idnits 2.17.1 draft-ietf-forces-sctptml-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** The document seems to lack a License Notice according IETF Trust Provisions of 28 Dec 2009, Section 6.b.i or Provisions of 12 Sep 2009 Section 6.b -- however, there's a paragraph with a matching beginning. Boilerplate error? (You're using the IETF Trust Provisions' Section 6.b License Notice from 12 Feb 2009 rather than one of the newer Notices. See https://trustee.ietf.org/license-info/.) Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([RFC3654], [RFC4960], [FE-PROTO]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. ** The document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 692: '...nded that the FE SHOULD do socket conn...' RFC 2119 keyword, line 695: '... however, MUST NOT assume that there...' RFC 2119 keyword, line 781: '...ded that the FE and CE SHOULD assume a...' RFC 2119 keyword, line 1043: '... ForCES FE or CE MUST support the foll...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors 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.) -- The document date (June 4, 2009) is 5441 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) -- Obsolete informational reference (is this intentional?): RFC 3768 (Obsoleted by RFC 5798) Summary: 4 errors (**), 0 flaws (~~), 2 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group J. Hadi Salim 3 Internet-Draft Mojatatu Networks 4 Expires: December 6, 2009 K. Ogawa 5 NTT Corporation 6 June 4, 2009 8 SCTP based TML (Transport Mapping Layer) for ForCES protocol 9 draft-ietf-forces-sctptml-03 11 Status of this Memo 13 This Internet-Draft is submitted to IETF in full conformance with the 14 provisions of BCP 78 and BCP 79. 16 Internet-Drafts are working documents of the Internet Engineering 17 Task Force (IETF), its areas, and its working groups. Note that 18 other groups may also distribute working documents as Internet- 19 Drafts. 21 Internet-Drafts are draft documents valid for a maximum of six months 22 and may be updated, replaced, or obsoleted by other documents at any 23 time. It is inappropriate to use Internet-Drafts as reference 24 material or to cite them other than as "work in progress." 26 The list of current Internet-Drafts can be accessed at 27 http://www.ietf.org/ietf/1id-abstracts.txt. 29 The list of Internet-Draft Shadow Directories can be accessed at 30 http://www.ietf.org/shadow.html. 32 This Internet-Draft will expire on December 6, 2009. 34 Copyright Notice 36 Copyright (c) 2009 IETF Trust and the persons identified as the 37 document authors. All rights reserved. 39 This document is subject to BCP 78 and the IETF Trust's Legal 40 Provisions Relating to IETF Documents in effect on the date of 41 publication of this document (http://trustee.ietf.org/license-info). 42 Please review these documents carefully, as they describe your rights 43 and restrictions with respect to this document. 45 Abstract 47 This document defines the SCTP based TML (Transport Mapping Layer) 48 for the ForCES protocol. It explains the rationale for choosing the 49 SCTP (Stream Control Transmission Protocol) [RFC4960] and also 50 describes how this TML addresses all the requirements described in 51 [RFC3654] and the ForCES protocol [FE-PROTO] draft. 53 Table of Contents 55 1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3 56 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 3. Protocol Framework Overview . . . . . . . . . . . . . . . . . 3 58 3.1. The PL . . . . . . . . . . . . . . . . . . . . . . . . . . 4 59 3.2. The TML . . . . . . . . . . . . . . . . . . . . . . . . . 5 60 3.2.1. TML and PL Interfaces . . . . . . . . . . . . . . . . 5 61 3.2.2. TML Parameterization . . . . . . . . . . . . . . . . . 6 62 4. SCTP TML overview . . . . . . . . . . . . . . . . . . . . . . 6 63 4.1. Rationale for using SCTP for TML . . . . . . . . . . . . . 8 64 4.2. Meeting TML requirements . . . . . . . . . . . . . . . . . 8 65 4.2.1. SCTP TML Channels . . . . . . . . . . . . . . . . . . 9 66 4.2.2. Satisfying TML Requirements . . . . . . . . . . . . . 14 67 5. SCTP TML Channel work . . . . . . . . . . . . . . . . . . . . 16 68 5.1. SCTP TML Channel Initialization . . . . . . . . . . . . . 16 69 5.2. Channel work scheduling . . . . . . . . . . . . . . . . . 16 70 5.2.1. FE Channel work scheduling . . . . . . . . . . . . . . 17 71 5.2.2. CE Channel work scheduling . . . . . . . . . . . . . . 17 72 5.3. SCTP TML Channel Termination . . . . . . . . . . . . . . . 18 73 5.4. SCTP TML NE level channel scheduling . . . . . . . . . . . 18 74 6. Service Interface . . . . . . . . . . . . . . . . . . . . . . 19 75 6.1. TML Boot-strapping . . . . . . . . . . . . . . . . . . . . 19 76 6.2. TML Shutdown . . . . . . . . . . . . . . . . . . . . . . . 21 77 6.3. TML Sending and Receiving . . . . . . . . . . . . . . . . 21 78 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 23 79 8. Security Considerations . . . . . . . . . . . . . . . . . . . 23 80 8.1. IPsec Usage . . . . . . . . . . . . . . . . . . . . . . . 24 81 8.1.1. SAD and SPD setup . . . . . . . . . . . . . . . . . . 24 82 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 24 83 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 25 84 10.1. Normative References . . . . . . . . . . . . . . . . . . . 25 85 10.2. Informative References . . . . . . . . . . . . . . . . . . 25 86 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 26 88 1. Definitions 90 The following definitions are taken from [RFC3654]and [RFC3746]: 92 ForCES Protocol -- The protocol used at the Fp reference point in the 93 ForCES Framework in [RFC3746]. 95 ForCES Protocol Layer (ForCES PL) -- A layer in ForCES protocol 96 architecture that defines the ForCES protocol architecture and the 97 state transfer mechanisms as defined in [FE-PROTO]. 