idnits 2.17.1 draft-ietf-soc-overload-control-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 641 has weird spacing: '...control param...' == The document seems to use 'NOT RECOMMENDED' as an RFC 2119 keyword, but does not include the phrase in its RFC 2119 key words list. -- The document date (October 28, 2011) is 4535 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) -- Looks like a reference, but probably isn't: '0' on line 686 -- Looks like a reference, but probably isn't: '100' on line 686 == Outdated reference: A later version (-13) exists of draft-ietf-soc-load-control-event-package-01 == Outdated reference: A later version (-02) exists of draft-noel-soc-overload-rate-control-01 Summary: 0 errors (**), 0 flaws (~~), 5 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 SOC Working Group V. Gurbani, Ed. 3 Internet-Draft Bell Laboratories, Alcatel-Lucent 4 Intended status: Standards Track V. Hilt 5 Expires: April 30, 2012 Bell Labs/Alcatel-Lucent 6 H. Schulzrinne 7 Columbia University 8 October 28, 2011 10 Session Initiation Protocol (SIP) Overload Control 11 draft-ietf-soc-overload-control-04 13 Abstract 15 Overload occurs in Session Initiation Protocol (SIP) networks when 16 SIP servers have insufficient resources to handle all SIP messages 17 they receive. Even though the SIP protocol provides a limited 18 overload control mechanism through its 503 (Service Unavailable) 19 response code, SIP servers are still vulnerable to overload. This 20 document defines the behaviour of SIP servers involved in overload 21 control, and in addition, it specifies a loss-based overload scheme 22 for SIP. 24 Status of this Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at http://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on April 30, 2012. 41 Copyright Notice 43 Copyright (c) 2011 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (http://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 59 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 5 60 3. Overview of operations . . . . . . . . . . . . . . . . . . . . 5 61 4. Via header parameters for overload control . . . . . . . . . . 5 62 4.1. The oc parameter . . . . . . . . . . . . . . . . . . . . . 6 63 4.2. The oc-algo parameter . . . . . . . . . . . . . . . . . . 6 64 4.3. The oc-validity parameter . . . . . . . . . . . . . . . . 7 65 4.4. The oc-seq parameter . . . . . . . . . . . . . . . . . . . 7 66 5. General behaviour . . . . . . . . . . . . . . . . . . . . . . 7 67 5.1. Handshake to determine support for overload control . . . 8 68 5.2. Creating and updating the overload control parameters . . 9 69 5.3. Determining the 'oc' Parameter Value . . . . . . . . . . . 10 70 5.4. Processing the Overload Control Parameters . . . . . . . . 11 71 5.5. Using the Overload Control Parameter Values . . . . . . . 11 72 5.6. Forwarding the overload control parameters . . . . . . . . 12 73 5.7. Terminating overload control . . . . . . . . . . . . . . . 12 74 5.8. Stabilizing overload control . . . . . . . . . . . . . . . 13 75 5.9. Self-Limiting . . . . . . . . . . . . . . . . . . . . . . 13 76 5.10. Responding to an Overload Indication . . . . . . . . . . . 14 77 5.10.1. Message prioritization at the hop before the 78 overloaded server . . . . . . . . . . . . . . . . . . 14 79 5.10.2. Rejecting requests at an overloaded server . . . . . 14 80 5.11. 100-Trying provisional response and overload control 81 parameters . . . . . . . . . . . . . . . . . . . . . . . . 15 82 6. The loss-based overload control scheme . . . . . . . . . . . . 15 83 6.1. Special parameter values for loss-based overload 84 control . . . . . . . . . . . . . . . . . . . . . . . . . 16 85 6.2. Example . . . . . . . . . . . . . . . . . . . . . . . . . 16 86 6.3. Default algorithm for loss-based overload control . . . . 18 87 7. Relationship with other IETF SIP load control efforts . . . . 20 88 8. Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 89 9. Design Considerations . . . . . . . . . . . . . . . . . . . . 20 90 9.1. SIP Mechanism . . . . . . . . . . . . . . . . . . . . . . 20 91 9.1.1. SIP Response Header . . . . . . . . . . . . . . . . . 20 92 9.1.2. SIP Event Package . . . . . . . . . . . . . . . . . . 21 93 9.2. Backwards Compatibility . . . . . . . . . . . . . . . . . 22 94 10. Security Considerations . . . . . . . . . . . . . . . . . . . 23 95 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 24 96 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 24 97 12.1. Normative References . . . . . . . . . . . . . . . . . . . 24 98 12.2. Informative References . . . . . . . . . . . . . . . . . . 25 99 Appendix A. Acknowledgements . . . . . . . . . . . . . . . . . . 25 100 Appendix B. RFC5390 requirements . . . . . . . . . . . . . . . . 25 101 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 31 103 1. Introduction 105 As with any network element, a Session Initiation Protocol (SIP) 106 [RFC3261] server can suffer from overload when the number of SIP 107 messages it receives exceeds the number of messages it can process. 108 Overload can pose a serious problem for a network of SIP servers. 109 During periods of overload, the throughput of a network of SIP 110 servers can be significantly degraded. In fact, overload may lead to 111 a situation in which the throughput drops down to a small fraction of 112 the original processing capacity. This is often called congestion 113 collapse. 