporting these 123 services may be pt-pt, pt-mpt, or mpt-mpt. 125 The locations in a network where these requirements apply are a Label 126 Edge Router (LER) or a Label Switch Router (LSR) as defined in RFC 127 3031 [RFC3031]. 129 The IP DSCP cannot be used for flow identification since L3VPN 130 requires Diffserv transparency (see RFC 4031 5.5.2 [RFC4031]), and in 131 general network operators do not rely on the DSCP of Internet 132 packets. 134 3. Definitions 136 ITU-T G.800 Based Composite and Component Link Definitions: 137 Section 6.9.2 of ITU-T-G.800 [ITU-T.G.800] defines composite and 138 component links as summarized in Appendix C. The following 139 definitions for composite and component links are derived from 140 and intended to be consistent with the cited ITU-T G.800 141 terminology. 143 Composite Link: A composite link is a logical link composed of a 144 set of parallel point-to-point component links, where all 145 links in the set share the same endpoints. A composite link 146 may itself be a component of another composite link, but only 147 a strict hierarchy of links is allowed. 149 Component Link: A point-to-point physical or logical link that 150 preserves ordering in the steady state. A component link may 151 have transient out of order events, but such events must not 152 exceed the network's specific NPO. Examples of a physical 153 link are: Lambda, Ethernet PHY, and OTN. Examples of a 154 logical link are: MPLS LSP, Ethernet VLAN, and MPLS-TP LSP. 156 Flow: A sequence of packets that must be transferred in order on one 157 component link. 159 Flow identification: The label stack and other information that 160 uniquely identifies a flow. Other information in flow 161 identification may include an IP header, PW control word, 162 Ethernet MAC address, etc. Note that an LSP may contain one or 163 more Flows or an LSP may be equivalent to a Flow. Flow 164 identification is used to locally select a component link, or a 165 path through the network toward the destination. 167 Network Performance Objective (NPO): Numerical values for 168 performance measures, principally availability, latency, and 169 delay variation. See Appendix A for more details. 171 4. Network Operator Functional Requirements 173 The Functional Requirements in this section are grouped in 174 subsections starting with the highest priority. 176 4.1. Availability, Stability and Transient Response 178 Limiting the period of unavailability in response to failures or 179 transient events is extremely important as well as maintaining 180 stability. The transient period between some service disrupting 181 event and the convergence of the routing and/or signaling protocols 182 MUST occur within a time frame specified by NPO values. Appendix A 183 provides references and a summary of service types requiring a range 184 of restoration times. 186 FR#1 The solution SHALL provide a means to summarize some routing 187 advertisements regarding the characteristics of a composite 188 link such that the routing protocol converges within the 189 timeframe needed to meet the network performance objective. A 190 composite link CAN be announced in conjunction with detailed 191 parameters about its component links, such as bandwidth and 192 latency. The composite link SHALL behave as a single IGP 193 adjacency. 195 FR#2 The solution SHALL ensure that all possible restoration 196 operations happen within the timeframe needed to meet the NPO. 197 The solution may need to specify a means for aggregating 198 signaling to meet this requirement. 200 FR#3 The solution SHALL provide a mechanism to select a path for a 201 flow across a network that contains a number of paths comprised 202 of pairs of nodes connected by composite links in such a way as 203 to automatically distribute the load over the network nodes 204 connected by composite links while meeting all of the other 205 mandatory requirements stated above. The solution SHOULD work 206 in a manner similar to that of current networks without any 207 composite link protocol enhancements when the characteristics 208 of the individual component links are advertised. 210 FR#4 If extensions to existing protocols are specified and/or new 211 protocols are defined, then the solution SHOULD provide a means 212 for a network operator to migrate an existing deployment in a 213 minimally disruptive manner. 