idnits 2.17.1 draft-ietf-tewg-diff-te-mar-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1.a on line 17. -- Found old boilerplate from RFC 3978, Section 5.5 on line 975. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 522. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 537. ** Found boilerplate matching RFC 3978, Section 5.4, paragraph 1 (on line 957), which is fine, but *also* found old RFC 2026, Section 10.4C, paragraph 1 text on line 36. ** The document seems to lack an RFC 3978 Section 5.1 IPR Disclosure Acknowledgement -- however, there's a paragraph with a matching beginning. Boilerplate error? ** This document has an original RFC 3978 Section 5.4 Copyright Line, instead of the newer IETF Trust Copyright according to RFC 4748. ** The document seems to lack an RFC 3978 Section 5.4 Reference to BCP 78 -- however, there's a paragraph with a matching beginning. Boilerplate error? ** This document has an original RFC 3978 Section 5.5 Disclaimer, instead of the newer disclaimer which includes the IETF Trust according to RFC 4748. ** The document seems to lack an RFC 3979 Section 5, para. 2 IPR Disclosure Acknowledgement -- however, there's a paragraph with a matching beginning. Boilerplate error? ** The document uses RFC 3667 boilerplate or RFC 3978-like boilerplate instead of verbatim RFC 3978 boilerplate. After 6 May 2005, submission of drafts without verbatim RFC 3978 boilerplate is not accepted. The following non-3978 patterns matched text found in the document. That text should be removed or replaced: By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- ** The document seems to lack a 1id_guidelines paragraph about Internet-Drafts being working documents. == The page length should not exceed 58 lines per page, but there was 14 longer pages, the longest (page 5) being 61 lines Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the RFC 3978 Section 5.4 Copyright Line does not match the current year -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- Couldn't find a document date in the document -- date freshness check skipped. Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Missing Reference: 'RFC2119' is mentioned on line 75, but not defined == Unused Reference: 'KEY' is defined on line 459, but no explicit reference was found in the text == Unused Reference: 'ASH2' is defined on line 473, but no explicit reference was found in the text == Unused Reference: 'ASH3' is defined on line 476, but no explicit reference was found in the text == Unused Reference: 'RFC2026' is defined on line 508, but no explicit reference was found in the text ** Obsolete normative reference: RFC 2434 (ref. 'IANA-CONS') (Obsoleted by RFC 5226) Summary: 10 errors (**), 0 flaws (~~), 7 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group Jerry Ash 3 Internet Draft AT&T 4 Category: Experimental 5 6 Expiration Date: June 2005 7 December, 2004 9 Max Allocation with Reservation Bandwidth Constraints Model for 10 DiffServ-aware MPLS Traffic Engineering & Performance Comparisons 12 Status of this Memo 14 By submitting this Internet-Draft, each author represents that any 15 applicable patent or other IPR claims of which he or she is aware have 16 been or will be disclosed, and any of which he or she becomes aware will 17 be disclosed, in accordance with Section 6 of RFC 3668. 19 Internet-Drafts are Working documents of the Internet Engineering Task 20 Force (IETF), its areas, and its working groups. Note that other groups 21 may also distribute working documents as Internet-Drafts. 23 Internet-Drafts are draft documents valid for a maximum of six months 24 and may be updated, replaced, or obsoleted by other documents at any 25 time. It is inappropriate to use Internet-Drafts as reference material 26 or to cite them other than as "work in progress." 28 The list of current Internet-Drafts can be accessed at 29 http://www.ietf.org/1id-abstracts.html. 31 The list of Internet-Draft Shadow Directories can be accessed at 32 http://www.ietf.org/shadow.html. 34 Copyright Notice 36 Copyright (C) The Internet Society (2004). All Rights Reserved. 38 Abstract 40 This document complements the DiffServ-aware MPLS TE (DS-TE) 41 requirements document by giving a functional specification for the 42 Maximum Allocation with Reservation (MAR) Bandwidth Constraints Model. 43 Assumptions, applicability, and examples of the operation of the MAR 44 Bandwidth Constraints Model are presented. MAR performance is analyzed 45 relative to the criteria for selecting a Bandwidth Constraints Model, in 46 order to provide guidance to user implementation of the model in their 47 networks. 49 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 51 Table of Contents 53 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 2. Definitions. . . . . . . . . . . . . . . . . . . . . . . . . . 4 55 3. Assumptions & Applicability . . . . . . . . . . . . . . . . . 5 56 4. Functional Specification of the MAR Bandwidth Constraints Model 6 57 5. Setting Bandwidth Constraints . . . . . . . . . . . . . . . . 7 58 6. Example of MAR Operation . . . . . . . . . . . . . . . . . . . 7 59 7. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 60 8. Security Considerations . . . . . . . . . . . . . . . . . . . 9 61 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 9 62 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 63 11. Normative References . . . . . . . . . . . . . . . . . . . . 9 64 12. Informative References . . . . . . . . . . . . . . . . . . . 9 65 13. Intellectual Property Considerations . . . . . . . . . . . . 10 66 14. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . 11 67 Appendix A. MAR Operation & Performance Analysis . . . . . . . . 11 68 Appendix B. Bandwidth Prediction for Path Computation . . . . . . 17 70 Specification of Requirements 72 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 73 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 74 document are to be interpreted as described in [RFC2119]. 