115 Overload is said to occur if a SIP server does not have sufficient 116 resources to process all incoming SIP messages. These resources may 117 include CPU processing capacity, memory, network bandwidth, input/ 118 output, or disk resources. 120 For overload control, we only consider failure cases where SIP 121 servers are unable to process all SIP requests due to resource 122 constraints. There are other cases where a SIP server can 123 successfully process incoming requests but has to reject them due to 124 failure conditions unrelated to the SIP server being overloaded. For 125 example, a PSTN gateway that runs out of trunks but still has plenty 126 of capacity to process SIP messages should reject incoming INVITEs 127 using a 488 (Not Acceptable Here) response [RFC4412]. Similarly, a 128 SIP registrar that has lost connectivity to its registration database 129 but is still capable of processing SIP requests should reject 130 REGISTER requests with a 500 (Server Error) response [RFC3261]. 131 Overload control does not apply to these cases and SIP provides 132 appropriate response codes for them. 134 The SIP protocol provides a limited mechanism for overload control 135 through its 503 (Service Unavailable) response code. However, this 136 mechanism cannot prevent overload of a SIP server and it cannot 137 prevent congestion collapse. In fact, the use of the 503 (Service 138 Unavailable) response code may cause traffic to oscillate and to 139 shift between SIP servers and thereby worsen an overload condition. 140 A detailed discussion of the SIP overload problem, the problems with 141 the 503 (Service Unavailable) response code and the requirements for 142 a SIP overload control mechanism can be found in [RFC5390]. 144 This document defines the general behaviour of SIP servers and 145 clients involved in overload control in Section 5. In addition, 146 Section 6 specifies a loss-based overload control scheme. SIP 147 clients and servers conformant to this specification MUST implement 148 the loss-based overload control scheme. They MAY implement other 149 overload control schemes as well. 151 2. Terminology 153 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 154 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 155 document are to be interpreted as described in RFC 2119 [RFC2119]. 157 The normative statements in this specification as they apply to SIP 158 clients and SIP servers assume that both the SIP clients and SIP 159 servers support this specification. If, for instance, only a SIP 160 client supports this specification and not the SIP server, then 161 follows that the normative statements in this specification pertinent 162 to the behavior of a SIP server do not apply to the server that does 163 not support this specification. 165 3. Overview of operations 167 We now explain the overview of how the overload control mechanism 168 operates by introducing the overload control parameters. Section 4 169 provides more details and normative behavior on the parameters listed 170 below. 172 Because overload control is best performed hop-by-hop, the Via 173 parameter is attractive since it allows two adjacent SIP entities to 174 indicate support for, and exchange information associated with 175 overload control. Additional advantages of this choice are discussed 176 in Section 9.1.1. An alternative mechanism using SIP event packages 177 was also considered, and the characteristics of that choice are 178 further outlined in Section 9.1.2. 180 This document defines four new parameters for the SIP Via header for 181 overload control. These parameters provide a mechanism for conveying 182 overload control information between adjacent SIP entities. The "oc" 183 parameter is used by a SIP server to indicate a reduction in the 184 amount of requests arriving at the server. The "oc-algo" parameter 185 contains a token or a list of tokens corresponding to the class of 186 overload control algorithms supported by the client. The server 187 chooses one algorithm from this list. The "oc-validity" parameter 188 establishes a time limit for which overload control is in effect, and 189 the "oc-seq" parameter aids in sequencing the responses at the 190 client. These parameters are discussed in detail in the next 191 section. 193 4. Via header parameters for overload control 195 The four Via header parameters are introduced below. Further context 196 in how to interpret these under various conditions is provided in 197 Section 5. 199 4.1. The oc parameter 201 This parameter is inserted by the SIP client and updated by the SIP 202 server. 204 A SIP client MUST add an "oc" parameter to the topmost Via header it 205 inserts into the SIP request. This provides an indication to 206 downstream neighbors that the client supports overload control. When 207 inserted into a request by a SIP client to indicate support for 208 overload control, there MUST NOT be a value associated with the 209 parameter. 211 The downstream server MUST add a value to the "oc" parameter in the 212 response going upstream. Inclusion of a value to the parameter 213 represents two things: one, upon an initial handshake (see Section 214 X.X), addition of a value by the server to this parameter indicates 215 (to the client) that the downstream server supports overload control 216 as defined in this document. Second, if the value added by the 217 server is non-zero, it indicates (to the client) that the server 218 wants to perform overload control. 