215 FR#5 Any automatic LSP routing and/or load balancing solutions MUST 216 not oscillate such that performance observed by users changes 217 such that an NPO is violated. Since oscillation may cause 218 reordering, there MUST be means to control the frequency of 219 changing the component link over which a flow is placed. 221 FR#6 Management and diagnostic protocols MUST be able to operate 222 over composite links. 224 4.2. Component Links Provided by Lower Layer Networks 226 Case 3 as defined in [ITU-T.G.800] involves a component link 227 supporting an MPLS layer network over another lower layer network 228 (e.g., circuit switched or another MPLS network (e.g., MPLS-TP)). 229 The lower layer network may change the latency (and/or other 230 performance parameters) seen by the MPLS layer network. Network 231 Operators have NPOs of which some components are based on performance 232 parameters. Currently, there is no protocol for the lower layer 233 network to inform the higher layer network of a change in a 234 performance parameter. Communication of the latency performance 235 parameter is a very important requirement. Communication of other 236 performance parameters (e.g., delay variation) is desirable. 238 FR#7 In order to support network NPOs and provide acceptable user 239 experience, the solution SHALL specify a protocol means to 240 allow a lower layer server network to communicate latency to 241 the higher layer client network. 243 FR#8 The precision of latency reporting SHOULD be at least 10% of 244 the one way latencies for latency of 1 ms or more. 246 FR#9 The solution SHALL provide a means to limit the latency on a 247 per LSP basis between nodes within a network to meet an NPO 248 target when the path between these nodes contains one or more 249 pairs of nodes connected via a composite link. 251 The NPOs differ across the services, and some services have 252 different NPOs for different QoS classes, for example, one QoS 253 class may have a much larger latency bound than another. 254 Overload can occur which would violate an NPO parameter (e.g., 255 loss) and some remedy to handle this case for a composite link 256 is required. 258 FR#10 If the total demand offered by traffic flows exceeds the 259 capacity of the composite link, the solution SHOULD define a 260 means to cause the LSPs for some traffic flows to move to some 261 other point in the network that is not congested. These 262 "preempted LSPs" may not be restored if there is no 263 uncongested path in the network. 265 4.3. Parallel Component Links with Different Characteristics 267 Corresponding to Case 1 of [ITU-T.G.800], as one means to provide 268 high availability, network operators deploy a topology in the MPLS 269 network using lower layer networks that have a certain degree of 270 diversity at the lower layer(s). Many techniques have been developed 271 to balance the distribution of flows across component links that 272 connect the same pair of nodes (See Appendix B.3). When the path for 273 a flow can be chosen from a set of candidate nodes connected via 274 composite links, other techniques have been developed (See 275 Appendix B.4). 277 FR#11 The solution SHALL measure traffic on a labeled traffic flow 278 and dynamically select the component link on which to place 279 this flow in order to balance the load so that no component 280 link in the composite link between a pair of nodes is 281 overloaded. 283 FR#12 When a traffic flow is moved from one component link to 284 another in the same composite link between a set of nodes (or 285 sites), it MUST be done so in a minimally disruptive manner. 287 When a flow is moved from a current link to a target link with 288 different latency, reordering can occur if the target link 289 latency is less than that of the current or clumping can occur 290 if target link latency is greater than that of the current. 291 Therefore, some flows (e.g., timing distribution, PW circuit 292 emulation) are quite sensitive to these effects, which may be 293 specified in an NPO or are needed to meet a user experience 294 objective (e.g. jitter buffer under/overrun). 296 FR#13 The solution SHALL provide a means to identify flows whose 297 rearrangement frequency needs to be bounded by a configured 298 value. 