76 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 78 1. Introduction 80 DiffServ-aware MPLS traffic engineering (DS-TE) requirements and 81 protocol extensions are specified in [DSTE-REQ, DSTE-PROTO]. A 82 requirement for DS-TE implementation is the specification of Bandwidth 83 Constraints Models for use with DS-TE. The Bandwidth Constraints Model 84 provides the 'rules' to support the allocation of bandwidth to 85 individual class types (CTs). CTs are groupings of service classes in 86 the DS-TE model, which are provided separate bandwidth allocations, 87 priorities, and QoS objectives. Several CTs can share a common 88 bandwidth pool on an integrated, multiservice MPLS/DiffServ network. 90 This document is intended to complement the DS-TE requirements document 91 [DSTE-REQ] by giving a functional specification for the Maximum 92 Allocation with Reservation (MAR) Bandwidth Constraints Model. Examples 93 of the operation of the MAR Bandwidth Constraints Model are presented. 94 MAR performance is analyzed relative to the criteria for selecting a 95 Bandwidth Constraints Model, in order to provide guidance to user 96 implementation of the model in their networks. 98 Two other Bandwidth Constraints Models are being specified for use in 99 DS-TE: 101 1. Maximum Allocation Model (MAM) [MAM] - the maximum allowable 102 bandwidth usage of each CT is explicitly specified. 103 2. Russian Doll Model (RDM) [RDM] - the maximum allowable bandwidth 104 usage is done cumulatively by grouping successive CTs according to 105 priority classes. 107 MAR is similar to MAM in that a maximum bandwidth allocation is given to 108 each CT. However, through the use of bandwidth reservation and 109 protection mechanisms, CTs are allowed to exceed their bandwidth 110 allocations under conditions of no congestion but revert to their 111 allocated bandwidths when overload and congestion occurs. 113 All Bandwidth Constraints Models should meet these objectives: 115 1. applies equally when preemption is either enabled or disabled (when 116 preemption is disabled, the model still works 'reasonably' well), 117 2. bandwidth efficiency, i.e., good bandwidth sharing among CTs under 118 both normal and overload conditions, 119 3. bandwidth isolation, i.e., a CT cannot hog the bandwidth of another 120 CT under overload conditions, 121 4. protection against QoS degradation, at least of the high-priority CTs 122 (e.g. high-priority voice, high-priority data, etc.), and 123 5. reasonably simple, i.e., does not require additional IGP extensions 124 and minimizes signaling load processing requirements. 126 In Appendix A modeling analysis is presented which shows that the MAR 127 Model meets all these objectives, and provides good network performance 128 relative to MAM and full sharing models, under normal and abnormal 129 operating conditions. It is demonstrated that MAR simultaneously 130 achieves bandwidth efficiency, bandwidth isolation, and protection 131 against QoS degradation without preemption. 133 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 135 In Section 3 we give the assumptions and applicability, in Section 4 a 136 functional specification of the MAR Bandwidth Constraints Model, and in 137 Section 5 we give examples of its operation. In Appendix A, MAR 138 performance is analyzed relative to the criteria for selecting a 139 Bandwidth Constraints Model, in order to provide guidance to user 140 implementation of the model in their networks. In Appendix B, 141 bandwidth prediction for path computation is discussed. 143 2. Definitions 145 For readability a number of definitions from [DSTE-REQ, DSTE-PROTO] are 146 repeated here: 148 Traffic Trunk: an aggregation of traffic flows of the same class (i.e. 149 which are to be treated equivalently from the DS-TE perspective) which 150 are placed inside an LSP. 152 Class-Type (CT): the set of Traffic Trunks crossing a link that is 153 governed by a specific set of Bandwidth constraints. CT is used for the 154 purposes of link bandwidth allocation, constraint based routing and 155 admission control. A given Traffic Trunk belongs to the same CT on all 156 links. 158 Up to 8 CTs (MaxCT = 8) are supported. They are referred to as CTc, 159 0 <= c <= MaxCT-1 = 7. Each CT is assigned either a Bandwidth 160 Constraint, or a set of Bandwidth Constraints. Up to 8 Bandwidth 161 Constraints (MaxBC = 8) are supported and they are referred to as BCc, 162 0 <= c <= MaxBC-1 = 7. 164 TE-Class: A pair of: a) a CT, and b) a preemption priority allowed for 165 that CT. This means that an LSP transporting a Traffic Trunk from that 166 CT can use that preemption priority as the set-up priority, as the 167 holding priority or both. 169 MAX_RESERVABLE_BWk: maximum reservable bandwidth on link k specifies the 170 maximum bandwidth that may be reserved; this may be greater than the 171 maximum link bandwidth in which case the link may be oversubscribed 172 [OSPF-TE]. 174 BCck: bandwidth constraint for CTc on link k = allocated (minimum 175 guaranteed) bandwidth for CTc on link k (see Section 4). 177 RBW_THRESk: reservation bandwidth threshold for link k (see Section 4). 179 RESERVED_BWck: reserved bandwidth-in-progress on CTc on link k (0 <= c 180 <= MaxCT-1), RESERVED_BWck = total amount of the bandwidth reserved 181 by all the established LSPs which belong to CTc. 183 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 185 UNRESERVED_BWk: unreserved link bandwidth on link k specifies the 186 amount of bandwidth not yet reserved for any CT, UNRESERVED_BWk = 187 MAX_RESERVABLE_BWk - sum [RESERVED_BWck (0 <= c <= MaxCT-1)]. 