220 When a SIP client receives a response with the value in the "oc" 221 parameter filled in, it SHOULD reduce, by the amount indicated, the 222 number of requests going downstream to the SIP server from which it 223 received the response (see Section 5.10 for pertinent discussion on 224 traffic reduction). 226 4.2. The oc-algo parameter 228 This parameter is inserted by the SIP client and updated by the SIP 229 server. 231 A SIP client conformant to this specification MUST add an "oc-algo" 232 parameter to the topmost Via header it inserts into the SIP request. 233 This parameter contains one or more overload control algorithms. A 234 SIP client conformant to this specification MUST support the loss- 235 based overload control scheme and MUST insert the token "loss" as the 236 "oc-algo" parameter value. In addition, the SIP client MAY insert 237 other tokens, separated by a comma, in the "oc-algo" parameter if it 238 supports other overload control schemes such as a rate-based scheme 239 ([I-D.noel-soc-overload-rate-control]). Each element in the comma- 240 separated list corresponds to the class of overload control 241 algorithms supported by the SIP client. When a downstream SIP server 242 receives a request with a choice of overload control algorithms 243 specified in the "oc-algo" parameter value, it MUST choose one 244 algorithm from the list and MUST pare the list down to include the 245 one chosen algorithm. The pared down list consisting of the chosen 246 algorithm MUST be returned to the upstream SIP client in the 247 response. 249 Once a SIP client and a SIP server have converged to a mutually 250 agreeable class of overload control algorithm, the agreed upon class 251 stays in effect for a non-trivial duration of time to allow the 252 overload control algorithm to stabilize its behaviour (see 253 Section 5.8). 255 4.3. The oc-validity parameter 257 This parameter is inserted by the SIP server. 259 This parameter contains a value that indicates an interval of time 260 (measured in milliseconds) that the load reduction specified value of 261 the "oc" parameter should be in effect. The default value of the 262 "oc-validity" parameter is 500 (millisecond). 264 A value of 0 in the "oc-validity" parameter is reserved to denote the 265 event that the server wishes to stop overload control (see 266 Section 5.7 for more information). 268 A non-zero value for the "oc-validity" parameter MUST only be present 269 in conjunction with an "oc" parameter. 271 4.4. The oc-seq parameter 273 This parameter is inserted by the SIP server. 275 This parameter contains a value that indicates the sequence number 276 associated with the "oc" parameter. Some implementations may be 277 capable of updating the overload control information before the 278 validity period specified by the "oc-validity" parameter expires. 279 Such implementations MUST have an increasing value in the "oc-seq" 280 parameter for each response sent to the upstream SIP client. This is 281 to allow the upstream SIP client to properly collate out-of-order 282 responses. 284 5. General behaviour 286 When forwarding a SIP request, a SIP client uses the SIP procedures 287 of [RFC3263] to determine the next hop SIP server. The procedures of 288 [RFC3263] take as input a SIP URI, extract the domain portion of that 289 URI for use as a lookup key, and query the Domain Name Service (DNS) 290 to obtain an ordered set of one or more IP addresses with a port 291 number and transport corresponding to each IP address in this set 292 (the "Expected Output"). 294 After selecting a specific SIP server from the Expected Output, a SIP 295 client compliant to this specification MUST determine if it is 296 operating under overload control mode with the server (see Section 297 Section 5.5 or if this is the initial contact with the server. 299 If the client determines that this is the initial contact with the 300 server, it proceeds in the following manner to determine if the 301 downstream server supports overload control and to choose an overload 302 control algorithm: A client compliant to this specification MUST 303 insert the "oc" parameter without any value, and MUST insert the "oc- 304 algo" parameter with a list of algorithms it supports. This list 305 MUST include "loss" and MAY include other algorithm names approved by 306 IANA and described in corresponding documents. The client transmits 307 the request to the chosen server. 309 If the server supports overload control as described in this 310 document, it MUST set the value of the "oc" parameter in the request 311 to "0". In addition, it MUST choose one algorithm from the list of 312 algorithms in "oc-algo" parameter and echo the chosen parameter as 313 the only value of the "oc-algo" parameter in the response sent back 314 to the client. A server compliant to this specification MAY insert 315 an "oc-validity=0" parameter in the response to further qualify the 316 value inserted in the "oc" parameter. 