300 FR#14 The solution SHALL provide a means that communicates whether 301 the flows within an LSP can be split across multiple component 302 links. The solution SHOULD provide a means to indicate the 303 flow identification field(s) which can be used along the flow 304 path which can be used to perform this function. 306 FR#15 The solution SHALL provide a means to indicate that a traffic 307 flow shall select a component link with the minimum latency 308 value. 310 FR#16 The solution SHALL provide a means to indicate that a traffic 311 flow shall select a component link with a maximum acceptable 312 latency value as specified by protocol. 314 FR#17 The solution SHALL provide a means to indicate that a traffic 315 flow shall select a component link with a maximum acceptable 316 delay variation value as specified by protocol. 318 FR#18 The solution SHALL provide a means local to a node that 319 automatically distributes flows across the component links in 320 the composite link such that NPOs are met. 322 FR#19 The solution SHALL provide a means to distribute flows from a 323 single LSP across multiple component links to handle at least 324 the case where the traffic carried in an LSP exceeds that of 325 any component link in the composite link. As defined in 326 section 3, a flow is a sequence of packets that must be 327 transferred on one component link. 329 FR#20 The solution SHOULD support the use case where a composite 330 link itself is a component link for a higher order composite 331 link. For example, a composite link comprised of MPLS-TP bi- 332 directional tunnels viewed as logical links could then be used 333 as a component link in yet another composite link that 334 connects MPLS routers. 336 5. Derived Requirements 338 This section takes the next step and derives high-level requirements 339 on protocol specification from the functional requirements. 341 DR#1 The solution SHOULD attempt to extend existing protocols 342 wherever possible, developing a new protocol only if this adds 343 a significant set of capabilities. 345 The vast majority of network operators have provisioned L3VPN 346 services over LDP. Many have deployed L2VPN services over LDP 347 as well. TE extensions to IGP and RSVP-TE are viewed as being 348 overly complex by some operators. 350 DR#2 A solution SHOULD extend LDP capabilities to meet functional 351 requirements (without using TE methods as decided in 352 [RFC3468]). 354 DR#3 Coexistence of LDP and RSVP-TE signaled LSPs MUST be supported 355 on a composite link. Other functional requirements should be 356 supported as independently of signaling protocol as possible. 358 DR#4 When the nodes connected via a composite link are in the same 359 MPLS network topology, the solution MAY define extensions to 360 the IGP. 362 DR#5 When the nodes are connected via a composite link are in 363 different MPLS network topologies, the solution SHALL NOT rely 364 on extensions to the IGP. 366 DR#6 The Solution SHOULD support composite link IGP advertisement 367 that results in convergence time better than that of 368 advertising the individual component links. The solution SHALL 369 be designed so that it represents the range of capabilities of 370 the individual component links such that functional 371 requirements are met, and also minimizes the frequency of 372 advertisement updates which may cause IGP convergence to occur. 374 Examples of advertisement update triggering events to be 375 considered include: LSP establishment/release, changes in 376 component link characteristics (e.g., latency, up/down state), 377 and/or bandwidth utilization. 379 DR#7 When a worst case failure scenario occurs, the number of 380 RSVP-TE LSPs to be resignaled will cause a period of 381 unavailability as perceived by users. The resignaling time of 382 the solution MUST meet the NPO objective for the duration of 383 unavailability. The resignaling time of the solution MUST not 384 increase significantly as compared with current methods. 386 6. Acknowledgements 388 Frederic Jounay of France Telecom and Yuji Kamite of NTT 389 Communications Corporation co-authored a version of this document. 391 A rewrite of this document occurred after the IETF77 meeting. 392 Dimitri Papadimitriou, Lou Berger, Tony Li, the WG chairs John Scuder 393 and Alex Zinin, and others provided valuable guidance prior to and at 394 the IETF77 RTGWG meeting. 