idnits 2.17.1 draft-ietf-rtgwg-cl-framework-01.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 -- The document date (August 12, 2012) is 4274 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-06) exists of draft-ietf-mpls-entropy-label-04 == Outdated reference: A later version (-16) exists of draft-ietf-rtgwg-cl-requirement-07 == Outdated reference: A later version (-06) exists of draft-ietf-rtgwg-cl-use-cases-01 == Outdated reference: A later version (-06) exists of draft-kompella-mpls-rsvp-ecmp-01 == Outdated reference: A later version (-03) exists of draft-villamizar-mpls-tp-multipath-02 == Outdated reference: A later version (-02) exists of draft-villamizar-mpls-tp-multipath-te-extn-01 -- Obsolete informational reference (is this intentional?): RFC 5316 (Obsoleted by RFC 9346) Summary: 0 errors (**), 0 flaws (~~), 7 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RTGWG S. Ning 3 Internet-Draft Tata Communications 4 Intended status: Informational D. McDysan 5 Expires: February 13, 2013 Verizon 6 E. Osborne 7 Cisco 8 L. Yong 9 Huawei USA 10 C. Villamizar 11 Outer Cape Cod Network 12 Consulting 13 August 12, 2012 15 Composite Link Framework in Multi Protocol Label Switching (MPLS) 16 draft-ietf-rtgwg-cl-framework-01 18 Abstract 20 This document specifies a framework for support of composite link in 21 MPLS networks. A composite link consists of a group of homogenous or 22 non-homogenous links that have the same forward adjacency and can be 23 considered as a single TE link or an IP link in routing. A composite 24 link relies on its component links to carry the traffic over the 25 composite link. Applicability is described for a single pair of 26 MPLS-capable nodes, a sequence of MPLS-capable nodes, or a set of 27 layer networks connecting MPLS-capable nodes. 29 Status of this Memo 31 This Internet-Draft is submitted in full conformance with the 32 provisions of BCP 78 and BCP 79. 34 Internet-Drafts are working documents of the Internet Engineering 35 Task Force (IETF). Note that other groups may also distribute 36 working documents as Internet-Drafts. The list of current Internet- 37 Drafts is at http://datatracker.ietf.org/drafts/current/. 39 Internet-Drafts are draft documents valid for a maximum of six months 40 and may be updated, replaced, or obsoleted by other documents at any 41 time. It is inappropriate to use Internet-Drafts as reference 42 material or to cite them other than as "work in progress." 44 This Internet-Draft will expire on February 13, 2013. 46 Copyright Notice 48 Copyright (c) 2012 IETF Trust and the persons identified as the 49 document authors. All rights reserved. 51 This document is subject to BCP 78 and the IETF Trust's Legal 52 Provisions Relating to IETF Documents 53 (http://trustee.ietf.org/license-info) in effect on the date of 54 publication of this document. Please review these documents 55 carefully, as they describe your rights and restrictions with respect 56 to this document. Code Components extracted from this document must 57 include Simplified BSD License text as described in Section 4.e of 58 the Trust Legal Provisions and are provided without warranty as 59 described in the Simplified BSD License. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 64 1.1. Architecture Summary . . . . . . . . . . . . . . . . . . . 4 65 1.2. Conventions used in this document . . . . . . . . . . . . 5 66 1.2.1. Terminology . . . . . . . . . . . . . . . . . . . . . 5 67 2. Composite Link Key Characteristics . . . . . . . . . . . . . . 5 68 2.1. Flow Identification . . . . . . . . . . . . . . . . . . . 6 69 2.2. Composite Link in Control Plane . . . . . . . . . . . . . 8 70 2.3. Composite Link in Data Plane . . . . . . . . . . . . . . . 11 71 3. Architecture Tradeoffs . . . . . . . . . . . . . . . . . . . . 11 72 3.1. Scalability Motivations . . . . . . . . . . . . . . . . . 12 73 3.2. Reducing Routing Information and Exchange . . . . . . . . 12 74 3.3. Reducing Signaling Load . . . . . . . . . . . . . . . . . 13 75 3.3.1. Reducing Signaling Load using LDP . . . . . . . . . . 14 76 3.3.2. Reducing Signaling Load using Hierarchy . . . . . . . 14 77 3.3.3. Using Both LDP and RSVP-TE Hierarchy . . . . . . . . . 14 78 3.4. Reducing Forwarding State . . . . . . . . . . . . . . . . 14 79 3.5. Avoiding Route Oscillation . . . . . . . . . . . . . . . . 15 80 4. New Challenges . . . . . . . . . . . . . . . . . . . . . . . . 16 81 4.1. Control Plane Challenges . . . . . . . . . . . . . . . . . 16 82 4.1.1. Delay and Jitter Sensitive Routing . . . . . . . . . . 17 83 4.1.2. Local Control of Traffic Distribution . . . . . . . . 17 84 4.1.3. Path Symmetry Requirements . . . . . . . . . . . . . . 17 85 4.1.4. Requirements for Contained LSP . . . . . . . . . . . . 18 86 4.1.5. Retaining Backwards Compatibility . . . . . . . . . . 19 87 4.2. Data Plane Challenges . . . . . . . . . . . . . . . . . . 19 88 4.2.1. Very Large LSP . . . . . . . . . . . . . . . . . . . . 20 89 4.2.2. Very Large Microflows . . . . . . . . . . . . . . . . 20 90 4.2.3. Traffic Ordering Constraints . . . . . . . . . . . . . 20 91 4.2.4. Accounting for IP and LDP Traffic . . . . . . . . . . 21 92 4.2.5. IP and LDP Limitations . . . . . . . . . . . . . . . . 21 93 5. Existing Mechanisms . . . . . . . . . . . . . . . . . . . . . 22 94 5.1. Link Bundling . . . . . . . . . . . . . . . . . . . . . . 22 95 5.2. Classic Multipath . . . . . . . . . . . . . . . . . . . . 24 97 6. Mechanisms Proposed in Other Documents . . . . . . . . . . . . 24 98 6.1. Loss and Delay Measurement . . . . . . . . . . . . . . . . 24 99 6.2. Link Bundle Extensions . . . . . . . . . . . . . . . . . . 25 100 6.3. Fat PW and Entropy Labels . . . . . . . . . . . . . . . . 26 101 6.4. Multipath Extensions . . . . . . . . . . . . . . . . . . . 26 102 7. Required Protocol Extensions and Mechanisms . . . . . . . . . 27 103 7.1. Brief Review of Requirements . . . . . . . . . . . . . . . 27 104 7.2. Required Document Coverage . . . . . . . . . . . . . . . . 28 105 7.2.1. Component Link Grouping . . . . . . . . . . . . . . . 28 106 7.2.2. Delay and Jitter Extensions . . . . . . . . . . . . . 29 107 7.2.3. Path Selection and Admission Control . . . . . . . . . 29 108 7.2.4. Dynamic Multipath Balance . . . . . . . . . . . . . . 30 109 7.2.5. Frequency of Load Balance . . . . . . . . . . . . . . 30 110 7.2.6. Inter-Layer Communication . . . . . . . . . . . . . . 30 111 7.2.7. Packet Ordering Requirements . . . . . . . . . . . . . 31 112 7.2.8. Minimally Disruption Load Balance . . . . . . . . . . 31 113 7.2.9. Path Symmetry . . . . . . . . . . . . . . . . . . . . 31 114 7.2.10. Performance, Scalability, and Stability . . . . . . . 32 115 7.2.11. IP and LDP Traffic . . . . . . . . . . . . . . . . . . 32 116 7.2.12. LDP Extensions . . . . . . . . . . . . . . . . . . . . 32 117 7.2.13. Pseudowire Extensions . . . . . . . . . . . . . . . . 33 118 7.2.14. Multi-Domain Composite Link . . . . . . . . . . . . . 33 119 7.3. Open Issues Regarding Requirements . . . . . . . . . . . . 34 120 7.4. Framework Requirement Coverage by Protocol . . . . . . . . 34 121 7.4.1. OSPF-TE and ISIS-TE Protocol Extensions . . . . . . . 35 122 7.4.2. PW Protocol Extensions . . . . . . . . . . . . . . . . 35 123 7.4.3. LDP Protocol Extensions . . . . . . . . . . . . . . . 35 124 7.4.4. RSVP-TE Protocol Extensions . . . . . . . . . . . . . 35 125 7.4.5. RSVP-TE Path Selection Changes . . . . . . . . . . . . 35 126 7.4.6. RSVP-TE Admission Control and Preemption . . . . . . . 35 127 7.4.7. Flow Identification and Traffic Balance . . . . . . . 35 128 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 35 129 9. Security Considerations . . . . . . . . . . . . . . . . . . . 36 130 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 36 131 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 36 132 11.1. Normative References . . . . . . . . . . . . . . . . . . . 36 133 11.2. Informative References . . . . . . . . . . . . . . . . . . 37 134 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 40 136 1. Introduction 138 Composite Link functional requirements are specified in 139 [I-D.ietf-rtgwg-cl-requirement]. Composite Link use cases are 140 described in [I-D.ietf-rtgwg-cl-use-cases]. This document specifies 141 a framework to meet these requirements. 143 Classic multipath, including Ethernet Link Aggregation has been 144 widely used in today's MPLS networks [RFC4385][RFC4928]. Classic 145 multipath using non-Ethernet links are often advertised using MPLS 146 Link bundling. A link bundle [RFC4201] bundles a group of 147 homogeneous links as a TE link to make IGP-TE information exchange 148 and RSVP-TE signaling more scalable. A composite link allows 149 bundling non-homogenous links together as a single logical link. The 150 motivations for using a composite link are descried in 151 [I-D.ietf-rtgwg-cl-requirement] and [I-D.ietf-rtgwg-cl-use-cases]. 153 This document describes a composite link framework in the context of 154 MPLS networks using an IGP-TE and RSVP-TE MPLS control plane with 155 GMPLS extensions [RFC3209][RFC3630][RFC3945][RFC5305]. 157 A composite link is a single logical link in MPLS network that 158 contains multiple parallel component links between two MPLS LSR. 159 Unlike a link bundle [RFC4201], the component links in a composite 160 link can have different properties such as cost or capacity. 162 Specific protocol solutions are outside the scope of this document, 163 however a framework for the extension of existing protocols is 164 provided. Backwards compatibility is best achieved by extending 165 existing protocols where practical rather than inventing new 166 protocols. The focus is on examining where existing protocol 167 mechanisms fall short with respect to [I-D.ietf-rtgwg-cl-requirement] 168 and on extensions that will be required to accommodate functionality 169 that is called for in [I-D.ietf-rtgwg-cl-requirement]. 171 1.1. Architecture Summary 173 Networks aggregate information, both in the control plane and in the 174 data plane, as a means to achieve scalability. A tradeoff exists 175 between the needs of scalability and the needs to identify differing 176 path and link characteristics and differing requirements among flows 177 contained within further aggregated traffic flows. These tradeoffs 178 are discussed in detail in Section 3. 180 Some aspects of Composite Link requirements present challenges for 181 which multiple solutions may exist. In Section 4 various challenges 182 and potential approaches are discussed. 184 A subset of the functionality called for in 185 [I-D.ietf-rtgwg-cl-requirement] is available through MPLS Link 186 Bundling [RFC4201]. Link bundling and other existing standards 187 applicable to Composite Link are covered in Section 5. 189 The most straightforward means of supporting Composite Link 190 requirements is to extend MPLS protocols and protocol semantics and 191 in particular to extend link bundling. Extensions which have already 192 been proposed in other documents which are applicable to Composite 193 Link are discussed in Section 6. 195 Goals of most new protocol work within IETF is to reuse existing 196 protocol encapsulations and mechanisms where they meet requirements 197 and extend existing mechanisms such that additional complexity is 198 minimized while meeting requirements and such that backwards 199 compatibility is preserved to the extent it is practical to do so. 200 These goals are considered in proposing a framework for further 201 protocol extensions and mechanisms in Section 7. 203 1.2. Conventions used in this document 205 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 206 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 207 document are to be interpreted as described in RFC 2119 [RFC2119]. 