idnits 2.17.1 draft-ietf-tsvwg-sctp-failover-14.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 (December 9, 2015) is 3033 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 4960 (Obsoleted by RFC 9260) Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group Y. Nishida 3 Internet-Draft GE Global Research 4 Intended status: Standards Track P. Natarajan 5 Expires: June 11, 2016 Cisco Systems 6 A. Caro 7 BBN Technologies 8 P. Amer 9 University of Delaware 10 K. Nielsen 11 Ericsson 12 December 9, 2015 14 SCTP-PF: Quick Failover Algorithm in SCTP 15 draft-ietf-tsvwg-sctp-failover-14.txt 17 Abstract 19 SCTP supports multi-homing. However, when the failover operation 20 specified in RFC4960 is followed, there can be significant delay and 21 performance degradation in the data transfer path failover. To 22 overcome this problem this document specifies a quick failover 23 algorithm (SCTP-PF) based on the introduction of a Potentially Failed 24 (PF) state in SCTP Path Management. 26 The document also specifies a dormant state operation of SCTP. This 27 dormant state operation is required to be followed by an SCTP-PF 28 implementation, but it may equally well be applied by a standard 29 RFC4960 SCTP implementation. 31 Additionally, the document introduces an alternative switchback 32 operation mode called Primary Path Switchover that will be beneficial 33 in certain situations. This mode of operation applies to both a 34 standard RFC4960 SCTP implementation as well as to a SCTP-PF 35 implementation. 37 The procedures defined in the document require only minimal 38 modifications to the RFC4960 specification. The procedures are 39 sender-side only and do not impact the SCTP receiver. 41 Status of This Memo 43 This Internet-Draft is submitted in full conformance with the 44 provisions of BCP 78 and BCP 79. 46 Internet-Drafts are working documents of the Internet Engineering 47 Task Force (IETF). Note that other groups may also distribute 48 working documents as Internet-Drafts. The list of current Internet- 49 Drafts is at http://datatracker.ietf.org/drafts/current/. 51 Internet-Drafts are draft documents valid for a maximum of six months 52 and may be updated, replaced, or obsoleted by other documents at any 53 time. It is inappropriate to use Internet-Drafts as reference 54 material or to cite them other than as "work in progress." 56 This Internet-Draft will expire on June 11, 2016. 58 Copyright Notice 60 Copyright (c) 2015 IETF Trust and the persons identified as the 61 document authors. All rights reserved. 63 This document is subject to BCP 78 and the IETF Trust's Legal 64 Provisions Relating to IETF Documents 65 (http://trustee.ietf.org/license-info) in effect on the date of 66 publication of this document. Please review these documents 67 carefully, as they describe your rights and restrictions with respect 68 to this document. Code Components extracted from this document must 69 include Simplified BSD License text as described in Section 4.e of 70 the Trust Legal Provisions and are provided without warranty as 71 described in the Simplified BSD License. 73 Table of Contents 75 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 76 2. Conventions and Terminology . . . . . . . . . . . . . . . . . 4 77 3. SCTP with Potentially Failed Destination State (SCTP-PF) . . 4 78 3.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 4 79 3.2. Specification of the SCTP-PF Procedures . . . . . . . . . 5 80 4. Dormant State Operation . . . . . . . . . . . . . . . . . . . 9 81 4.1. SCTP Dormant State Procedure . . . . . . . . . . . . . . 10 82 5. Primary Path Switchover . . . . . . . . . . . . . . . . . . . 11 83 6. Suggested SCTP Protocol Parameter Values . . . . . . . . . . 12 84 7. Socket API Considerations . . . . . . . . . . . . . . . . . . 12 85 7.1. Support for the Potentially Failed Path State . . . . . . 13 86 7.2. Peer Address Thresholds (SCTP_PEER_ADDR_THLDS) Socket 87 Option . . . . . . . . . . . . . . . . . . . . . . . . . 14 88 7.3. Exposing the Potentially Failed Path State 89 (SCTP_EXPOSE_POTENTIALLY_FAILED_STATE) Socket Option . . 15 90 8. Security Considerations . . . . . . . . . . . . . . . . . . . 16 91 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 16 92 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 16 93 11. Proposed Change of Status (to be Deleted before Publication) 17 94 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 17 95 12.1. Normative References . . . . . . . . . . . . . . . . . . 17 96 12.2. Informative References . . . . . . . . . . . . . . . . . 17 97 Appendix A. Discussions of Alternative Approaches . . . . . . . 18 98 A.1. Reduce Path.Max.Retrans (PMR) . . . . . . . . . . . . . . 18 99 A.2. Adjust RTO related parameters . . . . . . . . . . . . . . 19 100 Appendix B. Discussions for Path Bouncing Effect . . . . . . . . 20 101 Appendix C. SCTP-PF for SCTP Single-homed Operation . . . . . . 20 102 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 21 104 1. Introduction 106 The Stream Control Transmission Protocol (SCTP) specified in 107 [RFC4960] supports multi-homing at the transport layer. SCTP's 108 multi-homing features include failure detection and failover 109 procedures to provide network interface redundancy and improved end- 110 to-end fault tolerance. In SCTP's current failure detection 111 procedure, the sender must experience Path.Max.Retrans (PMR) number 112 of consecutive failed timer-based retransmissions on a destination 113 address before detecting a path failure. Until detecting the path 114 failure, the sender continues to transmit data on the failed path. 115 The prolonged time in which [RFC4960] SCTP continues to use a failed 116 path severely degrades the performance of the protocol. To address 117 this problem, this document specifies a quick failover algorithm 118 (SCTP-PF) based on the introduction of a new Potentially Failed (PF) 119 path state in SCTP path management. The performance deficiencies of 120 the [RFC4960] failover operation, and the improvements obtainable 121 from the introduction of a Potentially Failed state in SCTP, were 122 proposed and documented in [NATARAJAN09] for Concurrent Multipath 123 Transfer SCTP [IYENGAR06]. 