99 ForCES Protocol Transport Mapping Layer (ForCES TML) -- A layer in 100 ForCES protocol architecture that specifically addresses the protocol 101 message transportation issues, such as how the protocol messages are 102 mapped to different transport media (like SCTP, IP, ATM, Ethernet, 103 etc), and how to achieve and implement reliability, security, etc. 105 2. Introduction 107 The ForCES (Forwarding and Control Element Separation) working group 108 in the IETF defines the architecture and protocol for separation of 109 Control Elements(CE) and Forwarding Elements(FE) in Network 110 Elements(NE) such as routers. [RFC3654] and [RFC3746] respectively 111 define architectural and protocol requirements for the communication 112 between CE and FE. The ForCES protocol layer specification 113 [FE-PROTO] describes the protocol semantics and workings. The ForCES 114 protocol layer operates on top of an inter-connect hiding layer known 115 as the TML. The relationship is illustrated in Figure 1. 117 This document defines the SCTP based TML for the ForCES protocol 118 layer. It also addresses all the requirements for the TML including 119 security, reliability, etc as defined in [FE-PROTO]. 121 3. Protocol Framework Overview 123 The reader is referred to the Framework document [RFC3746], and in 124 particular sections 3 and 4, for an architectural overview and 125 explanation of where and how the ForCES protocol fits in. 127 There is some content overlap between the ForCES protocol draft 128 [FE-PROTO] and this section (Section 3) in order to provide basic 129 context to the reader of this document. 131 The ForCES protocol layering constitutes two pieces: the PL and TML 132 layer. This is depicted in Figure 1. 134 +----------------------------------------------+ 135 | CE PL | 136 +----------------------------------------------+ 137 | CE TML | 138 +----------------------------------------------+ 139 ^ 140 | 141 ForCES PL | messages 142 | 143 v 144 +-----------------------------------------------+ 145 | FE TML | 146 +-----------------------------------------------+ 147 | FE PL | 148 +-----------------------------------------------+ 150 Figure 1: Message exchange between CE and FE to establish an NE 151 association 153 The PL is in charge of the ForCES protocol. Its semantics and 154 message layout are defined in [FE-PROTO]. The TML is necessary to 155 connect two ForCES end-points as shown in Figure 1. 157 Both the PL and TML are standardized by the IETF. While only one PL 158 is defined, different TMLs are expected to be standardized. The TML 159 at each of the nodes (CE and FE) is expected to be of the same 160 definition in order to inter-operate. 162 When transmitting from a ForCES end-point, the PL delivers its 163 messages to the TML. The TML then delivers the PL message to the 164 destination TML(s). 166 On reception of a message, the TML delivers the message to its 167 destination PL level (as described in the ForCES header). 169 3.1. The PL 171 The PL is common to all implementations of ForCES and is standardized 172 by the IETF [FE-PROTO]. The PL level is responsible for associating 173 an FE or CE to an NE. It is also responsible for tearing down such 174 associations. 176 An FE may use the PL level to asynchronously send packets to the CE. 177 The FE may redirect via the PL (from outside the NE) various control 178 protocol packets (e.g. OSPF, etc) to the CE. Additionally, the FE 179 delivers various events that CE has subscribed-to via PL [FE-MODEL]. 181 The CE and FE may interact synchronously via the PL. The CE issues 182 status requests to the FE and receives responses via the PL. The CE 183 also configures the associated FE's LFBs' components using the PL 184 [FE-MODEL]. 186 3.2. The TML 188 The TML level is responsible for transport of the PL level messages. 189 [FE-PROTO] section 5 defines the requirements that need to be met by 190 a TML specification. The SCTP TML specified in this document meets 191 all the requirements specified in [FE-PROTO] section 5. 192 Section 4.2.2 describes how the TML requirements are met. 194 3.2.1. TML and PL Interfaces 196 There are two interfaces to the PL and TML, both of which are out of 197 scope for ForCES. The first one is the interface between the PL and 198 TML and the other is the CE Manager (CEM)/FE Manager (FEM)[RFC3746] 199 interface to both the PL and TML. Both interfaces are shown in 200 Figure 2. 202 +----------------------------+ 203 | +----------------------+ | 204 | | | | 205 +---------+ | | PL Layer | | 206 | | | +----------------------+ | 207 |FEM/CEM |<---->| ^ | 208 | | | | | 209 +---------+ | |TML API | 210 | | | 211 | V | 212 | +----------------------+ | 213 | | | | 214 | | TML Layer | | 215 | | | | 216 | +----------------------+ | 217 +----------------------------+ 219 Figure 2: The TML-PL interface 221 Figure 2 also shows an interface referred to as CEM/FEM[RFC3746] 222 which is responsible for bootstrapping and parameterization of the 223 TML. In its most basic form the CEM/FEM interface takes the form of 224 a simple static config file which is read on startup in the pre- 225 association phase. 227 Section 6 discusses in more details the service interfaces. 229 3.2.2. TML Parameterization 231 It is expected that it should be possible to use a configuration 232 reference point, such as the FEM or the CEM, to configure the TML. 234 Some of the configured parameters may include: 236 o PL ID 238 o Connection Type and associated data. For example if a TML uses 239 IP/SCTP then parameters such as SCTP ports and IP addresses need 240 to be configured. 242 o Number of transport connections 244 o Connection Capability, such as bandwidth, etc. 246 o Allowed/Supported Connection QoS policy (or Congestion Control 247 Policy) 249 4. SCTP TML overview 251 SCTP [RFC4960] is an end-to-end transport protocol that is equivalent 252 to TCP, UDP, or DCCP in many aspects. With a few exceptions, SCTP 253 can do most of what UDP, TCP, or DCCP can achieve. SCTP as well can 254 do most of what a combination of the other transport protocols can 255 achieve (e.g. TCP and DCCP or TCP and UDP). 