209 1.2.1. Terminology 211 Terminology defined in [I-D.ietf-rtgwg-cl-requirement] is used in 212 this document. 214 The abbreviation IGP-TE is used as a shorthand indicating either 215 OSPF-TE [RFC3630] or ISIS-TE [RFC5305]. 217 2. Composite Link Key Characteristics 219 [I-D.ietf-rtgwg-cl-requirement] defines external behavior of 220 Composite Links. The overall framework approach involves extending 221 existing protocols in a backwards compatible manner and reusing 222 ongoing work elsewhere in IETF where applicable, defining new 223 protocols or semantics only where necessary. Given the requirements, 224 and this approach of extending MPLS, Composite Link key 225 characteristics can be described in greater detail than given 226 requirements alone. 228 2.1. Flow Identification 230 Traffic mapping to component links is a data plane operation. 231 Control over how the mapping is done may be directly dictated or 232 constrained by the control plane or by the management plane. When 233 unconstrained by the control plane or management plane, distribution 234 of traffic is entirely a local matter. Regardless of constraints or 235 lack or constraints, the traffic distribution is required to keep 236 packets belonging to individual flows in sequence and meet QoS 237 criteria specified per LSP by either signaling or management 238 [RFC2475][RFC3260]. A key objective of the traffic distribution is 239 to not overload any component link, and be able to perform local 240 recovery when one of component link fails. 242 The network operator may have other objectives such as placing a 243 bidirectional flow or LSP on the same component link in both 244 direction, load balance over component links, composite link energy 245 saving, and etc. These new requirements are described in 246 [I-D.ietf-rtgwg-cl-requirement]. 248 Examples of means to identify a flow may in principle include: 250 1. an LSP identified by an MPLS label, 252 2. a sub-LSP [I-D.kompella-mpls-rsvp-ecmp] identified by an MPLS 253 label, 255 3. a pseudowire (PW) [RFC3985] identified by an MPLS PW label, 257 4. a flow or group of flows within a pseudowire (PW) [RFC6391] 258 identified by an MPLS flow label, 260 5. a flow or flow group in an LSP [I-D.ietf-mpls-entropy-label] 261 identified by an MPLS entropy label, 263 6. all traffic between a pair of IP hosts, identified by an IP 264 source and destination pair, 266 7. a specific connection between a pair of IP hosts, identified by 267 an IP source and destination pair, protocol, and protocol port 268 pair, 270 8. a layer-2 conversation within a pseudowire (PW), where the 271 identification is PW payload type specific, such as Ethernet MAC 272 addresses and VLAN tags within an Ethernet PW (RFC4448). 274 Although in principle a layer-2 conversation within a pseudowire 275 (PW), may be identified by PW payload type specific information, in 276 practice this is impractical at LSP midpoints when PW are carried. 277 The PW ingress may provide equivalent information in a PW flow label 278 [RFC6391]. Therefore, in practice, item #8 above is covered by 279 [RFC6391] and may be dropped from the list. 281 An LSR must at least be capable of identifying flows based on MPLS 282 labels. Most MPLS LSP do not require that traffic carried by the LSP 283 are carried in order. MPLS-TP is a recent exception. If it is 284 assumed that no LSP require strict packet ordering of the LSP itself 285 (only of flows within the LSP), then the entire label stack can be 286 used as flow identification. If some LSP may require strict packet 287 ordering but those LSP cannot be distinguished from others, then only 288 the top label can be used as a flow identifier. If only the top 289 label is used (for example, as specified by [RFC4201] when the "all- 290 ones" component described in [RFC4201] is not used), then there may 291 not be adequate flow granularity to accomplish well balanced traffic 292 distribution and it will not be possible to carry LSP that are larger 293 than any individual component link. 295 The number of flows can be extremely large. This may be the case 296 when the entire label stack is used and is always the case when IP 297 addresses are used in provider networks carrying Internet traffic. 298 Current practice for native IP load balancing at the time of writing 299 were documented in [RFC2991], [RFC2992]. These practices as 300 described, make use of IP addresses. The common practices were 301 extended to include the MPLS label stack and the common practice of 302 looking at IP addresses within the MPLS payload. These extended 303 practices are described in [RFC4385] and [RFC4928] due to their 304 impact on pseudowires without a PWE3 Control Word. Additional detail 305 on current multipath practices can be found in the appendices of 306 [I-D.ietf-rtgwg-cl-use-cases]. 308 Using only the top label supports too coarse a traffic balance. 309 Using the full label stack or IP addresses as flow identification 310 provides a sufficiently fine traffic balance, but is capable of 311 identifying such a high number of distinct flows, that a technique of 312 grouping flows, such as hashing on the flow identification criteria, 313 becomes essential to reduce the stored state, and is an essential 314 scaling technique. Other means of grouping flows may be possible. 316 In summary: 318 1. Load balancing using only the MPLS label stack provides too 319 coarse a granularity of load balance. 321 2. Tracking every flow is not scalable due to the extremely large 322 number of flows in provider networks. 324 3. Existing techniques, IP source and destination hash in 325 particular, have proven in over two decades of experience to be 326 an excellent way of identifying groups of flows. 328 4. If a better way to identify groups of flows is discovered, then 329 that method can be used. 331 5. IP address hashing is not required, but use of this technique is 332 strongly encouraged given the technique's long history of 333 successful deployment. 335 2.2. Composite Link in Control Plane 337 A composite Link is advertised as a single logical interface between 338 two connected routers, which forms forwarding adjacency (FA) between 339 the routers. The FA is advertised as a TE-link in a link state IGP, 340 using either OSPF-TE or ISIS-TE. The IGP-TE advertised interface 341 parameters for the composite link can be preconfigured by the network 342 operator or be derived from its component links. Composite link 343 advertisement requirements are specified in 344 [I-D.ietf-rtgwg-cl-requirement]. 346 In IGP-TE, a composite link is advertised as a single TE link between 347 two connected routers. This is similar to a link bundle [RFC4201]. 348 Link bundle applies to a set of homogenous component links. 349 Composite link allows homogenous and non-homogenous component links. 350 Due to the similarity, and for backwards compatibility, extending 351 link bundling is viewed as both simple and as the best approach. 353 In order for a route computation engine to calculate a proper path 354 for a LSP, it is necessary for composite link to advertise the 355 summarized available bandwidth as well as the maximum bandwidth that 356 can be made available for single flow (or single LSP where no finer 357 flow identification is available). If a composite link contains some 358 non-homogeneous component links, the composite link also should 359 advertise the summarized bandwidth and the maximum bandwidth for 360 single flow per each homogeneous component link group. 362 Both LDP [RFC5036] and RSVP-TE [RFC3209] can be used to signal a LSP 363 over a composite link. LDP cannot be extended to support traffic 364 engineering capabilities [RFC3468]. 366 When an LSP is signaled using RSVP-TE, the LSP MUST be placed on the 367 component link that meets the LSP criteria indicated in the signaling 368 message. 370 When an LSP is signaled using LDP, the LSP MUST be placed on the 371 component link that meets the LSP criteria, if such a component link 372 is available. LDP does not support traffic engineering capabilities, 373 imposing restrictions on LDP use of Composite Link. See 374 Section 4.2.5 for further details. 376 A composite link may contain non-homogeneous component links. The 377 route computing engine may select one group of component links for a 378 LSP. The routing protocol MUST make this grouping available in the 379 TE-LSDB. The route computation used in RSVP-TE MUST be extended to 380 include only the capacity of groups within a composite link which 381 meet LSP criteria. The signaling protocol MUST be able to indicate 382 either the criteria, or which groups may be used. A composite link 383 MUST place the LSP on a component link or group which meets or 384 exceeds the LSP criteria. 386 Composite link capacity is aggregated capacity. LSP capacity MAY be 387 larger than individual component link capacity. Any aggregated LSP 388 can determine a bounds on the largest microflow that could be carried 389 and this constraint can be handled as follows. 391 1. If no information is available through signaling, management 392 plane, or configuration, the largest microflow is bound by one of 393 the following: 395 A. the largest single LSP if most traffic is RSVP-TE signaled 396 and further aggregated, 398 B. the largest pseudowire if most traffic is carrying pseudowire 399 payloads that are aggregated within RSVP-TE LSP, 401 C. or the largest source and sink interface if a large amount of 402 IP or LDP traffic is contained within the aggregate. 404 If a very large amount of traffic being aggregated is IP or LDP, 405 then the largest microflow is bound by the largest component link 406 on which IP traffic can arrive. For example, if an LSR is acting 407 as an LER and IP and LDP traffic is arrving on 10 Gb/s edge 408 interfaces, then no microflow larger than 10 Gb/s will be present 409 on the RSVP-TE LSP that aggregate traffic across the core, even 410 if the core interfaces are 100 Gb/s interfaces. 412 2. The prior conditions provide a bound on the largest microflow 413 when no signaling extensions indicate a bounds. If an LSP is 414 aggregating smaller LSP for which the largest expected microflow 415 carried by the smaller LSP is signaled, then the largest 416 microflow expected in the containing LSP (the aggregate) is the 417 maximum of the largest expected microflow for any contained LSP. 418 For example, RSVP-TE LSP may be large but aggregate traffic for 419 which the source or sink are all 1 Gb/s or smaller interfaces 420 (such as in mobile applications in which cell sites backhauls are 421 no larger than 1 Gb/s). If this information is carried in the 422 LSP originated at the cell sites, then further aggregates across 423 a core may make use of this information. 425 3. The IGP must provide the bounds on the largest microflow that a 426 composite link can accommodate, which is the maximum capacity on 427 a component link that can be made available by moving other 428 traffic. This information is needed by the ingress LER for path 429 determination. 431 4. A means to signal an LSP whose capacity is larger than individual 432 component link capacity is needed [I-D.ietf-rtgwg-cl-requirement] 433 and also signal the largest microflow expected to be contained in 434 the LSP. If a bounds on the largest microflow is not signaled 435 there is no means to determine if an LSP which is larger than any 436 component link can be subdivided into flows and therefore should 437 be accepted by admission control. 