125 While SCTP-PF can accelerate failover process and improve 126 performance, the risks that an SCTP endpoint enters the dormant state 127 where all destination addresses are inactive can be increased. 128 [RFC4960] leaves the protocol operation during dormant state to 129 implementations and encourages to avoid entering the state as much as 130 possible by careful tuning of the Path.Max.Retrans (PMR) and 131 Association.Max.Retrans (AMR) parameters. We specify a dormant state 132 operation for SCTP-PF which makes SCTP-PF provide the same disruption 133 tolerance as [RFC4960] despite that the dormant state may be entered 134 more quickly. The dormant state operation may equally well be 135 applied by an [RFC4960] implementation and will here serve to provide 136 added fault tolerance for situations where the tuning of the 137 Path.Max.Retrans (PMR) and Association.Max.Retrans (AMR) parameters 138 fail to provide adequate prevention of the entering of the dormant 139 state. 141 The operation after the recovery of a failed path also impacts the 142 performance of the protocol. With the procedures specified in 143 [RFC4960] SCTP will, after a failover from the primary path, switch 144 back to use the primary path for data transfer as soon as this path 145 becomes available again. From a performance perspective such a 146 forced switchback of the data transmission path can be suboptimal as 147 the CWND towards the original primary destination address has to be 148 rebuilt once data transfer resumes, [CARO02]. As an optional 149 alternative to the switchback operation of [RFC4960], this document 150 specifies an alternative Primary Path Switchover procedure which 151 avoid such forced switchbacks of the data transfer path. The Primary 152 Path Switchover operation was originally proposed in [CARO02]. 154 While SCTP-PF primarily is motivated by a desire to improve the 155 multi-homed operation, the feature applies also to SCTP single-homed 156 operation. Here the algorithm serves to provide increased failure 157 detection on idle associations, whereas the failover or switchback 158 aspects of the algorithm will not be activated. This is discussed in 159 more detail in Appendix C. 161 A brief description of the motivation for the introduction of the 162 Potentially Failed state including a discussion of alternative 163 approaches to mitigate the deficiencies of the [RFC4960] failover 164 operation are given in the Appendices. Discussion of path bouncing 165 effects that might be caused by frequent switchovers, are also 166 provided there. 168 2. Conventions and Terminology 170 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 171 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 172 document are to be interpreted as described in [RFC2119]. 174 3. SCTP with Potentially Failed Destination State (SCTP-PF) 176 3.1. Overview 178 To minimize the performance impact during failover, the sender should 179 avoid transmitting data to a failed destination address as early as 180 possible. In the [RFC4960] SCTP path management scheme, the sender 181 stops transmitting data to a destination address only after the 182 destination address is marked inactive. This process takes a 183 significant amount of time as it requires the error counter of the 184 destination address to exceed the Path.Max.Retrans (PMR) threshold. 185 The issue cannot simply be mitigated by lowering of the PMR threshold 186 because this may result in spurious failure detection and unnecessary 187 prevention of the usage of a preferred primary path. Also due to the 188 coupled tuning of the Path.Max.Retrans (PMR) and the 189 Association.Max.Retrans (AMR) parameter values in [RFC4960], lowering 190 of the PMR threshold may result in lowering of the AMR threshold, 191 which may result in compromisation of the fault tolerance of SCTP. 193 The solution provided in this document is to extend the SCTP path 194 management scheme of [RFC4960] by the addition of the Potentially 195 Failed (PF) state as an intermediate state in between the active and 196 inactive state of a destination address in the [RFC4960] path 197 management scheme, and let the failover of data transfer away from a 198 destination address be driven by the entering of the PF state instead 199 of by the entering of the inactive state. Thereby SCTP may perform 200 quick failover without compromising the overall fault tolerance of 201 [RFC4960] SCTP. At the same time, RTO-based HEARTBEAT probing is 202 initiated towards a destination address once it enters PF state. 203 Thereby SCTP may quickly ascertain whether network connectivity 204 towards the destination address is broken or whether the failover was 205 spurious. In the case where the failover was spurious data transfer 206 may quickly resume towards the original destination address. 208 The new failure detection algorithm assumes that loss detected by a 209 timeout implies either severe congestion or network connectivity 210 failure. It recommends that by default a destination address is 211 classified as PF at the occurrence of the first timeout. 213 3.2. Specification of the SCTP-PF Procedures 215 The SCTP-PF operation is specified as follows: 217 1. The sender maintains a new tunable SCTP Protocol Parameter 218 called PotentiallyFailed.Max.Retrans (PFMR). The PFMR defines 219 the new intermediate PF threshold on the destination address 220 error counter at exceed of which the destination address is 221 classified as PF. The RECOMMENDED value of PFMR is 0, but other 222 values MAY be used. If PFMR is set to be greater than or equal 223 to Path.Max.Retrans (PMR), the resulting PF threshold will be so 224 high that the destination address will reach the inactive state 225 before it can be classified as PF. 227 2. The error counter of an active destination address is 228 incremented as specified in [RFC4960]. This means that the 229 error counter of the destination address will be incremented 230 each time the T3-rtx timer expires, or each time a HEARTBEAT 231 chunk is sent when idle and not acknowledged within an RTO. 232 When the value in the destination address error counter exceeds 233 PFMR, the endpoint MUST mark the destination address as in the 234 PF state. 