257 Like TCP, it provides ordered, reliable, connection-oriented, flow- 258 controlled, congestion controlled data exchange. Unlike TCP, it does 259 not provide byte streaming and instead provides message boundaries. 261 Like UDP, it can provide unreliable, unordered data exchange. Unlike 262 UDP, it does not provide multicast support 264 Like DCCP, it can provide unreliable, ordered, congestion controlled, 265 connection-oriented data exchange. 267 SCTP also provides other services that none of the 3 transport 268 protocols mentioned above provide. These include: 270 o Multi-homing 271 An SCTP connection can make use of multiple destination IP 272 addresses to communicate with its peer. 274 o Runtime IP address binding 275 With the SCTP Dynamic Address Reconfiguration ([RFC5061]) feature, 276 a new IP address can be bound at runtime. This allows for 277 migration of endpoints without restarting the association 278 (valuable for high availability). 280 o A range of reliability shades with congestion control 281 SCTP offers a range of services from full reliability to none, and 282 from full ordering to none. With SCTP, on a per message basis, 283 the application can specify a message's time-to-live. When the 284 expressed time expires, the message can be "skipped". 286 o Built-in heartbeats 287 SCTP has built-in heartbeat mechanism that validate the 288 reachability of peer addresses. 290 o Multi-streaming 291 A known problem with TCP is head of line (HOL) blocking. If you 292 have independent messages, TCP enforces ordering of such messages. 293 Loss at the head of the messages implies delays of delivery of 294 subsequent packets. SCTP allows for defining up to 64K 295 independent streams over the same socket connection, which are 296 ordered independently. 298 o Message boundaries with reliability 299 SCTP allows for easier message parsing (just like UDP but with 300 reliability built in) because it establishes boundaries on a PL 301 message basis. On a TCP stream, one would have to use techniques 302 such peeking into the message to figure the boundaries. 304 o Improved SYN DOS protection 305 Unlike TCP, which does a 3 way connection setup handshake, SCTP 306 does a 4 way handshake. This improves against SYN-flood attacks 307 because listening sockets do not set up state until a connection 308 is validated. 310 o Simpler transport events 311 An application (such as the TML) can subscribe to be notified of 312 both local and remote transport events. Events that can be 313 subscribed-to include indication of association changes, 314 addressing changes, remote errors, expiry of timed messages, etc. 315 These events are off by default and require explicit subscription. 317 o Simplified replicasting 318 Although SCTP does not allow for multicasting it allows for a 319 single message from an application to be sent to multiple peers. 320 This reduces the messaging that typically crosses different memory 321 domains within a host (example in a kernel to user space domain of 322 an operating system). 324 4.1. Rationale for using SCTP for TML 326 SCTP has all the features required to provide a robust TML. As a 327 transport that is all-encompassing, it negates the need for having 328 multiple transport protocols in order to satisfy the TML requirements 329 ([FE-PROTO] section 5). As a result it allows for simpler coding and 330 therefore reduces a lot of the interoperability concerns. 332 SCTP is also very mature and widely used making it a good choice for 333 ubiquitous deployment. 335 4.2. Meeting TML requirements 337 PL 338 +----------------------+ 339 | | 340 +-----------+----------+ 341 | TML API 342 TML | 343 +-----------+----------+ 344 | | | 345 | +------+------+ | 346 | | TML core | | 347 | +-+----+----+-+ | 348 | | | | | 349 | SCTP socket API | 350 | | | | | 351 | | | | | 352 | +-+----+----+-+ | 353 | | SCTP | | 354 | +------+------+ | 355 | | | 356 | | | 357 | +------+------+ | 358 | | IP | | 359 | +-------------+ | 360 +----------------------+ 362 Figure 3: The TML-SCTP interface 364 Figure 3 details the interfacing between the PL and SCTP TML and the 365 internals of the SCTP TML. The core of the TML interacts on its 366 north-bound interface to the PL (utilizing the TML API). On the 367 south-bound interface, the TML core interfaces to the SCTP layer 368 utilizing the standard socket interface[SCTP-API] There are three 369 SCTP socket connections opened between any two PL endpoints (whether 370 FE or CE). 372 4.2.1. SCTP TML Channels 374 +--------------------+ 375 | | 376 | TML core | 377 | | 378 +-+-------+--------+-+ 379 | | | 380 | Med prio, | 381 | Semi-reliable | 382 | channel | 383 | | Low prio, 384 | | Unreliable 385 | | channel 386 | | | 387 ^ ^ ^ 388 | | | 389 Y Y Y 390 High prio,| | | 391 reliable | | | 392 channel | | | 393 Y Y Y 394 +-+--------+--------+-+ 395 | | 396 | SCTP | 397 | | 398 +---------------------+ 400 Figure 4: The TML-SCTP channels 402 Figure 4 details further the interfacing between the TML core and 403 SCTP layers. There are 3 channels used to separate and prioritize 404 the different types of ForCES traffic. Each channel constitutes a 405 socket interface. It should be noted that all SCTP channels are 406 congestion aware (and for that reason that detail is left out of the 407 description of the 3 channels). SCTP port 6700, 6701, 6702 are used 408 for the higher, medium and lower priority channels respectively. 410 4.2.1.1. Justifying Choice of 3 Sockets 412 SCTP allows up to 64K streams to be sent over a single socket 413 interface. The authors initially envisioned using a single socket 414 for all three channels (mapping a channel to an SCTP stream). This 415 simplifies programming of the TML as well as conserves use of SCTP 416 ports. 