439 When a bidirectional LSP request is signaled over a composite link, 440 if the request indicates that the LSP must be placed on the same 441 component link, the routers of the composite link MUST place the LSP 442 traffic in both directions on a same component link. This is 443 particularly challenging for aggregated capacity which makes use of 444 the label stack for traffic distribution. The two requirements are 445 mutually exclusive for any one LSP. No one LSP may be both larger 446 than any individual component link and require symmetrical paths for 447 every flow. Both requirements can be accommodated by the same 448 composite link for different LSP, with any one LSP requiring no more 449 than one of these two features. 451 Individual component link may fail independently. Upon component 452 link failure, a composite link MUST support a minimally disruptive 453 local repair, preempting any LSP which can no longer be supported. 454 Available capacity in other component links MUST be used to carry 455 impacted traffic. The available bandwidth after failure MUST be 456 advertised immediately to avoid looped crankback. 458 When a composite link is not able to transport all flows, it preempts 459 some flows based upon local management configuration and informs the 460 control plane on these preempted flows. The composite link MUST 461 support soft preemption [RFC5712]. This action ensures the remaining 462 traffic is transported properly. FR#10 requires that the traffic be 463 restored. FR#12 requires that any change be minimally disruptive. 464 These two requirements are interpreted to include preemption among 465 the types of changes that must be minimally disruptive. 467 2.3. Composite Link in Data Plane 469 The data plane must first identify groups of flows. Flow 470 identification is covered in Section 2.1. Having identified groups 471 of flows the groups must be placed on individual component links. 472 This second step is called traffic distribution or traffic placement. 473 The two steps together are known as traffic balancing or load 474 balancing. 476 Traffic distribution may be determined by or constrained by control 477 plane or management plane. Traffic distribution may be changed due 478 to component link status change, subject to constraints imposed by 479 either the management plane or control plane. The distribution 480 function is local to the routers in which a composite link belongs to 481 and is not specified here. 483 When performing traffic placement, a composite link does not 484 differentiate multicast traffic vs. unicast traffic. 486 In order to maintain scalability, existing data plane forwarding 487 retains state associated with the top label only. The use of flow 488 group identification is in a second step in the forwarding process. 489 Data plane forwarding makes use of the top label to select a 490 composite link, or a group of components within a composite link or 491 for the case where an LSP is pinned (see [RFC4201]), a specific 492 component link. For those LSP for which the LSP selects only the 493 composite link or a group of components within a composite link, the 494 load balancing makes use of the flow group identification. 496 The most common traffic placement techniques uses the a flow group 497 identification as an index into a table. The table provides an 498 indirection. The number of bits of hash is constrained to keep table 499 size small. While this is not the best technique, it is the most 500 common. Better techniques exist but they are outside the scope of 501 this document and some are considered proprietary. 503 Requirements to limit frequency of load balancing can be adhered to 504 by keeping track of when a flow group was last moved and imposing a 505 minimum period before that flow group can be moved again. This is 506 straightforward for a table approach. For other approaches it may be 507 less straightforward but is acheivable. 509 3. Architecture Tradeoffs 511 Scalability and stability are critical considerations in protocol 512 design where protocols may be used in a large network such as today's 513 service provider networks. Composite Link is applicable to networks 514 which are large enough to require that traffic be split over multiple 515 paths. Scalability is a major consideration for networks that reach 516 a capacity large enough to require Composite Link. 518 Some of the requirements of Composite Link could potentially have a 519 negative impact on scalability. For example, Composite Link requires 520 additional information to be carried in situations where component 521 links differ in some significant way. 523 3.1. Scalability Motivations 525 In the interest of scalability information is aggregated in 526 situations where information about a large amount of network capacity 527 or a large amount of network demand provides is adequate to meet 528 requirements. Routing information is aggregated to reduce the amount 529 of information exchange related to routing and to simplify route 530 computation (see Section 3.2). 532 In an MPLS network large routing changes can occur when a single 533 fault occurs. For example, a single fault may impact a very large 534 number of LSP traversing a given link. As new LSP are signaled to 535 avoid the fault, resources are consumed elsewhere, and routing 536 protocol announcements must flood the resource changes. If 537 protection is in place, there is less urgency to converging quickly. 538 If multiple faults occur that are not covered by shared risk groups 539 (SRG), then some protection may fail, adding urgency to converging 540 quickly even where protection was deployed. 542 Reducing the amount of information allows the exchange of information 543 during a large routing change to be accomplished more quickly and 544 simplifies route computation. Simplifying route computation improves 545 convergence time after very significant network faults which cannot 546 be handled by preprovisioned or precomputed protection mechanisms. 547 Aggregating smaller LSP into larger LSP is a means to reduce path 548 computation load and reduce RSVP-TE signaling (see Section 3.3). 550 Neglecting scaling issues can result in performance issues, such as 551 slow convergence. Neglecting scaling in some cases can result in 552 networks which perform so poorly as to become unstable. 554 3.2. Reducing Routing Information and Exchange 556 Link bundling at the very least provides a means of aggregating 557 control plane information. Even where the all-ones component link 558 supported by link bundling is not used, the amount of control 559 information is reduced by the average number of component links in a 560 bundle. 562 Fully deaggregating link bundle information would negate this 563 benefit. If there is a need to deaggregate, such as to distinguish 564 between groups of links within specified ranges of delay, then no 565 more deaggregation than is necessary should be done. 567 For example, in supporting the requirement for heterogeneous 568 component links, it makes little sense to fully deaggregate link 569 bundles when adding support for groups of component links with common 570 attributes within a link bundle can maintain most of the benefit of 571 aggregation while adequately supporting the requirement to support 572 heterogeneous component links. 574 Routing information exchange is also reduced by making sensible 575 choices regarding the amount of change to link parameters that 576 require link readvertisement. For example, if delay measurements 577 include queuing delay, then a much more coarse granularity of delay 578 measurement would be called for than if the delay does not include 579 queuing and is dominated by geographic delay (speed of light delay). 581 3.3. Reducing Signaling Load 583 Aggregating traffic into very large hierarchical LSP in the core very 584 substantially reduces the number of LSP that need to be signaled and 585 the number of path computations any given LSR will be required to 586 perform when a major network fault occurs. 588 In the extreme, applying MPLS to a very large network without 589 hierarchy could exceed the 20 bit label space. For example, in a 590 network with 4,000 nodes, with 2,000 on either side of a cutset, 591 would have 4,000,000 LSP crossing the cutset. Even in a degree four 592 cutset, an uneven distribution of LSP across the cutset, or the loss 593 of one link would result in a need to exceed the size of the label 594 space. Among provider networks, 4,000 access nodes is not at all 595 large. 597 In less extreme cases, having each node terminate hundreds of LSP to 598 achieve a full mesh creates a very large computational load. The 599 time complexity of one CSPF computation is order(N log N), where L is 600 proportional to N, and N and L are the number of nodes and number of 601 links, respectively. If each node must perform order(N) computations 602 when a fault occurs, then the computational load increases as 603 order(N^2 log N) as the number of nodes increases. In practice at 604 the time of writing, this imposes a limit of a few hundred nodes in a 605 full mesh of MPLS LSP before the computational load is sufficient to 606 result in unacceptable convergence times. 608 Two solutions are applied to reduce the amount of RSVP-TE signaling. 609 Both involve subdividing the MPLS domain into a core and a set of 610 regions. 612 3.3.1. Reducing Signaling Load using LDP 614 LDP can be used for edge-to-edge LSP, using RSVP-TE to carry the LDP 615 intra-core traffic and also optionally also using RSVP-TE to carry 616 the LDP intra-region traffic within each region. LDP does not 617 support traffic engineering, but does support multipoint-to-point 618 (MPTP) LSP, which require less signaling than edge-to-edge RSVP-TE 619 point-to-point (PTP) LSP. A drawback of this approach is the 620 inability to use RSVP-TE protection (FRR or GMPLS protection) against 621 failure of the border LSR sitting at a core/region boundary. 623 3.3.2. Reducing Signaling Load using Hierarchy 625 When the number of nodes grows too large, the amount of RSVP-TE 626 signaling can be reduced using the MPLS PSC hierarchy [RFC4206]. A 627 core within the hierarchy can divide the topology into M regions of 628 on average N/M nodes. Within a region the computational load is 629 reduced by more than M^2. Within the core, the computational load 630 generally becomes quite small since M is usually a fairly small 631 number (a few tens of regions) and each region is generally attached 632 to the core in typically only two or three places on average. 634 Using hierarchy improves scaling but has two consequences. First, 635 hierarchy effectively forces the use of platform label space. When a 636 containing LSP is rerouted, the labels assigned to the contained LSP 637 cannot be changed but may arrive on a different interface. Second, 638 hierarchy results in much larger LSP. These LSP today are larger 639 than any single component link and therefore force the use of the 640 all-ones component in link bundles. 642 3.3.3. Using Both LDP and RSVP-TE Hierarchy 644 It is also possible to use both LDP and RSVP-TE hierarchy. MPLS 645 networks with a very large number of nodes may benefit from the use 646 of both LDP and RSVP-TE hierarchy. The two techniques are certainly 647 not mutually exclusive. 