236 3. A SCTP-PF sender SHOULD NOT send data to destination addresses 237 in PF state when alternative destination addresses in active 238 state are available. Specifically this means that: 240 i When there is outbound data to send and the destination 241 address presently used for data transmission is in PF state, 242 the sender SHOULD choose a destination address in active 243 state, if one exists, and failover to deploy this destination 244 address for data transmission. 246 ii When retransmitting data that has timed out and the sender 247 thus by [RFC4960], section 6.4.1, should attempt to pick a 248 new destination address for data retransmission, the sender 249 SHOULD choose an alternate destination transport address in 250 active state if one exists. 252 iii When there is outbound data to send and the SCTP user 253 explicitly requests to send data to a destination address in 254 PF state, the sender SHOULD send the data to an alternate 255 destination address in active state if one exists. 257 When choosing among multiple destination addresses in active 258 state the following considerations are given: 260 A. An SCTP sender should comply with [RFC4960], section 6.4.1, 261 principles of choosing most divergent source-destination 262 pairs compared with, for i.: the destination address in PF 263 state that it performs a failover from, and for ii.: the 264 destination address towards which the data timed out. Rules 265 for picking the most divergent source-destination pair are 266 an implementation decision and are not specified within this 267 document. 269 B. A SCTP-PF sender MAY choose to send data to a destination 270 address in PF state, even if destination addresses in active 271 state exist, have the SCTP-PF sender other means of 272 information available that disqualifies the destination 273 address in active state from being preferred. However, the 274 discussion of such mechanisms is outside of the scope of the 275 SCTP-PF operation specified in this document. 277 In all cases, the sender MUST NOT change the state of chosen 278 destination address, whether this state be active or PF, and it 279 MUST NOT clear the error counter of the destination address as a 280 result of choosing the destination address for data 281 transmission. 283 4. When the destination addresses are all in PF state or some in PF 284 state and some in inactive state, the sender MUST choose one 285 destination address in PF state and transmit or retransmit data 286 to this destination address using the following rules: 288 A. The sender SHOULD choose the destination in PF state with 289 the lowest error count (fewest consecutive timeouts) for 290 data transmission and transmit or retransmit data to this 291 destination. 293 B. When there are multiple destination addresses in PF state 294 with same error count, the sender should let the choice 295 among the multiple destination addresses in PF state with 296 equal error count be based on the [RFC4960], section 6.4.1, 297 principles of choosing most divergent source-destination 298 pairs when executing (potentially consecutive) 299 retransmission. Rules for picking the most divergent 300 source-destination pair are an implementation decision and 301 are not specified within this document. 303 C. A sender MAY choose to deploy other strategies than the 304 above when choosing among multiple destinations in PF state 305 have the SCTP-PF sender other means of information available 306 that qualifies a particular destination address for being 307 used. The SCTP-PF protocol operation specified in this 308 document makes no assumption of the existence of such other 309 means of information and specifies for the above as the 310 default operation of an SCTP-PF sender. 312 The sender MUST NOT change the state and the error counter of 313 any destination address regardless of whether it has been chosen 314 for transmission or not. 316 5. The HB.interval of the Path Heartbeat function of [RFC4960] MUST 317 be ignored for destination addresses in PF state. Instead 318 HEARTBEAT chunks are sent to destination addresses in PF state 319 once per RTO. HEARTBEAT chunks SHOULD be sent to destination 320 addresses in PF state, but the sending of HEARTBEATS MUST honor 321 whether the Path Heartbeat function (Section 8.3 of [RFC4960]) 322 is enabled for the destination address or not. I.e., if the 323 Path Heartbeat function is disabled for the destination address 324 in question, HEARTBEATS MUST NOT be sent. Note that when 325 Heartbeat function is disabled, it may take longer to transition 326 a destination address in PF state back to active state. 328 6. HEARTBEATs are sent when a destination address reaches the PF 329 state. When a HEARTBEAT chunk is not acknowledged within the 330 RTO, the sender increments the error counter and exponentially 331 backs off the RTO value. If the error counter is less than PMR, 332 the sender transmits another packet containing the HEARTBEAT 333 chunk immediately after timeout expiration on the previous 334 HEARTBEAT. When data is being transmitted to a destination 335 address in the PF state, the transmission of a HEARTBEAT chunk 336 MAY be omitted in case where the receipt of a SACK of the data 337 or a T3-rtx timer expiration on the data can provide equivalent 338 information, such as the case where the data chunk has been 339 transmitted to a single destination address only. Likewise, the 340 timeout of a HEARTBEAT chunk MAY be ignored if data is 341 outstanding towards the destination address. 343 7. When the sender receives a HEARTBEAT ACK from a HEARTBEAT sent 344 to a destination address in PF state, the sender SHOULD clear 345 the error counter of the destination address and transition the 346 destination address back to active state. When the sender 347 resumes data transmission on a destination address after a 348 transition of the destination address from PF to active state, 349 it MUST do this following the prescriptions of Section 7.2 of 350 [RFC4960]. In a situation where a HEARTBEAT ACK arrives while 351 there is data outstanding towards the destination address to 352 which the HEARTBEAT was sent, then an implementation MAY choose 353 to not have the HEARTBEAT ACK reset the error counter, but have 354 the error counter reset await the fate of the outstanding data 355 transmission. This situation can happen when data is sent to a 356 destination address in PF state. 