188 UNRESERVED_BWck: unreserved link bandwidth on CTc on link k specifies 189 the amount of bandwidth not yet reserved for CTc, UNRESERVED_BWck = 190 UNRESERVED_BWk - delta0/1(CTck) * RBW-THRESk 191 where 192 delta0/1(CTck) = 0 if RESERVED_BWck < BCck 193 delta0/1(CTck) = 1 if RESERVED_BWck >= BCck 195 A number of recovery mechanisms under investigation in the IETF take 196 advantage of the concept of bandwidth sharing across particular sets of 197 LSPs. "Shared Mesh Restoration" in [GMPLS-RECOV] and "Facility-based 198 Computation Model" in [MPLS-BACKUP] are example mechanisms which 199 increase bandwidth efficiency by sharing bandwidth across backup LSPs 200 protecting against independent failures. To ensure that the notion of 201 RESERVED_BWck introduced in [DSTE-REQ] is compatible with such a concept 202 of bandwidth sharing across multiple LSPs, the wording of the definition 203 provided in [DSTE-REQ] is generalized. With this generalization, the 204 definition is compatible with Shared Mesh Restoration defined in 205 [GMPLS-RECOV], so that DS-TE and Shared Mesh Protection can operate 206 simultaneously, under the assumption that Shared Mesh Restoration 207 operates independently within each DS-TE Class-Type and does not operate 208 across Class-Types. For example, backup LSPs protecting primary LSPs of 209 CTc need to also belong to CTc; excess traffic LSPs sharing bandwidth 210 with backup LSPs of CTc need to also belong to CTc. 212 3. Assumptions & Applicability 214 In general, DS-TE is a bandwidth allocation mechanism, for different 215 classes of traffic allocated to various CTs (e.g., voice, normal data, 216 best-effort data). Network operations functions such as capacity 217 design, bandwidth allocation, routing design, and network planning are 218 normally based on traffic measured load and forecast [ASH1]. 220 As such, the following assumptions are made according to the operation 221 of MAR: 223 1. connection admission control (CAC) allocates bandwidth for network 224 flows/LSPs according to the traffic load assigned to each CT, based on 225 traffic measurement and forecast. 226 2. CAC could allocate bandwidth per flow, per LSP, per traffic trunk, or 227 otherwise. That is, no specific assumption is made on a specific CAC 228 method, only that CT bandwidth allocation is related to the 229 measured/forecast traffic load, as per assumption #1. 230 3. CT bandwidth allocation is adjusted up or down according to 231 measured/forecast traffic load. No specific time period is assumed for 232 this adjustment, it could be short term (seconds, minutes, hours), 233 daily, weekly, monthly, or otherwise. 234 4. Capacity management and CT bandwidth allocation thresholds (e.g., 235 BCc) are designed according to traffic load, and are based on traffic 236 measurement and forecast. Again, no specific time period is assumed for 237 this adjustment, it could be short term (hours), daily, weekly, monthly, 238 or otherwise. 239 5. No assumption is made on the order in which traffic is allocated to 240 various CTs, again traffic allocation is assumed to be based only on 241 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 243 traffic load as it is measured and/or forecast. 244 6. If link bandwidth is exhausted on a given path for a flow/LSP/traffic 245 trunk, alternate paths may be attempted to satisfy CT bandwidth 246 allocation. 248 Note that the above assumptions are not unique to MAR, but are generic, 249 common assumptions for all BC Models. 251 4. Functional Specification of the MAR Bandwidth Constraints Model 253 A DS-TE LSR implementing MAR MUST support enforcement of bandwidth 254 constraints in compliance with the specifications in this Section. 256 In the MAR Bandwidth Constraints Model, the bandwidth allocation control 257 for each CT is based on estimated bandwidth needs, bandwidth use, and 258 status of links. The LER makes needed bandwidth allocation changes, and 259 uses [RSVP-TE], for example, to determine if link bandwidth can be 260 allocated to a CT. Bandwidth allocated to individual CTs is protected as 261 needed but otherwise shared. Under normal non-congested network 262 conditions, all CTs/services fully share all available bandwidth. When 263 congestion occurs for a particular CTc, bandwidth reservation acts to 264 prohibit traffic from other CTs from seizing the allocated capacity for 265 CTc. 267 On a given link k, a small amount of bandwidth RBW_THRESk, the 268 reservation bandwidth threshold for link k, is reserved and governs the 269 admission control on link k. Also associated with each CTc on link k 270 are the allocated bandwidth constraints BCck to govern bandwidth 271 allocation and protection. The reservation bandwidth on a link, 272 RBW_THRESk, can be accessed when a given CTc has bandwidth-in-use 273 RESERVED_BWck below its allocated bandwidth constraint BCck. However, 274 if RESERVED_BWck exceeds its allocated bandwidth constraint BCck, then 275 the reservation bandwidth RBW_THRESk cannot be accessed. In this way, 276 bandwidth can be fully shared among CTs if available, but is otherwise 277 protected by bandwidth reservation methods. 279 Bandwidth can be accessed for a bandwidth request = DBW for CTc on a 280 given link k based on the following rules: 282 Table 1: Rules for Admitting LSP Bandwidth Request = DBW on Link k 284 For LSP on a high priority or normal priority CTc: 285 If RESERVED_BWck <= BCc: admit if DBW <= UNRESERVED_BWk 286 If RESERVED_BWck > BCc: admit if DBW <= UNRESERVED_BWk - RBW_THRESk; 287 or, equivalently: 288 If DBW <= UNRESERVED_BWck, admit the LSP. 290 For LSP on a best-effort priority CTc: 291 allocated bandwidth BCc = 0; 292 DiffServ queuing admits BE packets only if there is available link 293 bandwidth. 295 The normal semantics of setup and holding priority are applied in the 296 MAR Bandwidth Constraints Model, and cross-CT preemption is permitted 297 when preemption is enabled. 299 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 301 The bandwidth allocation rules defined in Table 1 are illustrated with 302 an example in Section 6 and simulation analysis in Appendix A. 304 5. Setting Bandwidth Constraints 306 For a normal priority CTc, the bandwidth constraints BCck on link k are 307 set by allocating the maximum reservable bandwidth (MAX_RESERVABLE_BWk) 308 in proportion to the forecast or measured traffic load bandwidth 309 TRAF_LOAD_BWck for CTc on link k. That is: 311 PROPORTIONAL_BWck = TRAF_LOAD_BWck/[sum {TRAF_LOAD_BWck, c=0,MaxCT-1}] X 312 MAX_RESERVABLE_BWk 314 For normal priority CTc: 315 BCck = PROPORTIONAL_BWck 317 For a high priority CT, the bandwidth constraint BCck is set to a 318 multiple of the proportional bandwidth. That is: 320 For high priority CTc: 321 BCck = FACTOR X PROPORTIONAL_BWck 323 where FACTOR is set to a multiple of the proportional bandwidth (e.g., 324 FACTOR = 2 or 3 is typical). This results in some 'over-allocation' 325 of the maximum reservable bandwidth, and gives priority to the high 326 priority CTs. Normally the bandwidth allocated to high priority CTs 327 should be a relatively small fraction of the total link bandwidth, a 328 maximum of 10-15 percent being a reasonable guideline. 