396 Tony Li and John Drake have made numerous valuable comments on the 397 RTGWG mailing list that are reflected in versions following the 398 IETF77 meeting. 400 7. IANA Considerations 402 This memo includes no request to IANA. 404 8. Security Considerations 406 This document specifies a set of requirements. The requirements 407 themselves do not pose a security threat. If these requirements are 408 met using MPLS signaling as commonly practiced today with 409 authenticated but unencrypted OSPF-TE, ISIS-TE, and RSVP-TE or LDP, 410 then the requirement to provide additional information in this 411 communication presents additional information that could conceivably 412 be gathered in a man-in-the-middle confidentiality breach. Such an 413 attack would require a capability to monitor this signaling either 414 through a provider breach or access to provider physical transmission 415 infrastructure. A provider breach already poses a threat of numerous 416 tpes of attacks which are of far more serious consequence. Encrption 417 of the signaling can prevent or render more difficult any 418 confidentiality breach that otherwise might occur by means of access 419 to provider physical transmission infrastructure. 421 9. References 423 9.1. Normative References 425 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 426 Requirement Levels", BCP 14, RFC 2119, March 1997. 428 9.2. Informative References 430 [I-D.ietf-l2vpn-vpms-frmwk-requirements] 431 Kamite, Y., JOUNAY, F., Niven-Jenkins, B., Brungard, D., 432 and L. Jin, "Framework and Requirements for Virtual 433 Private Multicast Service (VPMS)", 434 draft-ietf-l2vpn-vpms-frmwk-requirements-03 (work in 435 progress), July 2010. 437 [ITU-T.G.800] 438 ITU-T, "Unified functional architecture of transport 439 networks", 2007, . 442 [RFC2702] Awduche, D., Malcolm, J., Agogbua, J., O'Dell, M., and J. 443 McManus, "Requirements for Traffic Engineering Over MPLS", 444 RFC 2702, September 1999. 446 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 447 Label Switching Architecture", RFC 3031, January 2001. 449 [RFC3468] Andersson, L. and G. Swallow, "The Multiprotocol Label 450 Switching (MPLS) Working Group decision on MPLS signaling 451 protocols", RFC 3468, February 2003. 453 [RFC3809] Nagarajan, A., "Generic Requirements for Provider 454 Provisioned Virtual Private Networks (PPVPN)", RFC 3809, 455 June 2004. 457 [RFC4031] Carugi, M. and D. McDysan, "Service Requirements for Layer 458 3 Provider Provisioned Virtual Private Networks (PPVPNs)", 459 RFC 4031, April 2005. 461 [RFC4364] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 462 Networks (VPNs)", RFC 4364, February 2006. 464 [RFC4664] Andersson, L. and E. Rosen, "Framework for Layer 2 Virtual 465 Private Networks (L2VPNs)", RFC 4664, September 2006. 467 [RFC4665] Augustyn, W. and Y. Serbest, "Service Requirements for 468 Layer 2 Provider-Provisioned Virtual Private Networks", 469 RFC 4665, September 2006. 471 [RFC4761] Kompella, K. and Y. Rekhter, "Virtual Private LAN Service 472 (VPLS) Using BGP for Auto-Discovery and Signaling", 473 RFC 4761, January 2007. 475 [RFC4762] Lasserre, M. and V. Kompella, "Virtual Private LAN Service 476 (VPLS) Using Label Distribution Protocol (LDP) Signaling", 477 RFC 4762, January 2007. 479 [RFC4797] Rekhter, Y., Bonica, R., and E. Rosen, "Use of Provider 480 Edge to Provider Edge (PE-PE) Generic Routing 481 Encapsulation (GRE) or IP in BGP/MPLS IP Virtual Private 482 Networks", RFC 4797, January 2007. 484 [RFC5254] Bitar, N., Bocci, M., and L. Martini, "Requirements for 485 Multi-Segment Pseudowire Emulation Edge-to-Edge (PWE3)", 486 RFC 5254, October 2008. 488 9.3. Appendix References 490 [I-D.ietf-pwe3-fat-pw] 491 Bryant, S., Filsfils, C., Drafz, U., Kompella, V., Regan, 492 J., and S. Amante, "Flow Aware Transport of Pseudowires 493 over an MPLS PSN", draft-ietf-pwe3-fat-pw-03 (work in 494 progress), January 2010. 496 [IEEE-802.1AX] 497 IEEE Standards Association, "IEEE Std 802.1AX-2008 IEEE 498 Standard for Local and Metropolitan Area Networks - Link 499 Aggregation", 2006, . 502 [ITU-T.Y.1540] 503 ITU-T, "Internet protocol data communication service - IP 504 packet transfer and availability performance parameters", 505 2007, . 