358 8. Additional (PMR - PFMR) consecutive timeouts on a destination 359 address in PF state confirm the path failure, upon which the 360 destination address transitions to the inactive state. As 361 described in [RFC4960], the sender (i) SHOULD notify the ULP 362 about this state transition, and (ii) transmit HEARTBEAT chunks 363 to the inactive destination address at a lower HB.interval 364 frequency as described in Section 8.3 of [RFC4960] (when the 365 Path Heartbeat function is enabled for the destination address). 367 9. Acknowledgments for chunks that have been transmitted to 368 multiple destinations (i.e., a chunk which has been 369 retransmitted to a different destination address than the 370 destination address to which the chunk was first transmitted) 371 SHOULD NOT clear the error count for an inactive destination 372 address and SHOULD NOT transition a destination address in PF 373 state back to active state, since a sender cannot disambiguate 374 whether the ACK was for the original transmission or the 375 retransmission(s). A SCTP sender MAY apply a different approach 376 for the error count handling based on unequivocally information 377 on which destination (including multiple destination addresses) 378 the chunk reached. This document makes no reference to what 379 such unequivocally information could consist of, nor how such 380 unequivocally information could be obtained. The design of such 381 an alternative approach is left to implementations. 383 10. Acknowledgments for data chunks that has been transmitted to one 384 destination address only MUST clear the error counter for the 385 destination address and MUST transition a destination address in 386 PF state back to active state. This situation can happen when 387 new data is sent to a destination address in the PF state. It 388 can also happen in situations where the destination address is 389 in the PF state due to the occurrence of a spurious T3-rtx timer 390 and acknowledgments start to arrive for data sent prior to 391 occurrence of the spurious T3-rtx and data has not yet been 392 retransmitted towards other destinations. This document does 393 not specify special handling for detection of or reaction to 394 spurious T3-rtx timeouts, e.g., for special operation vis-a-vis 395 the congestion control handling or data retransmission operation 396 towards a destination address which undergoes a transition from 397 active to PF to active state due to a spurious T3-rtx timeout. 398 But it is noted that this is an area which would benefit from 399 additional attention, experimentation and specification for 400 single-homed SCTP as well as for multi-homed SCTP protocol 401 operation. 403 11. When all destination addresses are in inactive state, and SCTP 404 protocol operation thus is said to be in dormant state, the 405 prescriptions given in Section 4 shall be followed. 407 12. The SCTP stack SHOULD expose the PF state of its destination 408 addresses to the ULP as well as provide the means to notify the 409 ULP of state transitions of its destination addresses from 410 active to PF, and vice-versa. However it is recommended that an 411 SCTP stack implementing SCTP-PF also allows for that the ULP is 412 kept ignorant of the PF state of its destinations and the 413 associated state transitions, thus allowing for retain of the 414 simpler state transition model of RFC4960 in the ULP. For this 415 reason it is recommended that an SCTP stack implementing SCTP-PF 416 also provides the ULP with the means to suppress exposure of the 417 PF state and the associated state transitions. 419 4. Dormant State Operation 421 In a situation with complete disruption of the communication in 422 between the SCTP Endpoints, the aggressive HEARTBEAT transmissions of 423 SCTP-PF on destination addresses in PF state may make the association 424 enter dormant state faster than a standard [RFC4960] SCTP 425 implementation given the same setting of Path.Max.Retrans (PMR) and 426 Association.Max.Retrans (AMR). For example, an SCTP association with 427 two destination addresses typically would reach dormant state in half 428 the time of an [RFC4960] SCTP implementation in such situations. 429 This is because a SCTP PF sender will send HEARTBEATS and data 430 retransmissions in parallel with RTO intervals when there are 431 multiple destinations addresses in PF state. This argument presumes 432 that RTO << HB.interval of [RFC4960]. With the design goal that 433 SCTP-PF shall provide the same level of disruption tolerance as an 434 [RFC4960] SCTP implementation with the same Path.Max.Retrans (PMR) 435 and Association.Max.Retrans (AMR) setting, we prescribe for that an 436 SCTP-PF implementation SHOULD operate as described below in 437 Section 4.1 during dormant state. 439 An SCTP-PF implementation MAY choose a different dormant state 440 operation than the one described below in Section 4.1 provided that 441 the solution chosen does not compromise the fault tolerance of the 442 SCTP-PF operation. 444 The below prescription for SCTP-PF dormant state handling SHOULD NOT 445 be coupled to the value of the PFMR, but solely to the activation of 446 SCTP-PF logic in an SCTP implementation. 448 It is noted that the below dormant state operation is considered to 449 provide added disruption tolerance also for an [RFC4960] SCTP 450 implementation, and that it can be sensible for an [RFC4960] SCTP 451 implementation to follow this mode of operation. For an [RFC4960] 452 SCTP implementation the continuation of data transmission during 453 dormant state makes the fault tolerance of SCTP be more robust 454 towards situations where some, or all, alternative paths of an SCTP 455 association approach, or reach, inactive state before the primary 456 path used for data transmission observes trouble. 