649 3.4. Reducing Forwarding State 651 Both LDP and MPLS hierarchy have the benefit of reducing the amount 652 of forwarding state. Using the example from Section 3.3, and using 653 MPLS hierarchy, the worst case generally occurs at borders with the 654 core. 656 For example, consider a network with approximately 1,000 nodes 657 divided into 10 regions. At the edges, each node requires 1,000 LSP 658 to other edge nodes. The edge nodes also require 100 intra-region 659 LSP. Within the core, if the core has only 3 attachments to each 660 region the core LSR have less than 100 intra-core LSP. At the border 661 cutset between the core and a given region, in this example there are 662 100 edge nodes with inter-region LSP crossing that cutset, destined 663 to 900 other edge nodes. That yields forwarding state for on the 664 order of 90,000 LSP at the border cutset. These same routers need 665 only reroute well under 200 LSP when a multiple fault occurs, as long 666 as only links are affected and a border LSR does not go down. 668 In the core, the forwarding state is greatly reduced. If inter- 669 region LSP have different characteristics, it makes sense to make use 670 of aggregates with different characteristics. Rather than exchange 671 information about every inter-region LSP within the intra-core LSP it 672 makes more sense to use multiple intra-core LSP between pairs of core 673 nodes, each aggregating sets of inter-region LSP with common 674 characteristics or common requirements. 676 3.5. Avoiding Route Oscillation 678 Networks can become unstable when a feedback loop exists such that 679 moving traffic to a link causes a metric such as delay to increase, 680 which then causes traffic to move elsewhere. For example, the 681 original ARPANET routing used a delay based cost metric and proved 682 prone to route oscillations [DBP]. 684 Delay may be used as a constraint in routing for high priority 685 traffic, where the movement of traffic cannot impact the delay. The 686 safest way to measure delay is to make measurements based on traffic 687 which is prioritized such that it is queued ahead of the traffic 688 which will be affected. This is a reasonable measure of delay for 689 high priority traffic for which constraints have been set which allow 690 this type of traffic to consume only a fraction of link capacities 691 with the remaining capacity available to lower priority traffic. 693 Any measurement of jitter (delay variation) that is used in route 694 decision is likely to cause oscillation. Jitter that is caused by 695 queuing effects and cannot be measured using a very high priority 696 measurement traffic flow. 698 It may be possible to find links with constrained queuing delay or 699 jitter using a theoretical maximum or a probability based bound on 700 queuing delay or jitter at a given priority based on the types and 701 amounts of traffic accepted and combining that theoretical limit with 702 a measured delay at very high priority. 704 Instability can occur due to poor performance and interaction with 705 protocol timers. In this way a computational scaling problem can 706 become a stability problem when a network becomes sufficiently large. 707 For this reason, [I-D.ietf-rtgwg-cl-requirement] has a number of 708 requirements focusing on minimally impacting scalability. 710 4. New Challenges 712 New technical challenges are posed by [I-D.ietf-rtgwg-cl-requirement] 713 in both the control plane and data plane. 715 Among the more difficult challenges are the following. 717 1. requirements related delay or jitter (see Section 4.1.1), 719 2. the combination of ingress control over LSP placement and 720 retaining an ability to move traffic as demands dictate can pose 721 challenges and such requirements can even be conflicting (see 722 target="sect.local-control" />), 724 3. path symmetry requires extensions and is particularly challenging 725 for very large LSP (see Section 4.1.3), 727 4. accommodating a very wide range of requirements among contained 728 LSP can lead to inefficiency if the most stringent requirements 729 are reflected in aggregates, or reduce scalability if a large 730 number of aggregates are used to provide a too fine a reflection 731 of the requirements in the contained LSP (see Section 4.1.4), 733 5. backwards compatibility is somewhat limited due to the need to 734 accommodate legacy multipath interfaces which provide too little 735 information regarding their configured default behavior, and 736 legacy LSP which provide too little information regarding their 737 requirements (see Section 4.1.5), 739 6. data plane challenges include those of accommodating very large 740 LSP, large microflows, traffic ordering constraints imposed by a 741 subsent of LSP, and accounting for IP and LDP traffic (see 742 Section 4.2). 744 4.1. Control Plane Challenges 746 Some of the control plane requirements are particularly challenging. 747 Handling large flows which aggregate smaller flows must be 748 accomplished with minimal impact on scalability. Potentially 749 conflicting are requirements for jitter and requirements for 750 stability. Potentially conflicting are the requirements for ingress 751 control of a large number of parameters, and the requirements for 752 local control needed to achieve traffic balance across a composite 753 link. These challenges and potential solutions are discussed in the 754 following sections. 756 4.1.1. Delay and Jitter Sensitive Routing 758 Delay and jitter sensitive routing are called for in 759 [I-D.ietf-rtgwg-cl-requirement] in requirements FR#2, FR#7, FR#8, 760 FR#9, FR#15, FR#16, FR#17, FR#18. Requirement FR#17 is particularly 761 problematic, calling for constraints on jitter. 763 A tradeoff exists between scaling benefits of aggregating 764 information, and potential benefits of using a finer granularity in 765 delay reporting. To maintain the scaling benefit, measured link 766 delay for any given composite link SHOULD be aggregated into a small 767 number of delay ranges. IGP-TE extensions MUST be provided which 768 advertise the available capacities for each of the selected ranges. 770 For path selection of delay sensitive LSP, the ingress SHOULD bias 771 link metrics based on available capacity and select a low cost path 772 which meets LSP total path delay criteria. To communicate the 773 requirements of an LSP, the ERO MUST be extended to indicate the per 774 link constraints. To communicate the type of resource used, the RRO 775 SHOULD be extended to carry an identification of the group that is 776 used to carry the LSP at each link bundle hop. 778 4.1.2. Local Control of Traffic Distribution 780 Many requirements in [I-D.ietf-rtgwg-cl-requirement] suggest that a 781 node immediately adjacent to a component link should have a high 782 degree of control over how traffic is distributed, as long as network 783 performance objectives are met. Particularly relevant are FR#18 and 784 FR#19. 786 The requirements to allow local control are potentially in conflict 787 with requirement FR#21 which gives full control of component link 788 select to the LSP ingress. While supporting this capability is 789 mandatory, use of this feature is optional per LSP. 791 A given network deployment will have to consider this pair of 792 conflicting requirements and make appropriate use of local control of 793 traffic placement and ingress control of traffic placement to best 794 meet network requirements. 796 4.1.3. Path Symmetry Requirements 798 Requirement FR#21 in [I-D.ietf-rtgwg-cl-requirement] includes a 799 provision to bind both directions of a bidirectional LSP to the same 800 component. This is easily achieved if the LSP is directly signaled 801 across a composite link. This is not as easily achieved if a set of 802 LSP with this requirement are signaled over a large hierarchical LSP 803 which is in turn carried over a composite link. The basis for load 804 distribution in such as case is the label stack. The labels in 805 either direction are completely independent. 807 This could be accommodated if the ingress, egress, and all midpoints 808 of the hierarchical LSP make use of an entropy label in the 809 distribution, and use only that entropy label. A solution for this 810 problem may add complexity with very little benefit. There is little 811 or no true benefit of using symmetrical paths rather than component 812 links of identical characteristics. 814 Traffic symmetry and large LSP capacity are a second pair of 815 conflicting requirements. Any given LSP can meet one of these two 816 requirements but not both. A given network deployment will have to 817 make appropriate use of each of these features to best meet network 818 requirements. 820 4.1.4. Requirements for Contained LSP 822 [I-D.ietf-rtgwg-cl-requirement] calls for new LSP constraints. These 823 constraints include frequency of load balancing rearrangement, delay 824 and jitter, packet ordering constraints, and path symmetry. 826 When LSP are contained within hierarchical LSP, there is no signaling 827 available at midpoint LSR which identifies the contained LSP let 828 alone providing the set of requirements unique to each contained LSP. 829 Defining extensions to provide this information would severely impact 830 scalability and defeat the purpose of aggregating control information 831 and forwarding information into hierarchical LSP. For the same 832 scalability reasons, not aggregating at all is not a viable option 833 for large networks where scalability and stability problems may occur 834 as a result. 836 As pointed out in Section 4.1.3, the benefits of supporting symmetric 837 paths among LSP contained within hierarchical LSP may not be 838 sufficient to justify the complexity of supporting this capability. 840 A scalable solution which accommodates multiple sets of LSP between 841 given pairs of LSR is to provide multiple hierarchical LSP for each 842 given pair of LSR, each hierarchical LSP aggregating LSP with common 843 requirements and a common pair of endpoints. This is a network 844 design technique available to the network operator rather than a 845 protocol extension. This technique can accommodate multiple sets of 846 delay and jitter parameters, multiple sets of frequency of load 847 balancing parameters, multiple sets of packet ordering constraints, 848 etc. 850 4.1.5. Retaining Backwards Compatibility 852 Backwards compatibility and support for incremental deployment 853 requires considering the impact of legacy LSR in the role of LSP 854 ingress, and considering the impact of legacy LSR advertising 855 ordinary links, advertising Ethernet LAG as ordinary links, and 856 advertising link bundles. 858 Legacy LSR in the role of LSP ingress cannot signal requirements 859 which are not supported by their control plane software. The 860 additional capabilities supported by other LSR has no impact on these 861 LSR. These LSR however, being unaware of extensions, may try to make 862 use of scarce resources which support specific requirements such as 863 low delay. To a limited extent it may be possible for a network 864 operator to avoid this issue using existing mechanisms such as link 865 administrative attributes and attribute affinities [RFC3209]. 867 Legacy LSR advertising ordinary links will not advertise attributes 868 needed by some LSP. For example, there is no way to determine the 869 delay or jitter characteristics of such a link. Legacy LSR 870 advertising Ethernet LAG pose additional problems. There is no way 871 to determine that packet ordering constraints would be violated for 872 LSP with strict packet ordering constraints, or that frequency of 873 load balancing rearrangement constraints might be violated. 875 Legacy LSR advertising link bundles have no way to advertise the 876 configured default behavior of the link bundle. Some link bundles 877 may be configured to place each LSP on a single component link and 878 therefore may not be able to accommodate an LSP which requires 879 bandwidth in excess of the size of a component link. Some link 880 bundles may be configured to spread all LSP over the all-ones 881 component. For LSR using the all-ones component link, there is no 882 documented procedure for correctly setting the "Maximum LSP 883 Bandwidth". There is currently no way to indicate the largest 884 microflow that could be supported by a link bundle using the all-ones 885 component link. 887 Having received the RRO, it is possible for an ingress to look for 888 the all-ones component to identify such link bundles after having 889 signaled at least one LSP. Whether any LSR collects this information 890 on legacy LSR and makes use of it to set defaults, is an 891 implementation choice. 