418 Further analysis revealed head of line blocking issues with this 419 initial approach. Lower priority packets not needing reliable 420 delivery could block higher priority packets (needing reliable 421 delivery) under congestion situation for an indeterminate period of 422 time (depending on how many outstanding lower priority packets are 423 pending). For this reason, we elected to go with mapping each of the 424 three channels to a different SCTP socket (instead of a different 425 stream within a single socket). 427 4.2.1.2. Higher Priority, Reliable channel 429 The higher priority (HP) channel uses a standard SCTP reliable socket 430 on port 6700. It is used for CE solicited messages and their 431 responses: 433 1. ForCES configuration messages flowing from CE to FE and responses 434 from the FE to CE. 436 2. ForCES query messages flowing from CE to FE and responses from 437 the FE to the CE. 439 It is recommended that PL priorities 4-7 be used for this channel and 440 that the following PL messages use the HP channel for transport: 442 o Association Setup 444 o Association Setup Response 446 o Association Teardown 448 o Config 450 o Config Response 452 o Query 454 o Query Response 456 4.2.1.3. Medium Priority, Semi-Reliable channel 458 The medium priority (MP) channel uses SCTP-PR on port 6701. Time 459 limits on how long a message is valid are set on each outgoing 460 message. This channel is used for events from the FE to the CE that 461 are obsoleted over time. Events that are accumulative in nature and 462 are recoverable by the CE (by issuing a query to the FE) can tolerate 463 lost events and therefore should use this channel. For example, a 464 generated event which carries the value of a counter that is 465 monotonically incrementing fits to use this channel. 467 It is recommended that PL priorities 2-3 be used for this channel and 468 that the following PL messages use the MP channel for transport: 470 o Event Notification 472 4.2.1.4. Lower Priority, Unreliable channel 474 The lower priority (LP) channel uses SCTP port 6702. This channel 475 also uses SCTP-PR with lower timeout values than the MP channel. The 476 reason an unreliable channel is used for redirect messages is to 477 allow the control protocol at both the CE and its peer-endpoint to 478 take charge of how the end-to-end semantics of the said control 479 protocol's operations. For example: 481 1. Some control protocols are reliable in nature, therefore making 482 this channel reliable introduces an extra layer of reliability 483 which could be harmful. So any end-to-end retransmits will 484 happen from remote. 486 2. Some control protocols may desire to have obsolescence of 487 messages over retransmissions; making this channel reliable 488 contradicts that desire. 490 Given ForCES PL level heartbeats are traffic sensitive, sending them 491 over the LP channel also makes sense. If the other end is not 492 processing other channels it will eventually get heartbeats; and if 493 it is busy processing other channels heartbeats will be obsoleted 494 locally over time (and it does not matter if they did not make it). 496 It is recommended that PL priorities 0-1 be used for this channel and 497 that that the following PL messages use the LP channel for transport: 499 o Packet Redirect 501 o Heartbeats 503 4.2.1.5. Scheduling of The 3 Channels 505 Strict priority work-conserving scheduling is used to process both on 506 sending and receiving (of the PL messages) by the TML Core as shown 507 in Figure 5. 509 This means that the HP messages are always processed first until 510 there are no more left. The LP channel is processed only if a 511 channel that is higher priority than itself has no more messages left 512 to process. This means that under congestion situation, a higher 513 priority channel with sufficient messages that occupy the available 514 bandwidth would starve lower priority channel(s). 516 The design intent of the SCTP TML is to tie prioritization as 517 described in Section 4.2.1.1 and transport congestion control to 518 provide implicit node congestion control. This is further detailed 519 in Section 5.2. 521 SCTP channel +----------+ 522 Work available | DONE +---<--<--+ 523 | +---+------+ | 524 Y ^ 525 | +-->--+ +-->---+ | 526 +-->-->-+ | | | | | 527 | | | | | | ^ 528 | ^ ^ Y ^ Y | 529 ^ / \ | | | | | 530 | / \ | ^ | ^ ^ 531 | / Is \ | / \ | / \ | 532 | / there \ | /Is \ | /Is \ | 533 ^ / HP work \ ^ /there\ ^ /there\ ^ 534 | \ ? / | /MP work\ | /LP work\ | 535 | \ / | \ ? / | \ ? / | 536 | \ / | \ / | \ / ^ 537 | \ / ^ \ / ^ \ / | 538 | \ / | \ / | \ / | 539 ^ Y-->-->-->+ Y-->-->-->+ Y->->->-+ 540 | | NO | NO | NO 541 | | | | 542 | Y Y Y 543 | | YES | YES | 544 ^ | | | 545 | Y Y Y 546 | +----+------+ +---|-------+ +----|------+ 547 | |- process | |- process | |- process | 548 | | HP work | | MP work | | LP work | 549 | +------+----+ +-----+-----+ +-----+-----+ 550 | | | | 551 ^ Y Y Y 552 | | | | 553 | Y Y Y 554 +--<--<---+--<--<----<----+-----<---<-----+ 556 Figure 5: SCTP TML Strict Priority Scheduling 558 4.2.1.6. SCTP TML Parameterization 560 The following is a list of parameters needed for booting the TML. It 561 is expected these parameters will be extracted via the FEM/CEM 562 interface for each PL ID. 564 1. The IP address or a resolvable DNS/hostname of the CE/FE. 566 2. Whether to use IPsec or not. If IPsec is used, how to 567 parameterize the different required ciphers, keys etc as 568 described in Section 8.1 570 3. The HP SCTP port, as discussed in Section 4.2.1.2. The default 571 HP port value is 6700 (Section 7). 573 4. The MP SCTP port, as discussed in Section 4.2.1.3. The default 574 MP port value is 6701 (Section 7). 