458 4.1. SCTP Dormant State Procedure 460 a. When the destination addresses are all in inactive state and data 461 is available for transfer, the sender MUST choose one destination 462 and transmit data to this destination address. 464 b. The sender MUST NOT change the state of the chosen destination 465 address (it remains in inactive state) and it MUST NOT clear the 466 error counter of the destination address as a result of choosing 467 the destination address for data transmission. 469 c. The sender SHOULD choose the destination in inactive state with 470 the lowest error count (fewest consecutive timeouts) for data 471 transmission. When there are multiple destinations with same 472 error count in inactive state, the sender SHOULD attempt to pick 473 the most divergent source - destination pair from the last source 474 - destination pair where failure was observed. Rules for picking 475 the most divergent source-destination pair are an implementation 476 decision and are not specified within this document. To support 477 differentiation of inactive destination addresses based on their 478 error count SCTP will need to allow for increment of the 479 destination address error counters up to some reasonable limit 480 above PMR+1, thus changing the prescriptions of [RFC4960], 481 section 8.3, in this respect. The exact limit to apply is not 482 specified in this document but it is considered reasonable to 483 require for the limit to be an order of magnitude higher than the 484 PMR value. A sender MAY choose to deploy other strategies that 485 the strategy defined here. The strategy to prioritize the last 486 active destination address, i.e., the destination address with 487 the fewest error counts is optimal when some paths are 488 permanently inactive, but suboptimal when a path instability is 489 transient. 491 5. Primary Path Switchover 493 The objective of the Primary Path Switchover operation is to allow 494 the SCTP sender to continue data transmission on a new working path 495 even when the old primary destination address becomes active again. 496 This is achieved by having SCTP perform a switchover of the primary 497 path to the new working path if the error counter of the primary path 498 exceeds a certain threshold. This mode of operation can be applied 499 not only to SCTP-PF implementations, but also to [RFC4960] 500 implementations. 502 The Primary Path Switchover operation requires only sender side 503 changes. The details are: 505 1. The sender maintains a new tunable parameter, called 506 Primary.Switchover.Max.Retrans (PSMR). For SCTP-PF 507 implementations, the PSMR MUST be set greater or equal to the 508 PFMR value. For [RFC4960] implementations the PSMR MUST be set 509 greater or equal to the PMR value. Implementations MUST reject 510 any other values of PSMR. 512 2. When the path error counter on a set primary path exceeds PSMR, 513 the SCTP implementation MUST autonomously select and set a new 514 primary path. 516 3. The primary path selected by the SCTP implementation MUST be the 517 path which at the given time would be chosen for data transfer. 518 A previously failed primary path can be used as data transfer 519 path as per normal path selection when the present data transfer 520 path fails. 522 4. For SCTP-PF, the recommended value of PSMR is PFMR when Primary 523 Path Switchover operation mode is used. This means that no 524 forced switchback to a previously failed primary path is 525 performed. An SCTP-PF implementation of Primary Path Switchover 526 MUST support the setting of PSMR = PFMR. A SCTP-PF 527 implementation of Primary Path Switchover MAY support setting of 528 PSMR > PFMR. 530 5. For [RFC4960] SCTP, the recommended value of PSMR is PMR when 531 Primary Path Switchover is used. This means that no forced 532 switchback to a previously failed primary path is performed. A 533 [RFC4960] SCTP implementation of Primary Path Switchover MUST 534 support the setting of PSMR = PMR. An [RFC4960] SCTP 535 implementation of Primary Path Switchover MAY support larger 536 settings of PSMR > PMR. 538 6. It MUST be possible to disable the Primary Path Switchover 539 operation and obtain the standard switchback operation of 540 [RFC4960]. 542 The manner of switchover operation that is most optimal in a given 543 scenario depends on the relative quality of a set primary path versus 544 the quality of alternative paths available as well as on the extent 545 to which it is desired for the mode of operation to enforce traffic 546 distribution over a number of network paths. I.e., load distribution 547 of traffic from multiple SCTP associations may be sought to be 548 enforced by distribution of the set primary paths with [RFC4960] 549 switchback operation. However as [RFC4960] switchback behavior is 550 suboptimal in certain situations, especially in scenarios where a 551 number of equally good paths are available, an SCTP implementation 552 MAY support also, as alternative behavior, the Primary Path 553 Switchover mode of operation and MAY enable it based on users' 554 requests. 556 For an SCTP implementation that implements the Primary Path 557 Switchover operation, this specification RECOMMENDS that the standard 558 RFC4960 switchback operation is retained as the default operation. 560 6. Suggested SCTP Protocol Parameter Values 562 This document does not alter the [RFC4960] value RECOMMENDATIONS for 563 the SCTP Protocol Parameters defined in [RFC4960]. 565 The following protocol parameter is RECOMMENDED: 567 PotentiallyFailed.Max.Retrans (PFMR) - 0 569 7. Socket API Considerations 571 This section describes how the socket API defined in [RFC6458] is 572 extended to provide a way for the application to control and observe 573 the SCTP-PF behavior as well as the Primary Path Switchover function. 575 Please note that this section is informational only. 577 A socket API implementation based on [RFC6458] is, by means of the 578 existing SCTP_PEER_ADDR_CHANGE event, extended to provide the event 579 notification when a peer address enters or leaves the potentially 580 failed state as well as the socket API implementation is extended to 581 expose the potentially failed state of a peer address in the existing 582 SCTP_GET_PEER_ADDR_INFO structure. 