330 As stated in Section 4, the bandwidth allocated to a best-effort 331 priority CTc should be set to zero. That is: 333 For best-effort priority CTc: 334 BCck = 0 336 6. Example of MAR Operation 338 In the example, assume there are three class-types: CT0, CT1, CT2. We 339 consider a particular link with 341 MAX-RESERVABLE_BW = 100 343 And with the allocated bandwidth constraints set as follows: 345 BC0 = 30 346 BC1 = 20 347 BC2 = 20 349 These bandwidth constraints are based on the normal traffic loads, as 350 discussed in Section 5. With MAR, any of the CTs is allowed to exceed 351 its bandwidth constraint BCc as long a there is at least RBW_THRES 352 (reservation bandwidth threshold on the link) units of spare bandwidth 353 remaining. Let's assume 355 RBW_THRES = 10 356 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 358 So under overload, if 360 RESERVED_BW0 = 50 361 RESERVED_BW1 = 30 362 RESERVED_BW2 = 10 364 Therefore, for this loading 366 UNRESERVED_BW = 100 - 50 - 30 - 10 = 10 368 CT0 and CT1 can no longer increase their bandwidth on the link, since 369 they are above their BC values and there is only RBW_THRES=10 units of 370 spare bandwidth left on the link. But CT2 can take the additional 371 bandwidth (up to 10 units) if the demand arrives, since it is below its 372 BC value. 374 As also discussed in Section 4, if best effort traffic is present, it 375 can always seize whatever spare bandwidth is available on the link at 376 the moment, but is subject to being lost at the queues in favor of the 377 higher priority traffic. 379 Let's say an LSP arrives for CT0 needing 5 units of bandwidth (i.e., DBW 380 = 5). We need to decide based on Table 1 whether to admit this LSP or 381 not. Since for CT0 383 RESERVED_BW0 > BC0 (50 > 30), and 384 DBW > UNRESERVED_BW - RBW_THRES (i.e., 5 > 10 - 10) 386 Table 1 says the LSP is rejected/blocked. 388 Now let's say an LSP arrives for CT2 needing 5 units of bandwidth (i.e., 389 DBW = 5). We need to decide based on Table 1 whether to admit this 390 LSP or not. Since for CT2 392 RESERVED_BW2 < BC2 (10 < 20), and 393 DBW < UNRESERVED_BW (i.e., 5 < 10) 395 Table 1 says to admit the LSP. 397 Hence, in the above example, in the current state of the link and the 398 current CT loading, CT0 and CT1 can no longer increase their bandwidth 399 on the link, since they are above their BCc values and there is only 400 RBW_THRES=10 units of spare bandwidth left on the link. But CT2 can 401 take the additional bandwidth (up to 10 units) if the demand arrives, 402 since it is below its BCc value. 404 7. Summary 406 The proposed MAR Bandwidth Constraints Model includes the following: 408 1. allocate bandwidth to individual CTs, 409 2. protect allocated bandwidth by bandwidth reservation methods, as 410 needed, but otherwise fully share bandwidth, 411 3. differentiate high-priority, normal-priority, and best-effort 412 priority services, and 413 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 415 4. provide admission control to reject connection requests when needed 416 to meet performance objectives. 418 Modeling results presented in Appendix A show that MAR bandwidth 419 allocation a) achieves greater efficiency in bandwidth sharing while 420 still providing bandwidth isolation and protection against QoS 421 degradation, and b) achieves service differentiation for high-priority, 422 normal-priority, and best-effort priority services. 424 8. Security Considerations 426 Security considerations related to the use of DS-TE are discussed in 427 [DSTE-PROTO]. Those apply independently of the Bandwidth Constraints 428 Model, including MAR specified in this document. 430 9. Acknowledgements 432 DS-TE and Bandwidth Constraints Models have been an active area of 433 discussion in the TEWG. I would like to thank Wai Sum Lai for his 434 support and review of this draft. I also appreciate helpful discussions 435 with Francois Le Faucheur. 437 10. IANA Considerations 439 [DSTE-PROTO] defines a new name space for "Bandwidth Constraints Model 440 Id". The guidelines for allocation of values in that name space are 441 detailed in Section 14 of [DSTE-PROTO]. In accordance with these 442 guidelines, IANA was requested to assign a Bandwidth Constraints Model 443 Id for MAR from the range 0-127 (which is to be managed as per the 444 "Specification Required" policy defined in [IANA-CONS]). 446 Bandwidth Constraints Model Id = TBD was allocated by IANA to MAR. 448 To be removed by the RFC editor at the time of publication 449 We request IANA to assign value 2 for the MAR model. Once the value 450 has been assigned, please replace "TBD" above by the assigned value. 451 453 11. Normative References 455 [DSTE-REQ] Le Faucheur, F., Lai, W., et. al., "Requirements for Support 456 of Diff-Serv-aware MPLS Traffic Engineering," RFC 3564, July 2003. 457 [DSTE-PROTO] Le Faucheur, F., et. al., "Protocol Extensions for Support 458 of Diff-Serv-aware MPLS Traffic Engineering," work in progress. 459 [KEY] Bradner, S., "Key words for Use in RFCs to Indicate Requirement 460 Levels", RFC 2119, March 1997. 461 [IANA-CONS] Narten, T., "Guidelines for Writing an IANA Considerations 462 Section in RFCs," RFC 2434, October 1998. 464 12. Informative References 466 [AKI] Akinpelu, J. M., "The Overload Performance of Engineered Networks 467 with Nonhierarchical & Hierarchical Routing," BSTJ, Vol. 63, 1984. 468 [ASH1] Ash, G. R., "Dynamic Routing in Telecommunications Networks," 469 McGraw-Hill, 1998. 471 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 473 [ASH2] Ash, G. R., et. al., "Routing Evolution in Multiservice 474 Integrated Voice/Data Networks," Proceeding of ITC-16, Edinburgh, June 475 1999. 476 [ASH3] Ash, G. R., "Performance Evaluation of QoS-Routing Methods for 477 IP-Based Multiservice Networks," Computer Communications Magazine, 478 May 2003. 479 [BUR] Burke, P. J., Blocking Probabilities Associated with Directional 480 Reservation, unpublished memorandum, 1961. 481 [DSTE-PERF] Lai, W., "Bandwidth Constraints Models for DiffServ-TE: 482 Performance Evaluation", work in progress. 