576 5. The LP SCTP port, as discussed in Section 4.2.1.4. The default 577 LP port value is 6702 (Section 7). 579 4.2.2. Satisfying TML Requirements 581 [FE-PROTO] section 5 lists requirements that a TML needs to meet. 582 This section describes how the SCTP TML satisfies those requirements. 584 4.2.2.1. Satisfying Reliability Requirement 586 As mentioned earlier, a shade of reliability ranges is possible in 587 SCTP. Therefore this requirement is met. 589 4.2.2.2. Satisfying Congestion Control Requirement 591 Congestion control is built into SCTP. Therefore, this requirement 592 is met. 594 4.2.2.3. Satisfying Timeliness and Prioritization Requirement 596 By using 3 sockets in conjunction with the partial-reliability 597 feature, both timeliness and prioritization can be achieved. 599 4.2.2.4. Satisfying Addressing Requirement 601 There are no extra headers required for SCTP to fulfil this 602 requirement. SCTP can be told to replicast packets to multiple 603 destinations. The TML implementation will need to translate PL level 604 addresses, to a variety of unicast IP addresses in order to emulate 605 multicast and broadcast PL addresses. 607 4.2.2.5. Satisfying HA Requirement 609 Transport link resiliency is one of SCTP's strongest point. Failure 610 detection and recovery is built in, as mentioned earlier. 612 o The SCTP multi-homing feature is used to provide path diversity. 613 Should one of the peer IP addresses become unreachable, the 614 other(s) are used without needing lower layer convergence 615 (routing, for example) or even the TML becoming aware. 617 o SCTP heartbeats and data transmission thresholds are used on a per 618 peer IP address to detect reachability faults. The faults could 619 be a result of an unreachable address or peer, which may be caused 620 by a variety of reasons, like interface, network, or endpoint 621 failures. The cause of the fault is noted. 623 o With the ADDIP feature, one can migrate IP addresses to other 624 nodes at runtime. This is not unlike the VRRP[RFC3768] protocol 625 use. This feature is used in addition to multi-homing in a 626 planned migration of activity from one FE/CE to another. In such 627 a case, part of the provisioning recipe at the CE for replacing an 628 FE involves migrating activity of one FE to another. 630 4.2.2.6. Satisfying Node Overload Prevention Requirement 632 The architecture of this TML defines three separate channels, one per 633 socket, to be used within any FE-CE setup. The scheduling design for 634 processing the TML channels (Section 4.2.1.5) is strict priority. A 635 fundamental desire of the strict prioritization is to ensure that 636 more important work always gets node resources such as CPU and 637 bandwidth over lesser important work. 639 When a ForCES node CPU is overwhelmed because the incoming packet 640 rate is higher than it can keep up with, the channel queues grow and 641 transport congestion subsequently follows. By virtue of using SCTP, 642 the congestion is propagated back to the source of the incoming 643 packets and eventually alleviated. 645 The HP channel work gets prioritized at the expense of the MP which 646 gets prioritized over LP channels. The preferential scheduling only 647 kicks in when there is node overload regardless of whether there is 648 transport congestion. As a result of the preferential work 649 treatment, the ForCES node achieves a robust steady processing 650 capacity. Refer to Section 5.2 for details on scheduling. 652 For an example of how the overload prevention works: consider a 653 scenario where an overwhelming amount redirected packets (from 654 outside the NE) coming into the NE may overload the FE while it has 655 outstanding config work from the CE. In such a case, the FE, while 656 it is busy processing config requests from the CE ignores processing 657 the redirect packets on the LP channel. If enough redirect packets 658 accumulate, they are dropped either because the LP channel threshold 659 is exceeded or because they are obsoleted. If on the other hand, the 660 FE has successfully processed the higher priority channels and their 661 related work, then it can proceed and process the LP channel. So as 662 demonstrated in this case, the TML ties transport and node overload 663 implicitly together. 665 4.2.2.7. Satisfying Encapsulation Requirement 667 There is no extra encapsulation added by the SCTP TML. 669 In the future, should the need arise, a new SCTP extension/chunk can 670 be defined to meet newer ForCES requirements [RFC4960]. 672 5. SCTP TML Channel work 674 There are two levels of TML channel work within an NE when a ForCES 675 node (CE or FE) is connected to multiple other ForCES nodes: 677 1. NE-level I/O work where a ForCES node (CE or FE) needs to choose 678 which of the peer nodes to process. 680 2. Node-level I/O work where a ForCES node, handles the three SCTP 681 TML channels separately for each single ForCES endpoint. 683 NE-level scheduling definition is left up to the implementation and 684 is considered out of scope for this document. Section 5.4 discuss 685 briefly some constraints that an implementor needs to worry about. 687 This document and in particular Section 5.1, Section 5.2 and 688 Section 5.3 discuss details of node-level I/O work. 690 5.1. SCTP TML Channel Initialization 692 It is recommended that the FE SHOULD do socket connections to the CE 693 in the order of incrementing priorities i.e. LP socket first, 694 followed by MP and ending with HP socket connection. The CE, 695 however, MUST NOT assume that there is ordering of socket connections 696 from any FE. Section 6.1 has more details on the expected 697 initialization of SCTP channel work. 699 5.2. Channel work scheduling 701 This section provides high level details of the scheduling view of 702 the SCTP TML core (Section 4.2.1). A practical scheduler 703 implementation takes care of many little details (such as timers, 704 work quanta, etc) not described in this document. The implementor is 705 left to take care of those details. 