893 4.2. Data Plane Challenges 895 Flow identification is briefly discussed in Section 2.1. Traffic 896 distribution is briefly discussed in Section 2.3. This section 897 discusses issues specific to particular requirements specified in 899 [I-D.ietf-rtgwg-cl-requirement]. 901 4.2.1. Very Large LSP 903 Very large LSP may exceed the capacity of any single component of a 904 composite link. In some cases contained LSP may exceed the capacity 905 of any single component. These LSP may the use of the equivalent of 906 the all-ones component of a link bundle, or may use a subset of 907 components which meet the LSP requirements. 909 Very large LSP can be accommodated as long as they can be subdivided 910 (see Section 4.2.2). A very large LSP cannot have a requirement for 911 symetric paths unless complex protocol extensions are proposed (see 912 Section 2.2 and Section 4.1.3). 914 4.2.2. Very Large Microflows 916 Within a very large LSP there may be very large microflows. A very 917 large microflow is a very large flows which cannot be further 918 subdivided. Flows which cannot be subdivided must be no larger that 919 the capacity of any single component. 921 Current signaling provides no way to specify the largest microflow 922 that a can be supported on a given link bundle in routing 923 advertisements. Extensions which address this are discussed in 924 Section 6.4. Absent extensions of this type, traffic containing 925 microflows that are too large for a given composite link may be 926 present. There is no data plane solution for this problem that would 927 not require reordering traffic at the composite link egress. 929 Some techniques are susceptible to statistical collisions where an 930 algorithm to distribute traffic is unable to disambiguate traffic 931 among two or more very large microflow where their sum is in excess 932 of the capacity of any single component. Hash based algorithms which 933 use too small a hash space are particularly susceptible and require a 934 change in hash seed in the event that this were to occur. A change 935 in hash seed is highly disruptive, causing traffic reordering among 936 all traffic flows over which the hash function is applied. 938 4.2.3. Traffic Ordering Constraints 940 Some LSP have strict traffic ordering constraints. Most notable 941 among these are MPLS-TP LSP. In the absence of aggregation into 942 hierarchical LSP, those LSP with strict traffic ordering constraints 943 can be placed on individual component links if there is a means of 944 identifying which LSP have such a constraint. If LSP with strict 945 traffic ordering constraints are aggregated in hierarchical LSP, the 946 hierarchical LSP capacity may exceed the capacity of any single 947 component link. In such a case the load balancing for the containing 948 may be constrained to look only at the top label and the first 949 contained label. This and related issues are discussed further in 950 Section 6.4. 952 4.2.4. Accounting for IP and LDP Traffic 954 Networks which carry RSVP-TE signaled MPLS traffic generally carry 955 low volumes of native IP traffic, often only carrying control traffic 956 as native IP. There is no architectural guarantee of this, it is 957 just how network operators have made use of the protocols. 959 [I-D.ietf-rtgwg-cl-requirement] requires that native IP and native 960 LDP be accommodated. In some networks, a subset of services may be 961 carried as native IP or carried as native LDP. Today this may be 962 accommodated by the network operator estimating the contribution of 963 IP and LDP and configuring a lower set of available bandwidth figures 964 on the RSVP-TE advertisements. 966 The only improvement that Composite Link can offer is that of 967 measuring the IP and LDP traffic levels and automatically reducing 968 the available bandwidth figures on the RSVP-TE advertisements. The 969 measurements would have to be significantly filtered. This is 970 similar to a feature in existing LSR, commonly known as 971 "autobandwidth" with a key difference. In the "autobandwidth" 972 feature, the bandwidth request of an RSVP-TE signaled LSP is adjusted 973 in response to traffic measurements. In this case the IP or LDP 974 traffic measurements are used to reduce the link bandwidth directly, 975 without first encapsulating in an RSVP-TE LSP. 977 This may be a subtle and perhaps even a meaningless distinction if 978 Composite Link is used to form a Sub-Path Maintenance Element (SPME). 979 A SPME is in practice essentially an unsignaled single hop LSP with 980 PHP enabled [RFC5921]. A Composite Link SPME looks very much like 981 classic multipath, where there is no signaling, only management plane 982 configuration creating the multipath entity (of which Ethernet Link 983 Aggregation is a subset). 985 4.2.5. IP and LDP Limitations 987 IP does not offer traffic engineering. LDP cannot be extended to 988 offer traffic engineering [RFC3468]. Therefore there is no traffic 989 engineered fallback to an alternate path for IP and LDP traffic if 990 resources are not adequate for the IP and/or LDP traffic alone on a 991 given link in the primary path. The only option for IP and LDP would 992 be to declare the link down. Declaring a link down due to resource 993 exhaustion would reduce traffic to zero and eliminate the resource 994 exhaustion. This would cause oscillations and is therefore not a 995 viable solution. 997 Congestion caused by IP or LDP traffic loads is a pathologic case 998 that can occur if IP and/or LDP are carried natively and there is a 999 high volume of IP or LDP traffic. This situation can be avoided by 1000 carrying IP and LDP within RSVP-TE LSP. 1002 It is also not possible to route LDP traffic differently for 1003 different FEC. LDP traffic engineering is specifically disallowed by 1004 [RFC3468]. It may be possible to support multi-topology IGP 1005 extensions to accommodate more than one set of criteria. If so, the 1006 additional IGP could be bound to the forwarding criteria, and the LDP 1007 FEC bound to a specific IGP instance, inheriting the forwarding 1008 criteria. Alternately, one IGP instance can be used and the LDP SPF 1009 can make use of the constraints, such as delay and jitter, for a 1010 given LDP FEC. [Note: WG needs to discuss this and decide first 1011 whether to solve this at all and then if so, how.] 1013 5. Existing Mechanisms 1015 In MPLS the one mechanisms which support explicit signaling of 1016 multiple parallel links is Link Bundling [RFC4201]. The set of 1017 techniques known as "classis multipath" support no explicit 1018 signaling, except in two cases. In Ethernet Link Aggregation the 1019 Link Aggregation Control Protocol (LACP) coordinates the addition or 1020 removal of members from an Ethernet Link Aggregation Group (LAG). 1021 The use of the "all-ones" component of a link bundle indicates use of 1022 classis multipath, however the ability to determine if a link bundle 1023 makes use of classis multipath is not yet supported. 1025 5.1. Link Bundling 1027 Link bundling supports advertisement of a set of homogenous links as 1028 a single route advertisement. Link bundling supports placement of an 1029 LSP on any single component link, or supports placement of an LSP on 1030 the all-ones component link. Not all link bundling implementations 1031 support the all-ones component link. There is no way for an ingress 1032 LSR to tell which potential midpoint LSR support this feature and use 1033 it by default and which do not. Based on [RFC4201] it is unclear how 1034 to advertise a link bundle for which the all-ones component link is 1035 available and used by default. Common practice is to violate the 1036 specification and set the Maximum LSP Bandwidth to the Available 1037 Bandwidth. There is no means to determine the largest microflow that 1038 could be supported by a link bundle that is using the all-ones 1039 component link. 1041 [RFC6107] extends the procedures for hierarchical LSP but also 1042 extends link bundles. An LSP can be explicitly signaled to indicate 1043 that it is an LSP to be used as a component of a link bundle. Prior 1044 to that the common practice was to simply not advertise the component 1045 link LSP into the IGP, since only the ingress and egress of the link 1046 bundle needed to be aware of their existence, which they would be 1047 aware of due to the RSVP-TE signaling used in setting up the 1048 component LSP. 1050 While link bundling can be the basis for composite links, a 1051 significant number of small extension needs to be added. 1053 1. To support link bundles of heterogeneous links, a means of 1054 advertising the capacity available within a group of homogeneous 1055 needs to be provided. 1057 2. Attributes need to be defined to support the following parameters 1058 for the link bundle or for a group of homogeneous links. 1060 A. delay range 1062 B. jitter (delay variation) range 1064 C. group metric 1066 D. all-ones component capable 1068 E. capable of dynamically balancing load 1070 F. largest supportable microflow 1072 G. abilities to support strict packet ordering requirements 1073 within contained LSP 1075 3. For each of the prior extended attributes, the constraint based 1076 routing path selection needs to be extended to reflect new 1077 constraints based on the extended attributes. 1079 4. For each of the prior extended attributes, LSP admission control 1080 needs to be extended to reflect new constraints based on the 1081 extended attributes. 1083 5. Dynamic load balance must be provided for flows within a given 1084 set of links with common attributes such that NPO are not 1085 violated including frequency of load balance adjustment for any 1086 given flow. 1088 5.2. Classic Multipath 1090 Classic multipath is defined in [I-D.ietf-rtgwg-cl-use-cases]. 1092 Classic multipath refers to the most common current practice in 1093 implementation and deployment of multipath. The most common current 1094 practice makes use of a hash on the MPLS label stack and if IPv4 or 1095 IPv6 are indicated under the label stack, makes use of the IP source 1096 and destination addresses [RFC4385] [RFC4928]. 1098 Classic multipath provides a highly scalable means of load balancing. 1099 Adaptive multipath has proven value in assuring an even loading on 1100 component link and an ability to adapt to change in offerred load 1101 that occurs over periods of hundreds of milliseconds or more. 1102 Classic multipath scalability is due to the ability to effectively 1103 work with an extremely large number of flows (IP host pairs) using 1104 relatively little resources (a data structure accessed using a hash 1105 result as a key or using ranges of hash results). 1107 Classic multipath meets a small subset of Composite Link 1108 requirements. Due to scalability of the approach, classic multipath 1109 seems to be an excellent candidate for extension to meet the full set 1110 of Composite Link forwarding requirements. 1112 Additional detail can be found in [I-D.ietf-rtgwg-cl-use-cases]. 1114 6. Mechanisms Proposed in Other Documents 1116 A number of documents which at the time of writing are works in 1117 progress address parts of the requirements of Composite Link, or 1118 assist in making some of the goals achievable. 