483 [E.360.1 --> E.360.7] ITU-T Recommendations, "QoS Routing & Related 484 Traffic Engineering Methods for Multiservice TDM-, ATM-, & IP-Based 485 Networks". 486 [GMPLS-RECOV] Lang, J., et. al., "Generalized MPLS Recovery Functional 487 Specification", work in progress. 488 [KRU] Krupp, R. S., "Stabilization of Alternate Routing Networks", 489 Proceedings of ICC, Philadelphia, 1982. 490 [LAI] Lai, W., "Traffic Engineering for MPLS, Internet Performance and 491 Control of Network Systems III Conference", SPIE Proceedings Vol. 4865, 492 pp. 256-267, Boston, Massachusetts, USA, 29 July-1 August 2002 493 (http://www.columbia.edu/~ffl5/waisum/bcmodel.pdf). 494 [MAM] Le Faucheur, F., Lai, W., "Maximum Allocation Bandwidth 495 Constraints Model for Diff-Serv-aware MPLS Traffic Engineering", work in 496 progress. 497 [MPLS-BACKUP] Vasseur, J. P., et. al., "MPLS Traffic Engineering Fast 498 Reroute: Bypass Tunnel Path Computation for Bandwidth Protection", work 499 in progress. 500 [MUM] Mummert, V. S., "Network Management and Its Implementation on the 501 No. 4ESS, International Switching Symposium", Japan, 1976. 502 [NAK] Nakagome, Y., Mori, H., Flexible Routing in the Global 503 Communication Network, Proceedings of ITC-7, Stockholm, 1973. 504 [OSPF-TE] Katz, D., et. al., "Traffic Engineering (TE) Extensions to 505 OSPF Version 2," RFC 3630, September 2003. 506 [RDM] Le Faucheur, F., "Russian Dolls Bandwidth Constraints Model for 507 Diff-Serv-aware MPLS Traffic Engineering", work in progress. 508 [RFC2026] Bradner, S., "The Internet Standards Process -- Revision 3", 509 BCP 9, RFC 2026, October 1996. 510 [RSVP-TE] Awduche, D., et. al., "RSVP-TE: Extensions to RSVP for LSP 511 Tunnels", RFC 3209, December 2001. 513 13. Intellectual Property Considerations 515 The IETF takes no position regarding the validity or scope of any 516 Intellectual Property Rights or other rights that might be claimed to 517 pertain to the implementation or use of the technology described in this 518 document or the extent to which any license under such rights might or 519 might not be available; nor does it represent that it has made any 520 independent effort to identify any such rights. Information on the 521 procedures with respect to rights in RFC documents can be found in BCP 522 78 and BCP 79. 524 Copies of IPR disclosures made to the IETF Secretariat and any 525 assurances of licenses to be made available, or the result of an attempt 526 made to obtain a general license or permission for the use of such 527 proprietary rights by implementers or users of this specification can be 528 obtained from the IETF on-line IPR repository at 529 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 531 http://www.ietf.org/ipr. 533 The IETF invites any interested party to bring to its attention any 534 copyrights, patents or patent applications, or other proprietary rights 535 that may cover technology that may be required to implement this 536 standard. Please address the information to the IETF at 537 ietf-ipr@ietf.org. 539 14. Authors' Addresses 541 Jerry Ash 542 AT&T 543 Room MT D5-2A01 544 200 Laurel Avenue 545 Middletown, NJ 07748, USA 546 Phone: +1 732-420-4578 547 Email: gash@att.com 549 Appendix A. MAR Operation & Performance Analysis 551 A.1 MAR Operation 553 In the MAR Bandwidth Constraints Model, the bandwidth allocation control 554 for each CT is based on estimated bandwidth needs, bandwidth use, and 555 status of links. The LER makes needed bandwidth allocation changes, and 556 uses [RSVP-TE], for example, to determine if link bandwidth can be 557 allocated to a CT. Bandwidth allocated to individual CTs is protected as 558 needed but otherwise shared. Under normal non-congested network 559 conditions, all CTs/services fully share all available bandwidth. When 560 congestion occurs for a particular CTc, bandwidth reservation acts to 561 prohibit traffic from other CTs from seizing the allocated capacity for 562 CTc. Associated with each CT is the allocated bandwidth constraint 563 (BCc) to govern bandwidth allocation and protection, these parameters 564 are illustrated with examples in this Appendix. 566 In performing MAR bandwidth allocation for a given flow/LSP, the LER 567 first determines the egress LSR address, service-identity, and CT. The 568 connection request is allocated an equivalent bandwidth to be routed on 569 a particular CT. The LER then accesses the CT priority, QoS/traffic 570 parameters, and routing table between the LER and egress LSR, and sets 571 up the connection request using the MAR bandwidth allocation rules. The 572 LER selects a first choice path and determines if bandwidth can be 573 allocated on the path based on the MAR bandwidth allocation rules given 574 in Section 4. If the first choice path has insufficient bandwidth, the 575 LER may then try alternate paths, and again applies the MAR bandwidth 576 allocation rules now described. 578 MAR bandwidth allocation is done on a per-CT basis, in which aggregated 579 CT bandwidth is managed to meet the overall bandwidth requirements of CT 580 service needs. Individual flows/LSPs are allocated bandwidth in the 581 corresponding CT according to CT bandwidth availability. A fundamental 582 principle applied in MAR bandwidth allocation methods is the use of 583 bandwidth reservation techniques. 585 Bandwidth reservation gives preference to the preferred traffic by 586 allowing it to seize idle bandwidth on a link more easily than the 587 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 589 non-preferred traffic. Burke [BUR] first analyzed bandwidth reservation 590 behavior from the solution of the birth-death equations for the 591 bandwidth reservation model. Burke's model showed the relative 592 lost-traffic level for preferred traffic, which is not subject to 593 bandwidth reservation restrictions, as compared to non-preferred 594 traffic, which is subject to the restrictions. Bandwidth reservation 595 protection is robust to traffic variations and provides significant 596 dynamic protection of particular streams of traffic. It is widely used 597 in large-scale network applications [ASH1, MUM, AKI, KRU, NAK]. 