584 Furthermore, two new read/write socket options for the level 585 IPPROTO_SCTP and the name SCTP_PEER_ADDR_THLDS and 586 SCTP_EXPOSE_POTENTIALLY_FAILED_STATE are defined as described below. 587 The first socket option is used to control the values of the PFMR and 588 PSMR parameters described in Section 3 and in Section 5. The second 589 one controls the exposition of the potentially failed path state. 591 Support for the SCTP_PEER_ADDR_THLDS and 592 SCTP_EXPOSE_POTENTIALLY_FAILED_STATE socket options need also to be 593 added to the function sctp_opt_info(). 595 7.1. Support for the Potentially Failed Path State 597 As defined in [RFC6458], the SCTP_PEER_ADDR_CHANGE event is provided 598 if the status of a peer address changes. In addition to the state 599 changes described in [RFC6458], this event is also provided, if a 600 peer address enters or leaves the potentially failed state. The 601 notification as defined in [RFC6458] uses the following structure: 603 struct sctp_paddr_change { 604 uint16_t spc_type; 605 uint16_t spc_flags; 606 uint32_t spc_length; 607 struct sockaddr_storage spc_aaddr; 608 uint32_t spc_state; 609 uint32_t spc_error; 610 sctp_assoc_t spc_assoc_id; 611 } 613 [RFC6458] defines the constants SCTP_ADDR_AVAILABLE, 614 SCTP_ADDR_UNREACHABLE, SCTP_ADDR_REMOVED, SCTP_ADDR_ADDED, and 615 SCTP_ADDR_MADE_PRIM to be provided in the spc_state field. This 616 document defines in addition to that the new constant 617 SCTP_ADDR_POTENTIALLY_FAILED, which is reported if the affected 618 address becomes potentially failed. 620 The SCTP_GET_PEER_ADDR_INFO socket option defined in [RFC6458] can be 621 used to query the state of a peer address. It uses the following 622 structure: 624 struct sctp_paddrinfo { 625 sctp_assoc_t spinfo_assoc_id; 626 struct sockaddr_storage spinfo_address; 627 int32_t spinfo_state; 628 uint32_t spinfo_cwnd; 629 uint32_t spinfo_srtt; 630 uint32_t spinfo_rto; 631 uint32_t spinfo_mtu; 632 }; 634 [RFC6458] defines the constants SCTP_UNCONFIRMED, SCTP_ACTIVE, and 635 SCTP_INACTIVE to be provided in the spinfo_state field. This 636 document defines in addition to that the new constant 637 SCTP_POTENTIALLY_FAILED, which is reported if the peer address is 638 potentially failed. 640 7.2. Peer Address Thresholds (SCTP_PEER_ADDR_THLDS) Socket Option 642 Applications can control the SCTP-PF behavior by getting or setting 643 the number of consecutive timeouts before a peer address is 644 considered potentially failed or unreachable. The same socket option 645 is used by applications to set and get the number of timeouts before 646 the primary path is changed automatically by the Primary Path 647 Switchover function. This socket option uses the level IPPROTO_SCTP 648 and the name SCTP_PEER_ADDR_THLDS. 650 The following structure is used to access and modify the thresholds: 652 struct sctp_paddrthlds { 653 sctp_assoc_t spt_assoc_id; 654 struct sockaddr_storage spt_address; 655 uint16_t spt_pathmaxrxt; 656 uint16_t spt_pathpfthld; 657 uint16_t spt_pathcpthld; 658 }; 660 spt_assoc_id: This parameter is ignored for one-to-one style 661 sockets. For one-to-many style sockets the application may fill 662 in an association identifier or SCTP_FUTURE_ASSOC. It is an error 663 to use SCTP_{CURRENT|ALL}_ASSOC in spt_assoc_id. 665 spt_address: This specifies which peer address is of interest. If a 666 wild card address is provided, this socket option applies to all 667 current and future peer addresses. 669 spt_pathmaxrxt: Each peer address of interest is considered 670 unreachable, if its path error counter exceeds spt_pathmaxrxt. 672 spt_pathpfthld: Each peer address of interest is considered 673 Potentially Failed, if its path error counter exceeds 674 spt_pathpfthld. 676 spt_pathcpthld: Each peer address of interest is not considered the 677 primary remote address anymore, if its path error counter exceeds 678 spt_pathcpthld. Using a value of 0xffff disables the selection of 679 a new primary peer address. If an implementation does not support 680 the automatically selection of a new primary address, it should 681 indicate an error with errno set to EINVAL if a value different 682 from 0xffff is used in spt_pathcpthld. For SCTP-PF, the setting 683 of spt_pathcpthld < spt_pathpfthld should be rejected with errno 684 set to EINVAL. For [RFC4960] SCTP, the setting of spt_pathcpthld 685 < spt_pathmaxrxt should be rejected with errno set to EINVAL. A 686 SCTP-PF implementation MAY support only setting of spt_pathcpthld 687 = spt_pathpfthld and spt_pathcpthld = 0xffff and a [RFC4960] SCTP 688 implementation MAY support only setting of spt_pathcpthld = 689 spt_pathmaxrxt and spt_pathcpthld = 0xffff. In these cases SCTP 690 shall reject setting of other values with errno set to EINVAL. 692 7.3. Exposing the Potentially Failed Path State 693 (SCTP_EXPOSE_POTENTIALLY_FAILED_STATE) Socket Option 695 Applications can control the exposure of the potentially failed path 696 state in the SCTP_PEER_ADDR_CHANGE event and the 697 SCTP_GET_PEER_ADDR_INFO as described in Section 7.1. The default 698 value is implementation specific. 700 This socket option uses the level IPPROTO_SCTP and the name 701 SCTP_EXPOSE_POTENTIALLY_FAILED_STATE. 703 The following structure is used to control the exposition of the 704 potentially failed path state: 706 struct sctp_assoc_value { 707 sctp_assoc_t assoc_id; 708 uint32_t assoc_value; 709 }; 711 assoc_id: This parameter is ignored for one-to-one style sockets. 712 For one-to-many style sockets the application may fill in an 713 association identifier or SCTP_FUTURE_ASSOC. It is an error to 714 use SCTP_{CURRENT|ALL}_ASSOC in assoc_id. 716 assoc_value: The potentially failed path state is exposed if and 717 only if this parameter is non-zero. 719 8. Security Considerations 721 Security considerations for the use of SCTP and its APIs are 722 discussed in [RFC4960] and [RFC6458]. 724 The logic introduced by this document does not impact existing SCTP 725 messages on the wire. Also, this document does not introduce any new 726 SCTP messages on the wire that require new security considerations. 