707 The CE(s) and FE(s) are coupled together in the principles of the 708 scheduling scheme described here to tie together node overload with 709 transport congestion. The design intent is to provide the highest 710 possible robust work throughput for the NE under any network or 711 processing congestion. 713 5.2.1. FE Channel work scheduling 715 The FE scheduling, in priority order, needs to I/O process: 717 1. The HP channel I/O in the following priority order: 719 1. Transmitting back to the CE any outstanding result of 720 executed work via the HP channel transmit path. 722 2. Taking new incoming work from the CE which creates ForCES 723 work to be executed by the FE. 725 2. ForCES events which result in transmission of unsolicited ForCES 726 packets to the CE via the MP channel. 728 3. Incoming Redirect work in the form of control packets that come 729 from the CE via LP channel. After redirect processing, these 730 packets get sent out on external (to the NE) interface. 732 4. Incoming Redirect work in the form of control packets that come 733 from other NEs via external (to the NE) interfaces. After some 734 processing, such packets are sent to the CE. 736 It is worth emphasizing at this point again that the SCTP TML 737 processes the channel work in strict priority. For example, as long 738 as there are messages to send to the CE on the HP channel, they will 739 be processed first until there are no more left before processing the 740 next priority work (which is to read new messages on the HP channel 741 incoming from the CE). 743 5.2.2. CE Channel work scheduling 745 The CE scheduling, in priority order, needs to deal with: 747 1. The HP channel I/O in the following priority order: 749 1. Process incoming responses to requests of work it made to the 750 FE(s). 752 2. Transmitting any outstanding HP work it needs for the FE(s) 753 to complete. 755 2. Incoming ForCES events from the FE(s) via the MP channel. 757 3. Outgoing Redirect work in the form of control packets that get 758 sent from the CE via LP channel destined to external (to the NE) 759 interface on FE(s). 761 4. Incoming Redirect work in the form of control packets that come 762 from other NEs via external (to the NE) interfaces on the FE(s). 764 It is worth to repeat for emphasis again that the SCTP TML processes 765 the channel work in strict priority. For example, if there are 766 messages incoming from an FE on the HP channel, they will be 767 processed first until there are no more left before processing the 768 next priority work which is to transmit any outstanding HP channel 769 messages going to the FE. 771 5.3. SCTP TML Channel Termination 773 Section 6.2 describes a controlled disassociation of the FE from the 774 NE. 776 It is also possible for connectivity to be lost between the FE and CE 777 on one or more sockets. In cases where SCTP multi-homing features 778 are used for path availability, the disconnection of a socket will 779 only occur if all paths are unreachable; otherwise, SCTP will ensure 780 reachability. In the situation of a total connectivity loss of even 781 one SCTP socket, it is recommended that the FE and CE SHOULD assume a 782 state equivalent to ForCES Association Teardown being issued and 783 follow the sequence described in Section 6.2. 785 A CE could also disconnect sockets to an FE to indicate an "emergency 786 teardown". The "emergency teardown" may be necessary in cases when a 787 CE needs to disconnect an FE but knows that an FE is busy processing 788 a lot of outstanding commands (some of which the FE hasn't got around 789 to processing yet). By virtue of the CE closing the connections, the 790 FE will immediately be asynchronously notified and will not have to 791 process any outstanding commands from the CE. 793 5.4. SCTP TML NE level channel scheduling 795 In handling NE-level I/O work, an implementation needs to worry about 796 being both fair and robust across peer ForCES nodes. 798 Fairness is desired so that each peer node makes progress across the 799 NE. For the sake of illustration consider two FEs connected to a CE; 800 whereas one FE has a few HP messages that need to be processed by the 801 CE, another may have infinite HP messages. The scheduling scheme may 802 decide to use a quota scheduling system to ensure that the second FE 803 does not hog the CE cycles. 805 Robustness is desired so that the NE does not succumb to a DoS attack 806 from hostile entities and always achieves a maximum stable workload 807 processing level. For the sake of illustration consider again two 808 FEs connected to a CE. Consider FE1 as having a large number of HP 809 and MP messages and FE2 having a large number of MP and LP messages. 810 The scheduling scheme needs to ensure that while FE1 always gets its 811 messages processed, at some point we allow FE2 messages to be 812 processed. A promotion and preemption based scheduling could be used 813 by the CE to resolve this issue. 815 6. Service Interface 817 This section provides high level service interface between FEM/CEM 818 and TML, the PL and TML, and between local and remote TMLs. The 819 intent of this interface discussion is to provide general guidelines. 820 The implementer is expected to worry about details and even follow a 821 different approach if needed. 823 The theory of operation for the PL-TML service is as follows: 825 1. The PL starts up and bootstraps the TML. The end result of a 826 successful TML bootstrap is that the CE TML and the FE TML 827 connect to each other at the transport level. 