1120 6.1. Loss and Delay Measurement 1122 Procedures for measuring loss and delay are provided in [RFC6374]. 1123 These are OAM based measurements. This work could be the basis of 1124 delay measurements and delay variation measurement used for metrics 1125 called for in [I-D.ietf-rtgwg-cl-requirement]. 1127 Currently there are two additional Internet-Drafts that address delay 1128 and delay variation metrics. 1130 draft-wang-ccamp-latency-te-metric 1131 [I-D.wang-ccamp-latency-te-metric] is designed specifically to 1132 meet this requirement. OSPF-TE and ISIS-TE extensions are 1133 defined to indicate link delay and delay variance. The RSVP-TE 1134 ERO is extended to include service level requirements. A latency 1135 accumulation object is defined to provide a means of verification 1136 of the service level requirements. This draft is intended to 1137 proceed in the CCAMP WG. It is currently and individual 1138 submission. The 03 version of this draft expired in September 1139 2012. 1141 draft-giacalone-ospf-te-express-path 1142 This document proposes to extend OSPF-TE only. Extensions 1143 support delay, delay variance, loss, residual bandwidth, and 1144 available bandwidth. No extensions to RSVP-TE are proposed. 1145 This draft is intended to proceed in the CCAMP WG. It is 1146 currently and individual submission. The 02 version will expire 1147 in March 2012. 1149 A possible course of action may be to combine these two drafts. The 1150 delay variance, loss, residual bandwidth, and available bandwidth 1151 extensions are particular prone to network instability. The question 1152 as to whether queuing delay and delay variation should be considered, 1153 and if so for which diffserv Per-Hop Service Class (PSC) is not 1154 addressed. 1156 Note to co-authors: The ccamp-latency-te-metric draft refers to 1157 [I-D.ietf-rtgwg-cl-requirement] and is well matched to those 1158 requirements, including stability. The ospf-te-express-path draft 1159 refers to the "Alto Protocol" (draft-ietf-alto-protocol) and 1160 therefore may not be intended for RSVP-TE use. The authors of the 1161 two drafts may be able to resolve this. It may be best to drop ospf- 1162 te-express-path from this framework document. 1164 6.2. Link Bundle Extensions 1166 A set of link bundling extensions are defined in 1167 [I-D.ietf-mpls-explicit-resource-control-bundle]. This document 1168 provides extensions to the ERO and RRO to explicitly control the 1169 labels and resources within a bundle used by an LSP. 1171 The extensions in this document could be further extended to support 1172 indicating a group of component links in the ERO or RRO, where the 1173 group is given an interface identification like the bundle itself. 1174 The extensions could also be further extended to support 1175 specification of the all-ones component link in the ERO or RRO. 1177 [I-D.ietf-mpls-explicit-resource-control-bundle] does not provide a 1178 means to advertise the link bundle components. It is not certain how 1179 the ingress LSR would determine the set of link bundle component 1180 links available for a given link bundle. 1182 [I-D.ospf-cc-stlv] provides a baseline draft for extending link 1183 bundling to advertise components. A new component TVL (C-TLV) is 1184 proposed, which must reference a Composite Link Link TLV. 1185 [I-D.ospf-cc-stlv] is intended for the OSPF WG and submitted for the 1186 "Experimental" track. The 00 version expired in February 2012. 1188 6.3. Fat PW and Entropy Labels 1190 Two documents provide a means to add entropy for the purpose of 1191 improving load balance. MPLS encapsulation can bury information that 1192 is needed to identify microflows. These two documents allow a 1193 pseudowire ingress and LSP ingress respectively to add a label solely 1194 for the purpose of providing a finer granularity of microflow groups. 1196 [RFC6391] allows pseudowires which carry a large volume of traffic, 1197 where microflows can be identified to be load balanced across 1198 multiple members of an Ethernet LAG or an MPLS link bundle. This is 1199 accomplished by adding a flow label below the pseudowire label in the 1200 MPLS label stack. For this to be effective the link bundle load 1201 balance must make use of the label stack up to and including this 1202 flow label. 1204 [I-D.ietf-mpls-entropy-label] provides a means for a LER to put an 1205 additional label known as an entropy label on the MPLS label stack. 1206 As defined, only the LER can add the entropy label. 1208 Core LSR acting as LER for aggregated LSP can add entropy labels 1209 based on deep packet inspection and place an entropy label indicator 1210 (ELI) and entropy label (EL) just below the label being acted on. 1211 This would be helpful in situations where the label stack depth to 1212 which load distribution can operate is limited by implementation or 1213 is limited for other reasons such as carrying both MPLS-TP and MPLS 1214 with entropy labels within the same hierarchical LSP. 1216 6.4. Multipath Extensions 1218 The multipath extensions drafts address one aspect of Composite Link. 1219 These drafts deal with the issue of accommodating LSP which have 1220 strict packet ordering constraints in a network containing multipath. 1221 MPLS-TP has become the one important instance of LSP with strict 1222 packet ordering constraints and has driven this work. 1224 [I-D.villamizar-mpls-tp-multipath] outlines requirements and gives a 1225 number of options for dealing with the apparent incompatibility of 1226 MPLS-TP and multipath. A preferred option is described. 1228 [I-D.villamizar-mpls-tp-multipath-te-extn] provides protocol 1229 extensions needed to implement the preferred option described in 1230 [I-D.villamizar-mpls-tp-multipath]. 1232 Other issues pertaining to multipath are also addressed. Means to 1233 advertise the largest microflow supportable are defined. Means to 1234 indicate the largest expected microflow within an LSP are defined. 1235 Issues related to hierarchy are addressed. 1237 7. Required Protocol Extensions and Mechanisms 1239 Prior sections have reviewed key characteristics, architecture 1240 tradeoffs, new challenges, existing mechanisms, and relevant 1241 mechanisms proposed in existing new documents. 1243 This section first summarizes and groups requirements. A set of 1244 documents coverage groupings are proposed with existing works-in- 1245 progress noted where applicable. The set of extensions are then 1246 grouped by protocol affected as a convenience to implementors. 1248 7.1. Brief Review of Requirements 1250 The following list provides a categorization of requirements 1251 specified in [I-D.ietf-rtgwg-cl-requirement] along with a short 1252 phrase indication what topic the requirement covers. 1254 routing information aggregation 1255 FR#1 (routing summarization), FR#20 (composite link may be a 1256 component of another composite link) 1258 restoration speed 1259 FR#2 (restoration speed meeting NPO), FR#12 (minimally disruptive 1260 load rebalance), DR#6 (fast convergence), DR#7 (fast worst case 1261 failure convergence) 1263 load distribution, stability, minimal disruption 1264 FR#3 (automatic load distribution), FR#5 (must not oscillate), 1265 FR#11 (dynamic placement of flows), FR#12 (minimally disruptive 1266 load rebalance), FR#13 (bounded rearrangement frequency), FR#18 1267 (flow placement must satisfy NPO), FR#19 (flow identification 1268 finer than per top level LSP), MR#6 (operator initiated flow 1269 rebalance) 1271 backward compatibility and migration 1272 FR#4 (smooth incremental deployment), FR#6 (management and 1273 diagnostics must continue to function), DR#1 (extend existing 1274 protocols), DR#2 (extend LDP, no LDP TE) 1276 delay and delay variation 1277 FR#7 (expose lower layer measured delay), FR#8 (precision of 1278 latency reporting), FR#9 (limit latency on per LSP basis), FR#15 1279 (minimum delay path), FR#16 (bounded delay path), FR#17 (bounded 1280 jitter path) 1282 admission control, preemption, traffic engineering 1283 FR#10 (admission control, preemption), FR#14 (packet ordering), 1284 FR#21 (ingress specification of path), FR#22 (path symmetry), 1285 DR#3 (IP and LDP traffic), MR#3 (management specification of 1286 path) 1288 single vs multiple domain 1289 DR#4 (IGP extensions allowed within single domain), DR#5 (IGP 1290 extensions disallowed in multiple domain case) 1292 general network management 1293 MR#1 (polling, configuration, and notification), MR#2 (activation 1294 and de-activation) 1296 path determination, connectivity verification 1297 MR#4 (path trace), MR#5 (connectivity verification) 1299 The above list is not intended as a substitute for 1300 [I-D.ietf-rtgwg-cl-requirement], but rather as a concise grouping and 1301 reminder or requirements to serve as a means of more easily 1302 determining requirements coverage of a set of protocol documents. 1304 7.2. Required Document Coverage 1306 The primary areas where additional protocol extensions and mechanisms 1307 are required include the topics described in the following 1308 subsections. 1310 There are candidate documents for a subset of the topics below. This 1311 grouping of topics does not require that each topic be addressed by a 1312 separate document. In some cases, a document may cover multiple 1313 topics, or a specific topic may be addressed as applicable in 1314 multiple documents. 1316 7.2.1. Component Link Grouping 1318 An extension to link bundling is needed to specify a group of 1319 components with common attributes. This can be a TLV defined within 1320 the link bundle that carries the same encapsulations as the link 1321 bundle. Two interface indices would be needed for each group. 1323 a. An index is needed that if included in an ERO would indicate the 1324 need to place the LSP on any one component within the group. 1326 b. A second index is needed that if included in an ERO would 1327 indicate the need to balance flows within the LSP across all 1328 components of the group. This is equivalent to the "all-ones" 1329 component for the entire bundle. 1331 [I-D.ospf-cc-stlv] can be extended to include multipath treatment 1332 capabilities. An ISIS solution is also needed. An extension of 1333 RSVP-TE signaling is needed to indicate multipath treatment 1334 preferences. 1336 If a component group is allowed to support all of the parameters of a 1337 link bundle, then a group TE metric would be accommodated. This can 1338 be supported with the component TLV (C-TLV) defined in 1339 [I-D.ospf-cc-stlv]. 1341 The primary focus of this document, among the sets of requirements 1342 listed in Section 7.1 is the "routing information aggregation" set of 1343 requirements. The "restoration speed", "backward compatibility and 1344 migration", and "general network management" requirements must also 1345 be considered. 1347 7.2.2. Delay and Jitter Extensions 1349 A extension is needed in the IGP-TE advertisement to support delay 1350 and delay variation for links, link bundles, and forwarding 1351 adjacencies. Whatever mechanism is described must take precautions 1352 that insure that route oscillations cannot occur. 1353 [I-D.wang-ccamp-latency-te-metric] may be a good starting point. 1355 The primary focus of this document, among the sets of requirements 1356 listed in Section 7.1 is the "delay and delay variation" set of 1357 requirements. The "restoration speed", "backward compatibility and 1358 migration", and "general network management" requirements must also 1359 be considered. 1361 7.2.3. Path Selection and Admission Control 1363 Path selection and admission control changes must be documented in 1364 each document that proposes a protocol extension that advertises a 1365 new capability or parameter that must be supported by changes in path 1366 selection and admission control. 