507 [ITU-T.Y.1541] 508 ITU-T, "Network performance objectives for IP-based 509 services", 2006, . 511 [RFC1717] Sklower, K., Lloyd, B., McGregor, G., and D. Carr, "The 512 PPP Multilink Protocol (MP)", RFC 1717, November 1994. 514 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., 515 and W. Weiss, "An Architecture for Differentiated 516 Services", RFC 2475, December 1998. 518 [RFC2615] Malis, A. and W. Simpson, "PPP over SONET/SDH", RFC 2615, 519 June 1999. 521 [RFC2991] Thaler, D. and C. Hopps, "Multipath Issues in Unicast and 522 Multicast Next-Hop Selection", RFC 2991, November 2000. 524 [RFC2992] Hopps, C., "Analysis of an Equal-Cost Multi-Path 525 Algorithm", RFC 2992, November 2000. 527 [RFC3260] Grossman, D., "New Terminology and Clarifications for 528 Diffserv", RFC 3260, April 2002. 530 [RFC4201] Kompella, K., Rekhter, Y., and L. Berger, "Link Bundling 531 in MPLS Traffic Engineering (TE)", RFC 4201, October 2005. 533 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 534 Internet Protocol", RFC 4301, December 2005. 536 [RFC4385] Bryant, S., Swallow, G., Martini, L., and D. McPherson, 537 "Pseudowire Emulation Edge-to-Edge (PWE3) Control Word for 538 Use over an MPLS PSN", RFC 4385, February 2006. 540 [RFC4928] Swallow, G., Bryant, S., and L. Andersson, "Avoiding Equal 541 Cost Multipath Treatment in MPLS Networks", BCP 128, 542 RFC 4928, June 2007. 544 Appendix A. More Details on Existing Network Operator Practices and 545 Protocol Usage 547 Often, network operators have a contractual Service Level Agreement 548 (SLA) with customers for services that are comprised of numerical 549 values for performance measures, principally availability, latency, 550 delay variation. Additionally, network operators may have Service 551 Level Sepcification (SLS) that is for internal use by the operator. 552 See [ITU-T.Y.1540], [ITU-T.Y.1541], RFC3809, Section 4.9 [RFC3809] 553 for examples of the form of such SLA and SLS specifications. In this 554 document we use the term Network Performance Objective (NPO) as 555 defined in section 5 of [ITU-T.Y.1541] since the SLA and SLS measures 556 have network operator and service specific implications. Note that 557 the numerical NPO values of Y.1540 and Y.1541 span multiple networks 558 and may be looser than network operator SLA or SLS objectives. 559 Applications and acceptable user experience have an important 560 relationship to these performance parameters. 562 Consider latency as an example. In some cases, minimizing latency 563 relates directly to the best customer experience (e.g., in TCP closer 564 is faster). I other cases, user experience is relatively insensitive 565 to latency, up to a specific limit at which point user perception of 566 quality degrades significantly (e.g., interactive human voice and 567 multimedia conferencing). A number of NPOs have. a bound on point- 568 point latency, and as long as this bound is met, the NPO is met -- 569 decreasing the latency is not necessary. In some NPOs, if the 570 specified latency is not met, the user considers the service as 571 unavailable. An unprotected LSP can be manually provisioned on a set 572 of to meet this type of NPO, but this lowers availability since an 573 alternate route that meets the latency NPO cannot be determined. 575 Historically, when an IP/MPLS network was operated over a lower layer 576 circuit switched network (e.g., SONET rings), a change in latency 577 caused by the lower layer network (e.g., due to a maintenance action 578 or failure) this was not known to the MPLS network. This resulted in 579 latency affecting end user experience, sometimes violating NPOs or 580 resulting in user complaints. 582 A response to this problem was to provision IP/MPLS networks over 583 unprotected circuits and set the metric and/or TE-metric proportional 584 to latency. This resulted in traffic being directed over the least 585 latency path, even if this was not needed to meet an NPO or meet user 586 experience objectives. This results in reduced flexibility and 587 increased cost for network operators. Using lower layer networks to 588 provide restoration and grooming is expected to be more efficient, 589 but the inability to communicate performance parameters, in 590 particular latency, from the lower layer network to the higher layer 591 network is an important problem to be solved before this can be done. 593 Latency NPOs for pt-pt services are often tied closely to geographic 594 locations, while latency for multipoint services may be based upon a 595 worst case within a region. 