599 Bandwidth reservation is used in MAR bandwidth allocation to control 600 sharing of link bandwidth across different CTs. On a given link, a 601 small amount of bandwidth RBW_THRES is reserved (say 1% of the total 602 link bandwidth), and the reservation bandwidth can be accessed when a 603 given CT has reserved bandwidth-in-progress RESERVED_BW below its 604 allocated bandwidth BC. That is, if the available link bandwidth 605 (unreserved idle link bandwidth UNRESERVED_BW) exceeds RBW_THRES, then 606 any CT is free to access the available bandwidth on the link. However, 607 if UNRESERVED_BW is less than RBW_THRES, then the CT can utilize the 608 available bandwidth only if its current bandwidth usage is below the 609 allocated amount BC. In this way, bandwidth can be fully shared among 610 CTs if available, but is protected by bandwidth reservation if below the 611 reservation level. 613 Through the bandwidth reservation mechanism, MAR bandwidth allocation 614 also gives preference to high-priority CTs, in comparison to 615 normal-priority and best-effort priority CTs. 617 Hence, bandwidth allocated to each CT is protected by bandwidth 618 reservation methods, as needed, but otherwise shared. Each LER monitors 619 CT bandwidth use on each CT, and determines if connection requests can 620 be allocated to the CT bandwidth. For example, for a bandwidth request 621 of DBW on a given flow/LSP, the LER determines the CT priority (high, 622 normal, or best-effort), CT bandwidth-in-use, and CT bandwidth 623 allocation thresholds, and uses these parameters to determine the 624 allowed load state threshold to which capacity can be allocated. In 625 allocating bandwidth DBW to a CT on given LSP, say A-B-E, each link in 626 the path is checked for available bandwidth in comparison to the allowed 627 load state. If bandwidth is unavailable on any link in path A-B-E, 628 another LSP could by tried, such as A-C-D-E. Hence determination of the 629 link load state is necessary for MAR bandwidth allocation, and two link 630 load states are distinguished: available (non-reserved) bandwidth 631 (ABW_STATE), and reserved-bandwidth (RBW_STATE). Management of CT 632 capacity uses the link state and the allowed load state threshold to 633 determine if a bandwidth allocation request can be accepted on a given 634 CT. 636 A.2 Analysis of MAR Performance 638 In this Appendix, modeling analysis is presented in which MAR bandwidth 639 allocation is shown to provide good network performance relative to full 640 sharing models, under normal and abnormal operating conditions. A 641 large-scale DiffServ-aware MPLS traffic engineering simulation model is 642 used, in which several CTs with different priority classes share the 643 pool of bandwidth on a multiservice, integrated voice/data network. MAR 644 methods have also been analyzed in practice for TDM-based networks 645 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 647 [ASH1], and in modeling studies for IP-based networks [ASH2, ASH3, 648 E.360]. 650 All Bandwidth Constraints Models should meet these objectives: 652 1. applies equally when preemption is either enabled or disabled (when 653 preemption is disabled, the model still works 'reasonably' well), 654 2. bandwidth efficiency, i.e., good bandwidth sharing among CTs under 655 both normal and overload conditions, 656 3. bandwidth isolation, i.e., a CT cannot hog the bandwidth of another 657 CT under overload conditions, 658 4. protection against QoS degradation, at least of the high-priority CTs 659 (e.g. high-priority voice, high-priority data, etc.), and 660 5. reasonably simple, i.e., does not require additional IGP extensions 661 and minimizes signaling load processing requirements. 663 The use of any given Bandwidth Constraints Model has significant impacts 664 on the performance of a network, as explained later. Therefore, the 665 criteria used to select a model need to enable us to evaluate how a 666 particular model delivers its performance, relative to other models. Lai 667 [LAI, DSTE-PERF] has analyzed the MAM and RDM Models and provided 668 valuable insights into the relative performance of these models under 669 various network conditions. 671 In environments where preemption is not used, MAM is attractive because 672 a) it is good at achieving isolation, and b) it achieves reasonable 673 bandwidth efficiency with some QoS degradation of lower classes. When 674 preemption is used, RDM is attractive because it can achieve bandwidth 675 efficiency under normal load. However, RDM cannot provide service 676 isolation under high load or when preemption is not used. 678 Our performance analysis of MAR bandwidth allocation methods is based on 679 a full-scale, 135-node simulation model of a national network together 680 with a multiservice traffic demand model to study various scenarios and 681 tradeoffs [ASH3, E.360]. Three levels of traffic priority - high, 682 normal, and best effort -- are given across 5 CTs: normal priority 683 voice, high priority voice, normal priority data, high priority data, 684 and best effort data. 686 The performance analyses for overloads and failures include a) the MAR 687 Bandwidth Constraints Model, as specified in Section 4, b) the MAM 688 Bandwidth Constraints Model, and c) the No-DSTE Bandwidth Constraints 689 Model. 691 The allocated bandwidth constraints for MAR are as described in Section 692 5: 694 Normal priority CTs: BCck = PROPORTIONAL_BWk, 695 High priority CTs: BCck = FACTOR X PROPORTIONAL_BWk 696 Best-effort priority CTs: BCck = 0 698 In the MAM Bandwidth Constraints Model, the bandwidth constraints for 699 each CT are set to a multiple of the proportional bandwidth allocation: 701 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 703 Normal priority CTs: BCck = FACTOR1 X PROPORTIONAL_BWk, 704 High priority CTs: BCck = FACTOR2 X PROPORTIONAL_BWk 705 Best-effort priority CTs: BCck = 0 707 Simulations show that for MAM, the sum (BCc) should exceed 708 MAX_RESERVABLE_BWk for better efficiency, as follows: 710 1. The normal priority CTs the BCc values need to be over-allocated to 711 get reasonable performance. It was found that over-allocating by 100%, 712 that is, setting FACTOR1 = 2, gave reasonable performance. 713 2. The high priority CTs can be over-allocated by a larger multiple 714 FACTOR2 in MAM and this gives better performance. 