728 SCTP-PF makes SCTP not only more robust during primary path failure/ 729 congestion but also more vulnerable to network connectivity/ 730 congestion attacks on the primary path. SCTP-PF makes it easier for 731 an attacker to trick SCTP to change data transfer path, since the 732 duration of time that an attacker needs to compromise the network 733 connectivity is much shorter than [RFC4960]. However, SCTP-PF does 734 not constitute a significant change in the duration of time and 735 effort an attacker needs to keep SCTP away from the primary path. 736 With the standard switchback operation [RFC4960] SCTP resumes data 737 transfer on its primary path as soon as the next HEARTBEAT succeeds. 739 On the other hand, usage of the Primary Path Switchover mechanism, 740 does change the threat analysis. This is because on-path attackers 741 can force a permanent change of the data transfer path by blocking 742 the primary path until the switchover of the primary path is 743 triggered by the Primary Path Switchover algorithm. This especially 744 will be the case when the Primary Path Switchover is used together 745 with SCTP-PF with the particular setting of PSMR = PFMR = 0, as 746 Primary Path Switchover here happens already at the first RTO timeout 747 experienced. Users of the Primary Path Switchover mechanism should 748 be aware of this fact. 750 The event notification of path state transfer from active to 751 potentially failed state and vice versa gives attackers an increased 752 possibility to generate more local events. However, it is assumed 753 that event notifications are rate-limited in the implementation to 754 address this threat. 756 9. IANA Considerations 758 This document does not create any new registries or modify the rules 759 for any existing registries managed by IANA. 761 10. Acknowledgements 763 The authors wish to thank Michael Tuexen for his many invaluable 764 comments and for his very substantial support with the making of this 765 document. 767 11. Proposed Change of Status (to be Deleted before Publication) 769 Initially this work looked to entail some changes of the Congestion 770 Control (CC) operation of SCTP and for this reason the work was 771 proposed as Experimental. These intended changes of the CC operation 772 have since been judged to be irrelevant and are no longer part of the 773 specification. As the specification entails no other potential 774 harmful features, consensus exists in the WG to bring the work 775 forward as PS. 777 Initially concerns have been expressed about the possibility for the 778 mechanism to introduce path bouncing with potential harmful network 779 impacts. These concerns are believed to be unfounded. This issue is 780 addressed in Appendix B. 782 It is noted that the feature specified by this document is 783 implemented by multiple SCTP SW implementations and furthermore that 784 various variants of the solution have been deployed in Tel co 785 signaling environments for several years with good results. 787 12. References 789 12.1. Normative References 791 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 792 Requirement Levels", BCP 14, RFC 2119, 793 DOI 10.17487/RFC2119, March 1997, 794 . 796 [RFC4960] Stewart, R., Ed., "Stream Control Transmission Protocol", 797 RFC 4960, DOI 10.17487/RFC4960, September 2007, 798 . 800 12.2. Informative References 802 [CARO02] Caro Jr., A., Iyengar, J., Amer, P., Heinz, G., and R. 803 Stewart, "A Two-level Threshold Recovery Mechanism for 804 SCTP", Tech report, CIS Dept, University of Delaware , 7 805 2002. 807 [CARO04] Caro Jr., A., Amer, P., and R. Stewart, "End-to-End 808 Failover Thresholds for Transport Layer Multi homing", 809 MILCOM 2004 , 11 2004. 811 [CARO05] Caro Jr., A., "End-to-End Fault Tolerance using Transport 812 Layer Multi homing", Ph.D Thesis, University of Delaware , 813 1 2005. 815 [FALLON08] 816 Fallon, S., Jacob, P., Qiao, Y., Murphy, L., Fallon, E., 817 and A. Hanley, "SCTP Switchover Performance Issues in WLAN 818 Environments", IEEE CCNC 2008, 1 2008. 820 [GRINNEMO04] 821 Grinnemo, K-J. and A. Brunstrom, "Performance of SCTP- 822 controlled failovers in M3UA-based SIGTRAN networks", 823 Advanced Simulation Technologies Conference , 4 2004. 825 [IYENGAR06] 826 Iyengar, J., Amer, P., and R. Stewart, "Concurrent 827 Multipath Transfer using SCTP Multihoming over Independent 828 End-to-end Paths.", IEEE/ACM Trans on Networking 14(5), 10 829 2006. 831 [JUNGMAIER02] 832 Jungmaier, A., Rathgeb, E., and M. Tuexen, "On the use of 833 SCTP in failover scenarios", World Multiconference on 834 Systemics, Cybernetics and Informatics , 7 2002. 836 [NATARAJAN09] 837 Natarajan, P., Ekiz, N., Amer, P., and R. Stewart, 838 "Concurrent Multipath Transfer during Path Failure", 839 Computer Communications , 5 2009. 841 [RFC6458] Stewart, R., Tuexen, M., Poon, K., Lei, P., and V. 842 Yasevich, "Sockets API Extensions for the Stream Control 843 Transmission Protocol (SCTP)", RFC 6458, 844 DOI 10.17487/RFC6458, December 2011, 845 . 847 Appendix A. Discussions of Alternative Approaches 849 This section lists alternative approaches for the issues described in 850 this document. Although these approaches do not require to update 851 RFC4960, we do not recommend them from the reasons described below. 853 A.1. Reduce Path.Max.Retrans (PMR) 855 Smaller values for Path.Max.Retrans shorten the failover duration and 856 in fact this is recommended in some research results [JUNGMAIER02] 857 [GRINNEMO04] [FALLON08]. However to significantly reduce the 858 failover time it is required to go down (as with PFMR) to 859 Path.Max.Retrans=0 and with this setting SCTP switches to another 860 destination address already on a single timeout which may result in 861 spurious failover. Spurious failover is a problem in [RFC4960] SCTP 862 as the transmission of HEARTBEATS on the left primary path, unlike in 863 SCTP-PF, is governed by 'HB.interval' also during the failover 864 process. 