597 Section 7 of [ITU-T.Y.1540] defines availability for an IP service in 598 terms of loss exceeding a threshold for a period on the order of 5 599 minutes. However, the timeframes for restoration (i.e., as 600 implemented by pre-determined protection, convergence of routing 601 protocols and/or signaling) for services range from on the order of 602 100 ms or less (e.g., for VPWS to emulate classical SDH/SONET 603 protection switching), to several minutes (e.g., to allow BGP to 604 reconverge for L3VPN) and may differ among the set of customers 605 within a single service. 607 The presence of only three Traffic Class (TC) bits (previously known 608 as EXP bits) in the MPLS shim header is limiting when a network 609 operator needs to support QoS classes for multiple services (e.g., 610 L2VPN VPWS, VPLS, L3VPN and Internet), each of which has a set of QoS 611 classes that need to be supported. In some cases one bit is used to 612 indicate conformance to some ingress traffic classification, leaving 613 only two bits for indicating the service QoS classes. The approach 614 that has been taken is to aggregate these QoS classes into similar 615 sets on LER-LSR and LSR-LSR links. 617 Labeled LSPs have been and use of link layer encapsulation have been 618 standardized in order to provide a means to meet these needs. 620 The IP DSCP cannot be used for flow identification since RFC 4301 621 Section 5.5 [RFC4301] requires Diffserv transparency, and in general 622 network operators do not rely on the DSCP of Internet packets. 624 A label is pushed onto Internet packets when they are carried along 625 with L2/L3VPN packets on the same link or lower layer network 626 provides a mean to distinguish between the QoS class for these 627 packets. 629 Operating an MPLS-TE network involves a different paradigm from 630 operating an IGP metric-based LDP signaled MPLS network. The mpt-pt 631 LDP signaled MPLS LSPs occur automatically, and balancing across 632 parallel links occurs if the IGP metrics are set "equally" (with 633 equality a locally definable relation). 635 Traffic is typically comprised of a few large (some very large) flows 636 and many small flows. In some cases, separate LSPs are established 637 for very large flows. This can occur even if the IP header 638 information is inspected by a router, for example an IPsec tunnel 639 that carries a large amount of traffic. An important example of 640 large flows is that of a L2/L3 VPN customer who has an access line 641 bandwdith comparable to a client-client composite link bandwidth -- 642 there could be flows that are on the order of the access line 643 bandwdith. 645 Appendix B. Existing Multipath Standards and Techniques 647 Today the requirement to handle large aggregations of traffic, much 648 larger than a single component link, can be handled by a number of 649 techniques which we will collectively call multipath. Multipath 650 applied to parallel links between the same set of nodes includes 651 Ethernet Link Aggregation [IEEE-802.1AX], link bundling [RFC4201], or 652 other aggregation techniques some of which may be vendor specific. 653 Multipath applied to diverse paths rather than parallel links 654 includes Equal Cost MultiPath (ECMP) as applied to OSPF, ISIS, or 655 even BGP, and equal cost LSP, as described in Appendix B.4. Various 656 mutilpath techniques have strengths and weaknesses. 658 The term composite link is more general than terms such as link 659 aggregate which is generally considered to be specific to Ethernet 660 and its use here is consistent with the broad definition in 661 [ITU-T.G.800]. The term multipath excludes inverse multiplexing and 662 refers to techniques which only solve the problem of large 663 aggregations of traffic, without addressing the other requirements 664 outlined in this document. 666 B.1. Common Multpath Load Spliting Techniques 668 Identical load balancing techniqes are used for multipath both over 669 parallel links and over diverse paths. 