716 The rather large amount of over-allocation improves efficiency but 717 somewhat defeats the 'bandwidth protection/isolation' needed with a BC 718 Model, since one CT can now invade the bandwidth allocated to another 719 CT. Each CT is restricted to its allocated bandwidth constraint BCck, 720 which is the maximum level of bandwidth allocated to each CT on each 721 link, as in normal operation of MAM. 723 In the No-DSTE Bandwidth Constraints Model, no reservation or protection 724 of CT bandwidth is applied, and bandwidth allocation requests are 725 admitted if bandwidth is available. Furthermore, no queuing priority 726 is applied to any of the CTs in the No-DSTE Bandwidth Constraints Model. 728 Table 2 gives performance results for a six-times overload on a single 729 network node at Oakbrook IL. The numbers given in the table are the 730 total network percent lost (blocked) or delayed traffic. Note that in 731 the focused overload scenario studied here, the percent lost/delayed 732 traffic on the Oakbrook node is much higher than the network-wide 733 average values given. 735 Table 2 736 Performance Comparison for MAR, MAM, & No-DSTE 737 Bandwidth Constraints (BC) Models 738 6X Focused Overload on Oakbrook (Total Network % Lost/Delayed Traffic) 740 Class Type MAR BC MAM BC No-DSTE BC 741 Model Model Model 742 NORMAL PRIORITY VOICE 0.00 1.97 10.30 743 HIGH PRIORITY VOICE 0.00 0.00 7.05 744 NORMAL PRIORITY DATA 0.00 6.63 13.30 745 HIGH PRIORITY DATA 0.00 0.00 7.05 746 BEST EFFORT PRIORITY DATA 12.33 11.92 9.65 748 Clearly the performance is better with MAR bandwidth allocation, and the 749 results show that performance improves when bandwidth reservation is 750 used. The reason for the poor performance of the No-DSTE Model, without 751 bandwidth reservation, is due to the lack of protection of allocated 752 bandwidth. If we add the bandwidth reservation mechanism, then 753 performance of the network is greatly improved. 755 The simulations showed that the performance of MAM is quite sensitive to 756 the over-allocation factors discussed above. For example, if the BCc 757 values are proportionally allocated with FACTOR1 = 1, then the results 758 are much worse, as shown in Table 3: 760 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 762 Table 3 763 Performance Comparison for MAM Bandwidth Constraints Model 764 with Different Over-allocation Factors 765 6X Focused Overload on Oakbrook (Total Network % Lost/Delayed Traffic) 767 Class Type (FACTOR1 = 1) (FACTOR1 = 2) 768 NORMAL PRIORITY VOICE 31.69 1.97 769 HIGH PRIORITY VOICE 0.00 0.00 770 NORMAL PRIORITY DATA 31.22 6.63 771 HIGH PRIORITY DATA 0.00 0.00 772 BEST EFFORT PRIORITY DATA 8.76 11.92 774 Table 4 illustrates the performance of the MAR, MAM, and No-DSTE 775 Bandwidth Constraints Models for a high-day network load pattern with a 776 50% general overload. The numbers given in the table are the total 777 network percent lost (blocked) or delayed traffic. 779 Table 4 780 Performance Comparison for MAR, MAM, & No-DSTE 781 Bandwidth Constraints (BC) Models 782 50% General Overload (Total Network % Lost/Delayed Traffic) 784 Class Type MAR BC MAM BC No-DSTE BC 785 Model Model Model 786 NORMAL PRIORITY VOICE 0.02 0.13 7.98 787 HIGH PRIORITY VOICE 0.00 0.00 8.94 788 NORMAL PRIORITY DATA 0.00 0.26 6.93 789 HIGH PRIORITY DATA 0.00 0.00 8.94 790 BEST EFFORT PRIORITY DATA 10.41 10.39 8.40 792 Again, we can see the performance is always better when MAR bandwidth 793 allocation and reservation is used. 795 Table 5 illustrates the performance of the MAR, MAM, and No-DSTE 796 Bandwidth Constraints Models for a single link failure scenario (3 797 OC-48). The numbers given in the table are the total network percent 798 lost (blocked) or delayed traffic. 800 Table 5 801 Performance Comparison for MAR, MAM, & No-DSTE 802 Bandwidth Constraints (BC) Models 803 Single Link Failure (2 OC-48) 804 (Total Network % Lost/Delayed Traffic) 806 Class Type MAR BC MAM BC No-DSTE BC 807 Model Model Model 808 NORMAL PRIORITY VOICE 0.00 0.62 0.63 809 HIGH PRIORITY VOICE 0.00 0.31 0.32 810 NORMAL PRIORITY DATA 0.00 0.48 0.50 811 HIGH PRIORITY DATA 0.00 0.31 0.32 812 BEST EFFORT PRIORITY DATA 0.12 0.72 0.63 814 Again, we can see the performance is always better when MAR bandwidth 815 allocation and reservation is used. 817 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 819 Table 6 illustrates the performance of the MAR, MAM, and No-DSTE 820 Bandwidth Constraints Models for a multiple link failure scenario (3 821 links with 3 OC-48, 3 OC-3, 4 OC-3 capacity, respectively). The numbers 822 given in the table are the total network percent lost (blocked) or 823 delayed traffic. 825 Table 6 826 Performance Comparison for MAR, MAM, & No-DSTE 827 Bandwidth Constraints (BC) Models 828 Multiple Link Failure 829 (3 Links with 2 OC-48, 2 OC-12, 1 OC-12, Respectively) 830 (Total Network % Lost/Delayed Traffic) 832 Class Type MAR BC MAM BC No-DSTE BC 833 Model Model Model 834 NORMAL PRIORITY VOICE 0.00 0.91 0.92 835 HIGH PRIORITY VOICE 0.00 0.44 0.44 836 NORMAL PRIORITY DATA 0.00 0.70 0.72 837 HIGH PRIORITY DATA 0.00 0.44 0.44 838 BEST EFFORT PRIORITY DATA 0.14 1.03 1.04 840 Again, we can see the performance is always better when MAR bandwidth 841 allocation and reservation is used. 843 Lai's results [LAI, DSTE-PERF] show the trade-off between bandwidth 844 sharing and service protection/isolation, using an analytic model of a 845 single link. He shows that RDM has a higher degree of sharing than MAM. 846 Furthermore, for a single link, the overall loss probability is the 847 smallest under full sharing and largest under MAM, with RDM being 848 intermediate. Hence, on a single link, Lai shows that the full sharing 849 model yields the highest link efficiency and MAM the lowest, and that 850 full sharing has the poorest service protection capability. 852 The results of the present study show that when considering a network 853 context, in which there are many links and multiple-link routing paths 854 are used, full sharing does not necessarily lead to maximum network-wide 855 bandwidth efficiency. In fact, the results in Table 4 show that the 856 No-DSTE Model not only degrades total network throughput, but also 857 degrades the performance of every CT that should be protected. Allowing 859 more bandwidth sharing may improve performance up to a point, but can 860 severely degrade performance if care is not taken to protect allocated 861 bandwidth under congestion. 