829 2. Sending and reception of the PL level messages commences after a 830 successful TML bootstrap. The PL uses send and receive PL-TML 831 interfaces to communicate to its peers. The TML is agnostic to 832 the nature of the messages being sent or received. The first 833 message exchanges that happen are to establish ForCES 834 association. Subsequent messages maybe either unsolicited events 835 from the FE PL, control message redirects from/to the CE to/from 836 FE, and configuration from the CE to the FE and their responses 837 flowing from the FE to the CE. 839 3. The PL does a shutdown of the TML after terminating ForCES 840 association. 842 6.1. TML Boot-strapping 844 Figure 6 illustrates a flow for the TML bootstrapped by the PL. 846 When the PL starts up (possibly after some internal initialization), 847 it boots up the TML. The TML first interacts with the FEM/CEM and 848 acquires the necessary TML parameterization (Section 4.2.1.6). Next 849 the TML uses the information it retrieved from the FEM/CEM interface 850 to initialize itself. 852 The TML on the FE proceeds to connect the 3 channels to the CE. The 853 socket interface is used for each of the channels. The TML continues 854 to re-try the connections to the CE until all 3 channels are 855 connected. It is advisable that the number of connection retry 856 attempts and the time between each retry is also configurable via the 857 FEM. On failure to connect one or more channels, and after the 858 configured number of retry thresholds is exceeded, the TML will 859 return an appropriate failure indicator to the PL. On success (as 860 shown in Figure 6), a success indication is presented to the TML. 862 FE PL FE TML FEM CEM CE TML CE PL 863 | | | | | | 864 | | | | | Bootup | 865 | | | | |<-------------------| 866 | Bootup | | | | | 867 |----------->| | |get CEM info| | 868 | |get FEM info | |<-----------| | 869 | |------------>| ~ ~ | 870 | ~ ~ |----------->| | 871 | |<------------| | | 872 | | |-initialize TML | 873 | | |-create the 3 chans.| 874 | | | to listen to FEs | 875 | | | | 876 | |-initialize TML |Bootup success | 877 | |-create the 3 chans. locally |------------------->| 878 | |-connect 3 chans. remotely | | 879 | |------------------------------>| | 880 | ~ ~ - FE TML connected ~ 881 | ~ ~ - FE TML info init ~ 882 | | channels connected | | 883 | |<------------------------------| | 884 | Bootup | | | 885 | succeeded | | | 886 |<-----------| | | 887 | | | | 889 Figure 6: SCTP TML Bootstrapping 891 On the CE things are slightly different. After initializing from the 892 CEM, the TML on the CE side proceeds to initialize the 3 channels to 893 listen to remote connections from the FEs. The success or failure 894 indication is passed on to the CE PL level (in the same manner as was 895 done in the FE). 897 Post boot-up, the CE TML waits for connections from the FEs. Upon a 898 successful connection by an FE, the CE TML level keeps track of the 899 transport level details of the FE. Note, at this stage only 900 transport level connection has been established; ForCES level 901 association follows using send/receive PL-TML interfaces (refer to 902 Section 6.3 and Figure 8). 904 6.2. TML Shutdown 906 Figure 7 shows an example of an FE shutting down the TML. It is 907 assumed at this point that the ForCES Association Teardown has been 908 issued by the CE. 910 When the FE PL issues a shutdown to its TML for a specific PL ID, the 911 TML releases all the channel connections to the CE. This is achieved 912 by closing the sockets used to communicate to the CE. 914 FE PL FE TML CE TML CE PL 915 | | | | 916 | Shutdown | | | 917 |----------->| | | 918 | |-disconnect 3 chans. | | 919 | |------------------------>| | 920 | | | | 921 | | |-FE TML info cleanup| 922 | | |-optionally tell PL | 923 | | |------------------->| 924 | |- clean up any state of | | 925 | | channels disconnected | | 926 | | | | 927 | |<------------------------| | 928 | Shutdown | | | 929 | succeeded | | | 930 |<-----------| | | 931 | | | | 933 Figure 7: FE Shutting down 935 On the CE side, a TML level disconnection would result in possible 936 cleanup of the FE state. Optionally, depending on the 937 implementation, there may be need to inform the PL about the TML 938 disconnection. 940 6.3. TML Sending and Receiving 942 The TML is agnostic to the nature of the PL message it delivers to 943 the remote TML (which subsequently delivers the message to its PL). 944 Figure 8 shows an example of a message exchange originated at the FE 945 and sent to the CE (such as a ForCES association message) which 946 illustrates all the necessary service interfaces for sending and 947 receiving. 949 When the FE PL sends a message to the TML, the TML is expected to 950 pick one of HP/MP/LP channels and send out the ForCES message. 952 FE PL FE TML CE TML CE PL 953 | | | | 954 |PL send | | | 955 |----------->| | | 956 | | | | 957 | |-Format msg. | | 958 | |-pick channel | | 959 | |-TML Send | | 960 | |------------->| | 961 | | |-TML Receive on chan. | 962 | | |-decapsulate | 963 | | |- mux to PL/PL recv | 964 | | |--------------------->| 965 | | | ~ 966 | | | ~ PL Process 967 | | | ~ 968 | | | PL send | 969 | | |<---------------------| 970 | | |-Format msg. for send | 971 | | |-pick chan to send on | 972 | | |-TML send | 973 | |<-------------| | 974 | |-TML Receive | | 975 | |-decapsulate | | 976 | |-mux to PL | | 977 | PL Recv | | | 978 |<---------- | | | 979 | | | | 981 Figure 8: Send and Recv Flow 983 When the CE TML receives the ForCES message on the channel it was 984 sent on, it demultiplexes the message to the CE PL. 986 The CE PL, after some processing (in this example dealing with the 987 FE's association), sends to the TML the response. And as in the case 988 of FE PL, the CE TML picks the channel to send on before sending. 990 The processing of the ForCES message upon arriving at the FE TML and 991 delivery to the FE PL is similar to the CE side equivalent as shown 992 above in Section 6.3. 