671 Large aggregates of IP traffic do not provide explicit signaling to 672 indicate the expected traffic loads. Large aggregates of MPLS 673 traffic are carried in MPLS tunnels supported by MPLS LSP. LSP which 674 are signaled using RSVP-TE extensions do provide explicit signaling 675 which includes the expected traffic load for the aggregate. LSP 676 which are signaled using LDP do not provide an expected traffic load. 678 MPLS LSP may contain other MPLS LSP arranged hierarchically. When an 679 MPLS LSR serves as a midpoint LSR in an LSP carrying other LSP as 680 payload, there is no signaling associated with these inner LSP. 681 Therefore even when using RSVP-TE signaling there may be insufficient 682 information provided by signaling to adequately distribute load 683 across a composite link. 685 Generally a set of label stack entries that is unique across the 686 ordered set of label numbers can safely be assumed to contain a group 687 of flows. The reordering of traffic can therefore be considered to 688 be acceptable unless reordering occurs within traffic containing a 689 common unique set of label stack entries. Existing load splitting 690 techniques take advantage of this property in addition to looking 691 beyond the bottom of the label stack and determining if the payload 692 is IPv4 or IPv6 to load balance traffic accordingly. 694 MPLS-TP OAM violates the assumption that it is safe to reorder 695 traffic within an LSP. If MPLS-TP OAM is to be accommodated, then 696 existing multipth techniques must be modified. Such modifications 697 are outside the scope of this document. 699 For example a large aggregate of IP traffic may be subdivided into a 700 large number of groups of flows using a hash on the IP source and 701 destination addresses. This is as described in [RFC2475] and 702 clarified in [RFC3260]. For MPLS traffic carrying IP, a similar hash 703 can be performed on the set of labels in the label stack. These 704 techniques are both examples of means to subdivide traffic into 705 groups of flows for the purpose of load balancing traffic across 706 aggregated link capacity. The means of identifying a flow should not 707 be confused with the definition of a flow. 709 Discussion of whether a hash based approach provides a sufficiently 710 even load balance using any particular hashing algorithm or method of 711 distributing traffic across a set of component links is outside of 712 the scope of this document. 714 The current load balancing techniques are referenced in [RFC4385] and 715 [RFC4928]. The use of three hash based approaches are described in 716 [RFC2991] and [RFC2992]. A mechanism to identify flows within PW is 717 described in [I-D.ietf-pwe3-fat-pw]. The use of hash based 718 approaches is mentioned as an example of an existing set of 719 techniques to distribute traffic over a set of component links. 720 Other techniques are not precluded. 722 B.2. Simple and Adaptive Load Balancing Multipath 724 Simple multipath generally relies on the mathematical probability 725 that given a very large number of small microflows, these microflows 726 will tend to be distributed evenly across a hash space. A common 727 simple multipath implementation assumes that all members (component 728 links) are of equal capacity and perform a modulo operation across 729 the hashed value. An alternate simple multipath technique uses a 730 table generally with a power of two size, and distributes the table 731 entries proportionally among members according to the capacity of 732 each member. 734 Simple load balancing works well if there are a very large number of 735 small microflows (i.e., microflow rate is much less than component 736 link capacity). However, the case where there are even a few large 737 microflows is not handled well by simple load balancing. 739 An adaptive multipath technique is one where the traffic bound to 740 each member (component link) is measured and the load split is 741 adjusted accordingly. As long as the adjustment is done within a 742 single network element, then no protocol extensions are required and 743 there are no interoperability issues. 745 Note that if the load balancing algorithm and/or its parameters is 746 adjusted, then packets in some flows may be delivered out of 747 sequence. 