863 Both Lai's study and this study show that increasing the degree of 864 bandwidth sharing among the different CTs leads to a tighter coupling 865 between CTs. Under normal loading conditions, there is adequate capacity 866 for each CT, which minimizes the effect of such coupling. Under overload 867 conditions, when there is a scarcity of capacity, such coupling can 868 cause severe degradation of service, especially for the lower priority 869 CTs. 871 Thus, the objective of maximizing efficient bandwidth usage, as stated 872 in Bandwidth Constraints Model objectives, needs to be exercised with 873 care. Due consideration needs to be given also to achieving bandwidth 874 isolation under overload, in order to minimize the effect of 875 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 877 interactions among the different CTs. The proper tradeoff of bandwidth 878 sharing and bandwidth isolation needs to be achieved in the selection of 879 a Bandwidth Constraints Model. Bandwidth reservation supports greater 880 efficiency in bandwidth sharing while still providing bandwidth 881 isolation and protection against QoS degradation. 883 In summary, the proposed MAR Bandwidth Constraints Model includes the 884 following: a) allocate bandwidth to individual CTs, b) protect allocated 885 bandwidth by bandwidth reservation methods, as needed, but otherwise 886 fully share bandwidth, c) differentiate high-priority, normal-priority, 887 and best-effort priority services, and d) provide admission control to 888 reject connection requests when needed to meet performance objectives. 890 In the modeling results, the MAR Bandwidth Constraints Model compares 891 favorably with methods that do not use bandwidth reservation. In 892 particular, some of the conclusions from the modeling are as follows: 894 o MAR bandwidth allocation is effective in improving performance over 895 methods that lack bandwidth reservation and that allow more bandwidth 896 sharing under congestion, 897 o MAR achieves service differentiation for high-priority, 898 normal-priority, and best-effort priority services, 899 o bandwidth reservation supports greater efficiency in bandwidth sharing 900 while still providing bandwidth isolation and protection against QoS 901 degradation, and is critical to stable and efficient network 902 performance. 904 Appendix B. Bandwidth Prediction for Path Computation 906 As discussed in [DSTE-PROTO], there there are potential advantages for a 907 Head-end in trying to predict the impact of an LSP on the unreserved 908 bandwidth when computing the path for the LSP. One example would be to 909 perform better load-distribution of multiple LSPs across multiple 910 paths. Another example would be to avoid CAC rejection when the LSP 911 would no longer fit on a link after establishment. 913 Where such predictions are used on Head-ends, the optional Bandwidth 914 Constraints sub-TLV and the optional Maximum Reservable Bandwidth 915 sub-TLV MAY be advertised in the IGP. This can be used by Head-ends 916 to predict how an LSP affects unreserved bandwidth values. Such 917 predictions can be made with MAR by using the unreserved bandwidth 918 values advertised by the IGP, as discussed in Sections 2 and 4: 920 UNRESERVED_BWck = MAX_RESERVABLE_BWk - UNRESERVED_BWk - 921 delta0/1(CTck) * RBW-THRESk 923 where 925 delta0/1(CTck) = 0 if RESERVED_BWck < BCck 926 delta0/1(CTck) = 1 if RESERVED_BWck >= BCck 928 Furthermore, the following estimate can be made for RBW_THRESk: 930 RBW_THRESk = RBW_% * MAX_RESERVABLE_BWk, 931 Internet Draft MAR Bandwidth Constraints Model for DS-TE December 04 933 where RBW_% is a locally configured variable, which could take on 934 different values for different link speeds. This information 935 could be used in conjunction with the BC sub-TLV, 936 MAX_RESERVABLE_BW sub-TLV, and UNRESERVED_BW sub-TLV to make 937 predictions of available bandwidth on each link for each CT. 938 Since admission control algorithms are left for vendor differentiation, 939 predictions can only be performed effectively when the Head-end LSR 940 predictions are based on the same (or a very close) admission control 941 algorithm as used by other LSRs. 943 There may be occasional rejected LSPs when head-ends are establishing 944 LSPs through a common link. As an example, consider some link L, and 945 two head-ends H1 and H2. If only H1 or only H2 is establishing LSPs 946 through L, then the prediction is accurate. But, if both H1 and H2 are 947 establishing LSPs through L at the same time, then the prediction 948 would not work perfectly. That is, the CAC will occasionally run into a 949 rejected LSP on a link with such 'race' conditions. Also, as mentioned 950 in Appendix A, such prediction is optional and outside the scope of the 951 document. 953 Full Copyright Statement 955 Copyright (C) The Internet Society (2004). This document is subject to 956 the rights, licenses and restrictions contained in BCP 78 and except as 957 set forth therein, the authors retain all their rights. 959 This document and the information contained herein are provided on an 960 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR 961 IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET 962 ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, 963 INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE 964 INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 965 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 967 Disclaimer of Validity 969 This document and the information contained herein are provided on an 970 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR 971 IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET 972 ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, 973 INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE 974 INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 975 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.