1368 The primary focus of this document, among the sets of requirements 1369 listed in Section 7.1 are the "load distribution, stability, minimal 1370 disruption" and "admission control, preemption, traffic engineering" 1371 sets of requirements. The "restoration speed" and "path 1372 determination, connectivity verification" requirements must also be 1373 considered. The "backward compatibility and migration", and "general 1374 network management" requirements must also be considered. 1376 7.2.4. Dynamic Multipath Balance 1378 FR#11 explicitly calls for dynamic load balancing similar to existing 1379 adaptive multipath. In implementations where flow identification 1380 uses a coarse granularity, the adjustments would have to be equally 1381 coarse, in the worst case moving entire LSP. The impact of flow 1382 identification granularity and potential adaptive multipath 1383 approaches may need to be documented in greater detail than provided 1384 here. 1386 The primary focus of this document, among the sets of requirements 1387 listed in Section 7.1 are the "restoration speed" and the "load 1388 distribution, stability, minimal disruption" sets of requirements. 1389 The "path determination, connectivity verification" requirements must 1390 also be considered. The "backward compatibility and migration", and 1391 "general network management" requirements must also be considered. 1393 7.2.5. Frequency of Load Balance 1395 IGP-TE and RSVP-TE extensions are needed to support frequency of load 1396 balancing rearrangement called for in FR#13, and FR#15-FR#17. 1397 Constraints are not defined in RSVP-TE, but could be modeled after 1398 administrative attribute affinities in RFC3209 and elsewhere. 1400 The primary focus of this document, among the sets of requirements 1401 listed in Section 7.1 is the "load distribution, stability, minimal 1402 disruption" set of requirements. The "path determination, 1403 connectivity verification" must also be considered. The "backward 1404 compatibility and migration" and "general network management" 1405 requirements must also be considered. 1407 7.2.6. Inter-Layer Communication 1409 Lower layer to upper layer communication called for in FR#7 and 1410 FR#20. This is addressed for a subset of parameters related to 1411 packet ordering in [I-D.villamizar-mpls-tp-multipath] where layers 1412 are MPLS. Remaining parameters, specifically delay and delay 1413 variation, need to be addressed. Passing information from a lower 1414 non-MPLS layer to an MPLS layer needs to be addressed, though this 1415 may largely be generic advice encouraging a coupling of MPLS to lower 1416 layer management plane or control plane interfaces. This topic can 1417 be addressed in each document proposing a protocol extension, where 1418 applicable. 1420 The primary focus of this document, among the sets of requirements 1421 listed in Section 7.1 is the "restoration speed" set of requirements. 1422 The "backward compatibility and migration" and "general network 1423 management" requirements must also be considered. 1425 7.2.7. Packet Ordering Requirements 1427 A document is needed to define extensions supporting various packet 1428 ordering requirements, ranging from requirements to preservce 1429 microflow ordering only, to requirements to preservce full LSP 1430 ordering (as in MPLS-TP). This is covered by 1431 [I-D.villamizar-mpls-tp-multipath] and 1432 [I-D.villamizar-mpls-tp-multipath-te-extn]. 1434 The primary focus of this document, among the sets of requirements 1435 listed in Section 7.1 are the "admission control, preemption, traffic 1436 engineering" and the "path determination, connectivity verification" 1437 sets of requirements. The "backward compatibility and migration" and 1438 "general network management" requirements must also be considered. 1440 7.2.8. Minimally Disruption Load Balance 1442 The behavior of hash methods used in classic multipath needs to be 1443 described in terms of FR#12 which calls for minimally disruptive load 1444 adjustments. For example, reseeding the hash violates FR#12. Using 1445 modulo operations is significantly disruptive if a link comes or goes 1446 down, as pointed out in [RFC2992]. In addition, backwards 1447 compatibility with older hardware needs to be accommodated. 1449 The primary focus of this document, among the sets of requirements 1450 listed in Section 7.1 is the "load distribution, stability, minimal 1451 disruption" set of requirements. 1453 7.2.9. Path Symmetry 1455 Protocol extensions are needed to support dynamic load balance as 1456 called for to meet FR#22 (path symmetry) and to meet FR#11 (dynamic 1457 placement of flows). Currently path symmetry can only be supported 1458 in link bundling if the path is pinned. When a flow is moved both 1459 ingress and egress must make the move as close to simultaneously as 1460 possible to satisfy FR#22 and FR#12 (minimally disruptive load 1461 rebalance). If a group of flows are identified using a hash, then 1462 the hash must be identical on the pair of LSR at the endpoint, using 1463 the same hash seed and with one side swapping source and destination. 1464 If the label stack is used, then either the entire label stack must 1465 be a special case flow identification, since the set of labels in 1466 either direction are not correlated, or the two LSR must conspire to 1467 use the same flow identifier. For example, using a common entropy 1468 label value, and using only the entropy label in the flow 1469 identification would satisfy this requirement. 1471 The primary focus of this document, among the sets of requirements 1472 listed in Section 7.1 are the "load distribution, stability, minimal 1473 disruption" and the "admission control, preemption, traffic 1474 engineering" sets of requirements. The "backward compatibility and 1475 migration" and "general network management" requirements must also be 1476 considered. Path symetry simplifies support for the "path 1477 determination, connectivity verification" set of requirements, but 1478 with significant complexity added elsewhere. 1480 7.2.10. Performance, Scalability, and Stability 1482 A separate document providing analysis of performance, scalability, 1483 and stability impacts of changes may be needed. The topic of traffic 1484 adjustment oscillation must also be covered. If sufficient coverage 1485 is provided in each document covering a protocol extension, a 1486 separate document would not be needed. 1488 The primary focus of this document, among the sets of requirements 1489 listed in Section 7.1 is the "restoration speed" set of requirements. 1490 This is not a simple topic and not a topic that is well served by 1491 scattering it over multiple documents, therefore it may be best to 1492 put this in a separate document and put citations in documents called 1493 for in Section 7.2.1, Section 7.2.2, Section 7.2.3, Section 7.2.9, 1494 Section 7.2.11, Section 7.2.12, Section 7.2.13, and Section 7.2.14. 1495 Citation may also be helpful in Section 7.2.4, and Section 7.2.5. 1497 7.2.11. IP and LDP Traffic 1499 A document is needed to define the use of measurements native IP and 1500 native LDP traffic levels to reduce link advertised bandwidth 1501 amounts. 1503 The primary focus of this document, among the sets of requirements 1504 listed in Section 7.1 are the "load distribution, stability, minimal 1505 disruption" and the "admission control, preemption, traffic 1506 engineering" set of requirements. The "path determination, 1507 connectivity verification" must also be considered. The "backward 1508 compatibility and migration" and "general network management" 1509 requirements must also be considered. 1511 7.2.12. LDP Extensions 1513 Extending LDP is called for in DR#2. LDP can be extended to couple 1514 FEC admission control to local resource availability without 1515 providing LDP traffic engineering capability. Other LDP extensions 1516 such as signaling a bound on microflow size and LDP LSP requirements 1517 would provide useful information without providing LDP traffic 1518 engineering capability. 1520 The primary focus of this document, among the sets of requirements 1521 listed in Section 7.1 is the "admission control, preemption, traffic 1522 engineering" set of requirements. The "backward compatibility and 1523 migration" and "general network management" requirements must also be 1524 considered. 1526 7.2.13. Pseudowire Extensions 1528 PW extensions such as signaling a bound on microflow size and PW 1529 requirements would provide useful information. 1531 The primary focus of this document, among the sets of requirements 1532 listed in Section 7.1 is the "admission control, preemption, traffic 1533 engineering" set of requirements. The "backward compatibility and 1534 migration" and "general network management" requirements must also be 1535 considered. 1537 7.2.14. Multi-Domain Composite Link 1539 DR#5 calls for Composite Link to span multiple network topologies. 1540 Component LSP may already span multiple network topologies, though 1541 most often in practice these are LDP signaled. Component LSP which 1542 are RSVP-TE signaled may also span multiple network topologies using 1543 at least three existing methods (per domain [RFC5152], BRPC 1544 [RFC5441], PCE [RFC4655]). When such component links are combined in 1545 a Composite Link, the Composite Link spans multiple network 1546 topologies. It is not clear in which document this needs to be 1547 described or whether this description in the framework is sufficient. 1548 The authors and/or the WG may need to discuss this. DR#5 mandates 1549 that IGP-TE extension cannot be used. This would disallow the use of 1550 [RFC5316] or [RFC5392] in conjunction with [RFC5151]. 1552 The primary focus of this document, among the sets of requirements 1553 listed in Section 7.1 are "single vs multiple domain" and "admission 1554 control, preemption, traffic engineering". The "routing information 1555 aggregation" and "load distribution, stability, minimal disruption" 1556 requirements need attention due to their use of the IGP in single 1557 domain Composite Link. Other requirements such as "delay and delay 1558 variation", can more easily be accomodated by carrying metrics within 1559 BGP. The "path determination, connectivity verification" 1560 requirements need attention due to requirements to restrict 1561 disclosure of topology information across domains in multi-domain 1562 deployments. The "backward compatibility and migration" and "general 1563 network management" requirements must also be considered. 1565 7.3. Open Issues Regarding Requirements 1567 Note to co-authors: This section needs to be reduced to an empty 1568 section and then removed. 1570 The following topics in the requirements document are not addressed. 1571 Since they are explicitly mentioned in the requirements document some 1572 mention of how they are supported is needed, even if to say nother 1573 needed to be done. If we conclude any particular topic is 1574 irrelevant, maybe the topic should be removed from the requirement 1575 document. At that point we could add the management requirements 1576 that have come up and were missed. 