749 B.3. Traffic Split over Parallel Links 751 The load spliting techniques defined in Appendix B.1 and Appendix B.2 752 are both used in splitting traffic over parallel links between the 753 same pair of nodes. The best known technique, though far from being 754 the first, is Ethernet Link Aggregation [IEEE-802.1AX]. This same 755 technique had been applied much earlier using OSPF or ISIS Equal Cost 756 MultiPath (ECMP) over parallel links between the same nodes. 758 Multilink PPP [RFC1717] uses a technique that provides inverse 759 multiplexing, however a number of vendors had provided proprietary 760 extensions to PPP over SONET/SDH [RFC2615] that predated Ethernet 761 Link Aggregation but are no longer used. 763 Link bundling [RFC4201] provides yet another means of handling 764 parallel LSP. RFC4201 explicitly allow a special value of all ones 765 to indicate a split across all members of the bundle. 767 B.4. Traffic Split over Multiple Paths 769 OSPF or ISIS Equal Cost MultiPath (ECMP) is a well known form of 770 traffic split over multiple paths that may traverse intermediate 771 nodes. ECMP is often incorrectly equated to only this case, and 772 multipath over multiple diverse paths is often incorrectly equated to 773 ECMP. 775 Many implementations are able to create more than one LSP between a 776 pair of nodes, where these LSP are routed diversely to better make 777 use of available capacity. The load on these LSP can be distributed 778 proportionally to the reserved bandwidth of the LSP. These multiple 779 LSP may be advertised as a single PSC FA and any LSP making use of 780 the FA may be split over these multiple LSP. 782 Link bundling [RFC4201] component links may themselves be LSP. When 783 this technique is used, any LSP which specifies the link bundle may 784 be split across the multiple paths of the LSP that comprise the 785 bundle. 787 Appendix C. ITU-T G.800 Composite Link Definitions and Terminology 789 Composite Link: 790 Section 6.9.2 of ITU-T-G.800 [ITU-T.G.800] defines composite link 791 in terms of three cases, of which the following two are relevant 792 (the one describing inverse (TDM) multiplexing does not apply). 793 Note that these case definitions are taken verbatim from section 794 6.9, "Layer Relationships". 796 Case 1: "Multiple parallel links between the same subnetworks 797 can be bundled together into a single composite link. Each 798 component of the composite link is independent in the sense 799 that each component link is supported by a separate server 800 layer trail. The composite link conveys communication 801 information using different server layer trails thus the 802 sequence of symbols crossing this link may not be preserved. 803 This is illustrated in Figure 14." 805 Case 3: "A link can also be constructed by a concatenation of 806 component links and configured channel forwarding 807 relationships. The forwarding relationships must have a 1:1 808 correspondence to the link connections that will be provided 809 by the client link. In this case, it is not possible to 810 fully infer the status of the link by observing the server 811 layer trails visible at the ends of the link. This is 812 illustrated in Figure 16." 814 Subnetwork: A set of one or more nodes (i.e., LER or LSR) and links. 815 As a special case it can represent a site comprised of multiple 816 nodes. 818 Forwarding Relationship: Configured forwarding between ports on a 819 subnetwork. It may be connectionless (e.g., IP, not considered 820 in this draft), or connection oriented (e.g., MPLS signaled or 821 configured). 823 Component Link: A topolological relationship between subnetworks 824 (i.e., a connection between nodes), which may be a wavelength, 825 circuit, virtual circuit or an MPLS LSP. 827 Authors' Addresses 829 Curtis Villamizar (editor) 830 Infinera Corporation 831 169 W. Java Drive 832 Sunnyvale, CA 94089 834 Email: cvillamizar@infinera.com 836 Dave McDysan (editor) 837 Verizon 838 22001 Loudoun County PKWY 839 Ashburn, VA 20147 841 Email: dave.mcdysan@verizon.com 842 So Ning 843 Verizon 844 2400 N. Glenville Ave. 845 Richardson, TX 75082 847 Phone: +1 972-729-7905 848 Email: ning.so@verizonbusiness.com 850 Andrew Malis 851 Verizon 852 117 West St. 853 Waltham, MA 02451 855 Phone: +1 781-466-2362 856 Email: andrew.g.malis@verizon.com 858 Lucy Yong 859 Huawei USA 860 1700 Alma Dr. Suite 500 861 Plano, TX 75075 863 Phone: +1 469-229-5387 864 Email: lucyyong@huawei.com