'HB.interval' is usually set in the order of seconds 865 (recommended value is 30 seconds) and when the primary path becomes 866 inactive, the next HEARTBEAT may be transmitted only many seconds 867 later. Indeed as recommended, only 30 secs later. Meanwhile, the 868 primary path may since long have recovered, if it needed recovery at 869 all (indeed the failover could be truly spurious). In such 870 situations, post failover, an endpoint is forced to wait in the order 871 of many seconds before the endpoint can resume transmission on the 872 primary path and furthermore once it returns on the primary path the 873 CWND needs to be rebuild anew - a process which the throughput 874 already have had to suffer from on the alternate path. Using a 875 smaller value for 'HB.interval' might help this situation, but it 876 would result in a general waste of bandwidth as such more frequent 877 HEARTBEATING would take place also when there are no observed 878 troubles. The bandwidth overhead may be diminished by having the ULP 879 use a smaller 'HB.interval' only on the path which at any given time 880 is set to be the primary path, but this adds complication in the ULP. 882 In addition, smaller Path.Max.Retrans values also affect the 883 'Association.Max.Retrans' value. When the SCTP association's error 884 count exceeds Association.Max.Retrans threshold, the SCTP sender 885 considers the peer endpoint unreachable and terminates the 886 association. Section 8.2 in [RFC4960] recommends that 887 Association.Max.Retrans value should not be larger than the summation 888 of the Path.Max.Retrans of each of the destination addresses. Else 889 the SCTP sender considers its peer reachable even when all 890 destinations are INACTIVE and to avoid this dormant state operation, 891 [RFC4960] SCTP implementation SHOULD reduce Association.Max.Retrans 892 accordingly whenever it reduces Path.Max.Retrans. However, smaller 893 Association.Max.Retrans value compromises the fault tolerance of SCTP 894 as it increases the chances of association termination during minor 895 congestion events. 897 A.2. Adjust RTO related parameters 899 As several research results indicate, we can also shorten the 900 duration of failover process by adjusting RTO related parameters 901 [JUNGMAIER02] [FALLON08]. During failover process, RTO keeps being 902 doubled. However, if we can choose smaller value for RTO.max, we can 903 stop the exponential growth of RTO at some point. Also, choosing 904 smaller values for RTO.initial or RTO.min can contribute to keep the 905 RTO value small. 907 Similar to reducing Path.Max.Retrans, the advantage of this approach 908 is that it requires no modification to the current specification, 909 although it needs to ignore several recommendations described in the 910 Section 15 of [RFC4960]. However, this approach requires to have 911 enough knowledge about the network characteristics between end 912 points. Otherwise, it can introduce adverse side-effects such as 913 spurious timeouts. 915 The significant issue with this approach, however, is that even if 916 the RTO.max is lowered to an optimal low value, then as long as the 917 Path.Max.Retrans is kept at the [RFC4960] recommended value, the 918 reduction of the RTO.max doesn't reduce the failover time 919 sufficiently enough to prevent severe performance degradation during 920 failover. 922 Appendix B. Discussions for Path Bouncing Effect 924 The methods described in the document can accelerate the failover 925 process. Hence, they might introduce the path bouncing effect where 926 the sender keeps changing the data transmission path frequently. 927 This sounds harmful to the data transfer, however several research 928 results indicate that there is no serious problem with SCTP in terms 929 of path bouncing effect [CARO04] [CARO05]. 931 There are two main reasons for this. First, SCTP is basically 932 designed for multipath communication, which means SCTP maintains all 933 path related parameters (CWND, ssthresh, RTT, error count, etc) per 934 each destination address. These parameters cannot be affected by 935 path bouncing. In addition, when SCTP migrates the data transfer to 936 another path, it starts with the minimal or the initial CWND. Hence, 937 there is little chance for packet reordering or duplicating. 939 Second, even if all communication paths between the end-nodes share 940 the same bottleneck, the SCTP-PF results in a behavior already 941 allowed by [RFC4960]. 943 Appendix C. SCTP-PF for SCTP Single-homed Operation 945 For a single-homed SCTP association the only tangible effect of the 946 activation of SCTP-PF operation is enhanced failure detection in 947 terms of potential notification of the PF state of the sole 948 destination address as well as, for idle associations, more rapid 949 entering, and notification, of inactive state of the destination 950 address and more rapid end-point failure detection. It is believed 951 that neither of these effects are harmful, provided adequate dormant 952 state operation is implemented, and furthermore that they may be 953 particularly useful for applications that deploys multiple SCTP 954 associations for load balancing purposes. The early notification of 955 the PF state may be used for preventive measures as the entering of 956 the PF state can be used as a warning of potential congestion. 957 Depending on the PMR value, the aggressive HEARTBEAT transmission in 958 PF state may speed up the end-point failure detection (exceed of AMR 959 threshold on the sole path error counter) on idle associations in 960 case where relatively large HB.interval value compared to RTO (e.g. 961 30secs) is used. 963 Authors' Addresses 965 Yoshifumi Nishida 966 GE Global Research 967 2623 Camino Ramon 968 San Ramon, CA 94583 969 USA 971 Email: nishida@wide.ad.jp 973 Preethi Natarajan 974 Cisco Systems 975 510 McCarthy Blvd 976 Milpitas, CA 95035 977 USA 979 Email: prenatar@cisco.com 981 Armando Caro 982 BBN Technologies 983 10 Moulton St. 984 Cambridge, MA 02138 985 USA 987 Email: acaro@bbn.com 989 Paul D. Amer 990 University of Delaware 991 Computer Science Department - 434 Smith Hall 992 Newark, DE 19716-2586 993 USA 995 Email: amer@udel.edu 996 Karen E. E. Nielsen 997 Ericsson 998 Kistavaegen 25 999 Stockholm 164 80 1000 Sweden 1002 Email: karen.nielsen@tieto.com