994 7. IANA Considerations 996 This document makes request of IANA to reserve SCTP ports 6700, 6701, 997 and 6702. 999 8. Security Considerations 1001 The SCTP TML provides the following security services to the PL 1002 level: 1004 o A mechanism to authenticate ForCES CEs and FEs at transport level 1005 in order to prevent the participation of unauthorized CEs and 1006 unauthorized FEs in the control and data path processing of a 1007 ForCES NE. 1009 o A mechanism to ensure message authentication of PL data and 1010 headers transferred from the CE to FE (and vice-versa) in order to 1011 prevent the injection of incorrect data into PL messages. 1013 o A mechanism to ensure the confidentiality of PL data and headers 1014 transferred from the CE to FE (and vice-versa), in order to 1015 prevent disclosure of PL level information transported via the 1016 TML. 1018 Security choices provided by the TML are made by the operator and 1019 take effect during the pre-association phase of the ForCES protocol. 1020 An operator may choose to use all, some or none of the security 1021 services provided by the TML in a CE-FE connection. 1023 When operating under a secured environment, or for other operational 1024 concerns (in some cases performance issues) the operator may turn off 1025 all the security functions between CE and FE. 1027 IP Security Protocol (IPsec) [RFC4301] is used to provide needed 1028 security mechanisms. 1030 IPsec is an IP level security scheme transparent to the higher-layer 1031 applications and therefore can provide security for any transport 1032 layer protocol. This gives IPsec the advantage that it can be used 1033 to secure everything between the CE and FE without expecting the TML 1034 implementation to be aware of the details. 1036 The IPsec architecture is designed to provide message integrity and 1037 message confidentiality outlined in the TML security requirements 1038 ([FE-PROTO]). Mutual authentication and key exchange protocol are 1039 provided by Internet Key Exchange (IKE)[RFC4109]. 1041 8.1. IPsec Usage 1043 A ForCES FE or CE MUST support the following: 1045 o Internet Key Exchange (IKE)[RFC4109] with certificates for 1046 endpoint authentication. 1048 o Transport Mode Encapsulating Security Payload (ESP)[RFC4303]. 1050 o HMAC-SHA1-96 [RFC2404] for message integrity protection 1052 o AES-CBC with 128-bit keys [RFC3602] for message confidentiality. 1054 o Replay protection[RFC4301]. 1056 It is expected to be possible for the CE or FE to be operationally 1057 configured to negotiate other cipher suites and even use manual 1058 keying. 1060 8.1.1. SAD and SPD setup 1062 To minimize the operational configuration it is recommended that only 1063 the IANA issued SCTP protocol number(132) be used as a selector in 1064 the Security Policy Database (SPD) for ForCES. In such a case only a 1065 single SPD and SAD entry is needed. 1067 It should be straightforward to extend such a policy to alternatively 1068 use the 3 SCTP TML port numbers as SPD selectors. But as noted above 1069 this choice will require increased number of SPD entries. 1071 In scenarios where multiple IP addresses are used within a single 1072 association, and there is desire to configure different policies on a 1073 per IP address, then it is recommended to follow [RFC3554] 1075 9. Acknowledgements 1077 The authors would like to thank Joel Halpern, Michael Tuxen, Randy 1078 Stewart and Evangelos Haleplidis for engaging us in discussions that 1079 have made this draft better. 1081 10. References 1082 10.1. Normative References 1084 [RFC2404] Madson, C. and R. Glenn, "The Use of HMAC-SHA-1-96 within 1085 ESP and AH", RFC 2404, November 1998. 1087 [RFC3554] Bellovin, S., Ioannidis, J., Keromytis, A., and R. 1088 Stewart, "On the Use of Stream Control Transmission 1089 Protocol (SCTP) with IPsec", RFC 3554, July 2003. 1091 [RFC3602] Frankel, S., Glenn, R., and S. Kelly, "The AES-CBC Cipher 1092 Algorithm and Its Use with IPsec", RFC 3602, 1093 September 2003. 1095 [RFC4109] Hoffman, P., "Algorithms for Internet Key Exchange version 1096 1 (IKEv1)", RFC 4109, May 2005. 1098 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 1099 Internet Protocol", RFC 4301, December 2005. 1101 [RFC4303] Kent, S., "IP Encapsulating Security Payload (ESP)", 1102 RFC 4303, December 2005. 1104 [RFC4960] Stewart, R., "Stream Control Transmission Protocol", 1105 RFC 4960, September 2007. 1107 [RFC5061] Stewart, R., Xie, Q., Tuexen, M., Maruyama, S., and M. 1108 Kozuka, "Stream Control Transmission Protocol (SCTP) 1109 Dynamic Address Reconfiguration", RFC 5061, 1110 September 2007. 1112 10.2. Informative References 1114 [FE-MODEL] 1115 Halpern, J. and J. Hadi Salim, "ForCES Forwarding Element 1116 Model", October 2008. 1118 [FE-PROTO] 1119 Doria (Ed.), A., Haas (Ed.), R., Hadi Salim (Ed.), J., 1120 Khosravi (Ed.), H., M. Wang (Ed.), W., Dong, L., and R. 1121 Gopal, "ForCES Protocol Specification", November 2008. 1123 [RFC3654] Khosravi, H. and T. Anderson, "Requirements for Separation 1124 of IP Control and Forwarding", RFC 3654, November 2003. 1126 [RFC3746] Yang, L., Dantu, R., Anderson, T., and R. Gopal, 1127 "Forwarding and Control Element Separation (ForCES) 1128 Framework", RFC 3746, April 2004. 1130 [RFC3768] Hinden, R., "Virtual Router Redundancy Protocol (VRRP)", 1131 RFC 3768, April 2004. 1133 [SCTP-API] 1134 Stewart, R., Poon, K., Tuexen, M., Yasevich, V., and P. 1135 Lei, "Sockets API Extensions for Stream Control 1136 Transmission Protocol (SCTP)", Feb. 2009. 1138 Authors' Addresses 1140 Jamal Hadi Salim 1141 Mojatatu Networks 1142 Ottawa, Ontario 1143 Canada 1145 Email: hadi@mojatatu.com 1147 Kentaro Ogawa 1148 NTT Corporation 1149 3-9-11 Midori-cho 1150 Musashino-shi, Tokyo 180-8585 1151 Japan 1153 Email: ogawa.kentaro@lab.ntt.co.jp