1578 1. L3VPN RFC 4364, RFC 4797,L2VPN RFC 4664, VPWS, VPLS RFC 4761, RFC 1579 4762 and VPMS VPMS Framework 1580 (draft-ietf-l2vpn-vpms-frmwk-requirements). It is not clear what 1581 additional Composite Link requirements these references imply, if 1582 any. If no additional requirements are implied, then these 1583 references are considered to be informational only. 1585 2. Migration may not be adequately covered in Section 4.1.5. It 1586 might also be necessary to say more here on performance, 1587 scalability, and stability as it related to migration. Comments 1588 on this from co-authors or the WG? 1590 3. We may need a performance section in this document to 1591 specifically address #DR6 (fast convergence), and #DR7 (fast 1592 worst case failure convergence), though we do already have 1593 scalability discussion. The performance section would have to 1594 say "no worse than before, except were there was no alternative 1595 to make it very slightly worse" (in a bit more detail than that). 1596 It would also have to better define the nature of the performance 1597 criteria. 1599 7.4. Framework Requirement Coverage by Protocol 1601 As an aid to implementors, this section summarizes requirement 1602 coverage listed in Section 7.2 by protocol or LSR functionality 1603 affected. 1605 Some documentation may be purely informational, proposing no changes 1606 and proposing usage at most. This includes Section 7.2.3, 1607 Section 7.2.8, Section 7.2.10, and Section 7.2.14. 1609 Section 7.2.9 may require a new protocol. 1611 7.4.1. OSPF-TE and ISIS-TE Protocol Extensions 1613 Many of the changes listed in Section 7.2 require IGP-TE changes, 1614 though most are small extensions to provide additional information. 1615 This set includes Section 7.2.1, Section 7.2.2, Section 7.2.5, 1616 Section 7.2.6, and Section 7.2.7. An adjustment to existing 1617 advertised parameters is suggested in Section 7.2.11. 1619 7.4.2. PW Protocol Extensions 1621 The only suggestion of pseudowire (PW) extensions is in 1622 Section 7.2.13. 1624 7.4.3. LDP Protocol Extensions 1626 Potential LDP extensions are described in Section 7.2.12. 1628 7.4.4. RSVP-TE Protocol Extensions 1630 RSVP-TE protocol extensions are called for in Section 7.2.1, 1631 Section 7.2.5, Section 7.2.7, and Section 7.2.9. 1633 7.4.5. RSVP-TE Path Selection Changes 1635 Section 7.2.3 calls for path selection to be addressed in individual 1636 documents that require change. These changes would include those 1637 proposed in Section 7.2.1, Section 7.2.2, Section 7.2.5, and 1638 Section 7.2.7. 1640 7.4.6. RSVP-TE Admission Control and Preemption 1642 When a change is needed to path selection, a corresponding change is 1643 needed in admission control. The same set of sections applies: 1644 Section 7.2.1, Section 7.2.2, Section 7.2.5, and Section 7.2.7. Some 1645 resource changes such as a link delay change might trigger 1646 preemption. The rules of preemption remain unchanged, still based on 1647 holding priority. 1649 7.4.7. Flow Identification and Traffic Balance 1651 The following describe either the state of the art in flow 1652 identification and traffic balance or propose changes: Section 7.2.4, 1653 Section 7.2.5, Section 7.2.7, and Section 7.2.8. 1655 8. IANA Considerations 1657 This memo includes no request to IANA. 1659 9. Security Considerations 1661 The security considerations for MPLS/GMPLS and for MPLS-TP are 1662 documented in [RFC5920] and [I-D.ietf-mpls-tp-security-framework]. 1664 The types protocol extensions proposed in this framework document 1665 provide additional information about links, forwarding adjacencies, 1666 and LSP requirements. The protocol semantics changes described in 1667 this framework document propose additional LSP constraints applied at 1668 path computation time and at LSP admission at midpoints LSR. The 1669 additional information and constraints provide no additional security 1670 considerations beyond the security considerations already documented 1671 in [RFC5920] and [I-D.ietf-mpls-tp-security-framework]. 1673 10. Acknowledgments 1675 Authors would like to thank Adrian Farrel, Fred Jounay, Yuji Kamite 1676 for his extensive comments and suggestions regarding early versions 1677 of this document, Ron Bonica, Nabil Bitar, Eric Gray, Lou Berger, and 1678 Kireeti Kompella for their reviews of early versions and great 1679 suggestions. 1681 Authors would like to thank Iftekhar Hussain for review and 1682 suggestions regarding recent versions of this document. 1684 In the interest of full disclosure of affiliation and in the interest 1685 of acknowledging sponsorship, past affiliations of authors are noted. 1686 Much of the work done by Ning So occurred while Ning was at Verizon. 1687 Much of the work done by Curtis Villamizar occurred while at 1688 Infinera. Infinera continues to sponsor this work on a consulting 1689 basis. 1691 11. References 1693 11.1. Normative References 1695 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1696 Requirement Levels", BCP 14, RFC 2119, March 1997. 1698 [RFC3209] Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan, V., 1699 and G. Swallow, "RSVP-TE: Extensions to RSVP for LSP 1700 Tunnels", RFC 3209, December 2001. 1702 [RFC3630] Katz, D., Kompella, K., and D. Yeung, "Traffic Engineering 1703 (TE) Extensions to OSPF Version 2", RFC 3630, 1704 September 2003. 1706 [RFC4201] Kompella, K., Rekhter, Y., and L. Berger, "Link Bundling 1707 in MPLS Traffic Engineering (TE)", RFC 4201, October 2005. 1709 [RFC4206] Kompella, K. and Y. Rekhter, "Label Switched Paths (LSP) 1710 Hierarchy with Generalized Multi-Protocol Label Switching 1711 (GMPLS) Traffic Engineering (TE)", RFC 4206, October 2005. 1713 [RFC5036] Andersson, L., Minei, I., and B. Thomas, "LDP 1714 Specification", RFC 5036, October 2007. 1716 [RFC5305] Li, T. and H. Smit, "IS-IS Extensions for Traffic 1717 Engineering", RFC 5305, October 2008. 1719 [RFC5712] Meyer, M. and JP. Vasseur, "MPLS Traffic Engineering Soft 1720 Preemption", RFC 5712, January 2010. 1722 [RFC6107] Shiomoto, K. and A. Farrel, "Procedures for Dynamically 1723 Signaled Hierarchical Label Switched Paths", RFC 6107, 1724 February 2011. 1726 [RFC6374] Frost, D. and S. Bryant, "Packet Loss and Delay 1727 Measurement for MPLS Networks", RFC 6374, September 2011. 1729 [RFC6391] Bryant, S., Filsfils, C., Drafz, U., Kompella, V., Regan, 1730 J., and S. Amante, "Flow-Aware Transport of Pseudowires 1731 over an MPLS Packet Switched Network", RFC 6391, 1732 November 2011. 1734 11.2. Informative References 1736 [DBP] Bertsekas, D., "Dynamic Behavior of Shortest Path Routing 1737 Algorithms for Communication Networks", IEEE Trans. Auto. 1738 Control 1982. 1740 [I-D.ietf-mpls-entropy-label] 1741 Kompella, K., Drake, J., Amante, S., Henderickx, W., and 1742 L. Yong, "The Use of Entropy Labels in MPLS Forwarding", 1743 draft-ietf-mpls-entropy-label-04 (work in progress), 1744 July 2012. 1746 [I-D.ietf-mpls-explicit-resource-control-bundle] 1747 Zamfir, A., Ali, Z., and P. Dimitri, "Component Link 1748 Recording and Resource Control for TE Links", 1749 draft-ietf-mpls-explicit-resource-control-bundle-10 (work 1750 in progress), April 2011. 1752 [I-D.ietf-mpls-tp-security-framework] 1753 Fang, L., Niven-Jenkins, B., Mansfield, S., and R. 1755 Graveman, "MPLS-TP Security Framework", 1756 draft-ietf-mpls-tp-security-framework-04 (work in 1757 progress), July 2012. 1759 [I-D.ietf-rtgwg-cl-requirement] 1760 Villamizar, C., McDysan, D., Ning, S., Malis, A., and L. 1761 Yong, "Requirements for MPLS Over a Composite Link", 1762 draft-ietf-rtgwg-cl-requirement-07 (work in progress), 1763 June 2012. 1765 [I-D.ietf-rtgwg-cl-use-cases] 1766 Ning, S., Malis, A., McDysan, D., Yong, L., and C. 1767 Villamizar, "Composite Link Use Cases and Design 1768 Considerations", draft-ietf-rtgwg-cl-use-cases-01 (work in 1769 progress), August 2012. 1771 [I-D.kompella-mpls-rsvp-ecmp] 1772 Kompella, K., "Multi-path Label Switched Paths Signaled 1773 Using RSVP-TE", draft-kompella-mpls-rsvp-ecmp-01 (work in 1774 progress), October 2011. 1776 [I-D.ospf-cc-stlv] 1777 Osborne, E., "Component and Composite Link Membership in 1778 OSPF", draft-ospf-cc-stlv-00 (work in progress), 1779 August 2011. 1781 [I-D.villamizar-mpls-tp-multipath] 1782 Villamizar, C., "Use of Multipath with MPLS-TP and MPLS", 1783 draft-villamizar-mpls-tp-multipath-02 (work in progress), 1784 February 2012. 1786 [I-D.villamizar-mpls-tp-multipath-te-extn] 1787 Villamizar, C., "Multipath Extensions for MPLS Traffic 1788 Engineering", 1789 draft-villamizar-mpls-tp-multipath-te-extn-01 (work in 1790 progress), February 2012. 1792 [I-D.wang-ccamp-latency-te-metric] 1793 Fu, X., Betts, M., Wang, Q., McDysan, D., and A. Malis, 1794 "GMPLS extensions to communicate latency as a traffic 1795 engineering performance metric", 1796 draft-wang-ccamp-latency-te-metric-03 (work in progress), 1797 March 2011. 1799 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., 1800 and W. Weiss, "An Architecture for Differentiated 1801 Services", RFC 2475, December 1998. 1803 [RFC2991] Thaler, D. and C. Hopps, "Multipath Issues in Unicast and 1804 Multicast Next-Hop Selection", RFC 2991, November 2000. 1806 [RFC2992] Hopps, C., "Analysis of an Equal-Cost Multi-Path 1807 Algorithm", RFC 2992, November 2000. 1809 [RFC3260] Grossman, D., "New Terminology and Clarifications for 1810 Diffserv", RFC 3260, April 2002. 1812 [RFC3468] Andersson, L. and G. Swallow, "The Multiprotocol Label 1813 Switching (MPLS) Working Group decision on MPLS signaling 1814 protocols", RFC 3468, February 2003. 1816 [RFC3945] Mannie, E., "Generalized Multi-Protocol Label Switching 1817 (GMPLS) Architecture", RFC 3945, October 2004. 1819 [RFC3985] Bryant, S. and P. Pate, "Pseudo Wire Emulation Edge-to- 1820 Edge (PWE3) Architecture", RFC 3985, March 2005. 1822 [RFC4385] Bryant, S., Swallow, G., Martini, L., and D. McPherson, 1823 "Pseudowire Emulation Edge-to-Edge (PWE3) Control Word for 1824 Use over an MPLS PSN", RFC 4385, February 2006. 1826 [RFC4655] Farrel, A., Vasseur, J., and J. Ash, "A Path Computation 1827 Element (PCE)-Based Architecture", RFC 4655, August 2006. 1829 [RFC4928] Swallow, G., Bryant, S., and L. Andersson, "Avoiding Equal 1830 Cost Multipath Treatment in MPLS Networks", BCP 128, 1831 RFC 4928, June 2007. 1833 [RFC5151] Farrel, A., Ayyangar, A., and JP. Vasseur, "Inter-Domain 1834 MPLS and GMPLS Traffic Engineering -- Resource Reservation 1835 Protocol-Traffic Engineering (RSVP-TE) Extensions", 1836 RFC 5151, February 2008. 1838 [RFC5152] Vasseur, JP., Ayyangar, A., and R. Zhang, "A Per-Domain 1839 Path Computation Method for Establishing Inter-Domain 1840 Traffic Engineering (TE) Label Switched Paths (LSPs)", 1841 RFC 5152, February 2008. 1843 [RFC5316] Chen, M., Zhang, R., and X. Duan, "ISIS Extensions in 1844 Support of Inter-Autonomous System (AS) MPLS and GMPLS 1845 Traffic Engineering", RFC 5316, December 2008. 1847 [RFC5392] Chen, M., Zhang, R., and X. Duan, "OSPF Extensions in 1848 Support of Inter-Autonomous System (AS) MPLS and GMPLS 1849 Traffic Engineering", RFC 5392, January 2009. 1851 [RFC5441] Vasseur, JP., Zhang, R., Bitar, N., and JL. Le Roux, "A 1852 Backward-Recursive PCE-Based Computation (BRPC) Procedure 1853 to Compute Shortest Constrained Inter-Domain Traffic 1854 Engineering Label Switched Paths", RFC 5441, April 2009. 1856 [RFC5920] Fang, L., "Security Framework for MPLS and GMPLS 1857 Networks", RFC 5920, July 2010. 1859 [RFC5921] Bocci, M., Bryant, S., Frost, D., Levrau, L., and L. 1860 Berger, "A Framework for MPLS in Transport Networks", 1861 RFC 5921, July 2010. 1863 Authors' Addresses 1865 So Ning 1866 Tata Communications 1868 Email: ning.so@tatacommunications.com 1870 Dave McDysan 1871 Verizon 1872 22001 Loudoun County PKWY 1873 Ashburn, VA 20147 1875 Email: dave.mcdysan@verizon.com 1877 Eric Osborne 1878 Cisco 1880 Email: eosborne@cisco.com 1882 Lucy Yong 1883 Huawei USA 1884 5340 Legacy Dr. 1885 Plano, TX 75025 1887 Phone: +1 469-277-5837 1888 Email: lucy.yong@huawei.com 1889 Curtis Villamizar 1890 Outer Cape Cod Network Consulting 1892 Email: curtis@occnc.com