318 A client that supports the rate-based overload control scheme 319 [I-D.noel-soc-overload-rate-control] will consider a value of "0" 320 in the "oc" parameter as an indication not to send any requests 321 downstream at all. Thus, when the server inserts "oc-validity=0" 322 as well, it is indicating that it does support overload control, 323 but it is not under overload mode right now. 325 5.1. Handshake to determine support for overload control 327 When a client contacts a server whose overload control support is not 328 known, the client MUST insert the "oc" parameter without any value. 329 Additionally, the client MUST insert the "oc-algo" parameter with a 330 list of algorithms it supports for overload control. This list MUST 331 include "loss" and MAY also include other algorithm names approved by 332 IANA and described in their corresponding documents in the future. 334 A server that supports overload control MUST set the value of the 335 "oc" parameter to be 0. In addition, it MUST choose one algorithm 336 from the list of algorithms in the "oc-algo" parameter and echo the 337 chosen algorithm as the sole parameter value in the "oc-algo" 338 parameter. A server that supports overload control MAY insert an 339 "oc-validity=0" parameter in the response to further qualify the 340 value in the "oc" parameter. 342 Note that the rate-based overload control scheme considers "oc=0" 343 as an indication not to send any requests downstream at all. 344 Thus, having the "oc-validity=0" parameter further imparts the 345 semantics that overload control is supported, but turned off (see 346 Section 5.7.) 348 5.2. Creating and updating the overload control parameters 350 A SIP server can provide overload control feedback to its upstream 351 neighbors by providing a value for the "oc" parameter to the topmost 352 Via header field of a SIP response. The topmost Via header is 353 determined after the SIP server has removed its own Via header; i.e., 354 it is the Via header that was generated by the upstream neighbor. 356 Since the topmost Via header of a response will be removed by an 357 upstream neighbor after processing it, overload control feedback 358 contained in the "oc" parameter will not travel beyond the upstream 359 SIP client. A Via header parameter therefore provides hop-by-hop 360 semantics for overload control feedback (see [RFC6357]) even if the 361 next hop neighbor does not support this specification. 363 The "oc" parameter can be used in all response types, including 364 provisional, success and failure responses (please see Section 5.11 365 for special consideration on transporting overload control parameters 366 in a 100-Trying response). A SIP server MAY update the "oc" 367 parameter in all responses it is sending. A SIP server MUST update 368 the "oc" parameter to responses when the transmission of overload 369 control feedback is required by the overload control algorithm to 370 limit the traffic received by the server. I.e., a SIP server MUST 371 update the "oc" parameter when the overload control algorithm sets 372 the value of an "oc" parameter to a value different than the default 373 value. 375 A SIP server that has updated the "oc" parameter to Via header SHOULD 376 also add a "oc-validity" parameter to the same Via header. The "oc- 377 validity" parameter defines the time in milliseconds during which the 378 content (i.e., the overload control feedback) of the "oc" parameter 379 is valid. The default value of the "oc-validity" parameter is 500 380 (millisecond). A SIP server SHOULD use a shorter "oc-validity" time 381 if its overload status varies quickly and MAY use a longer "oc- 382 validity" time if this status is more stable. If the "oc-validity" 383 parameter is not present, its default value is used. The "oc- 384 validity" parameter MUST NOT be used in a Via header that did not 385 originally contain an "oc" parameter when received. Furthermore, 386 when a SIP server receives a request with the topmost Via header 387 containing only an "oc-validity" parameter without the accompanying 388 "oc" parameter, it MUST ignore the "oc-validity" 390 When a SIP server retransmits a response, it SHOULD use the "oc" 391 parameter value and "oc-validity" parameter value consistent with the 392 overload state at the time the retransmitted response is sent. This 393 implies that the values in the "oc" and "oc-validity" parameters may 394 be different then the ones used in previous retransmissions of the 395 response. Due to the fact that responses sent over UDP may be 396 subject to delays in the network and arrive out of order, the "oc- 397 seq" parameter aids in detecting a stale "oc" parameter value. 399 Implementations that are capable of updating the "oc" and "oc- 400 validity" parameter values for retransmissions MUST insert the "oc- 401 seq" parameter. The value of this parameter MUST be a set of numbers 402 drawn from an increasing sequence. 404 Implementations that are not capable of updating the "oc" and "oc- 405 validity" parameter values for retransmissions --- or implementations 406 that do not want to do so because they will have to regenerate the 407 message to be retransmitted --- MUST still insert a "oc-seq" 408 parameter in the first response associated with a transaction; 409 however, they do not have to update the value in subsequent 410 retransmissions. 412 The "oc-validity" and "oc-seq" Via header parameters are only defined 413 in SIP responses and MUST NOT be used in SIP requests. These 414 parameters are only useful to the upstream neighbor of a SIP server 415 (i.e., the entity that is sending requests to the SIP server) since 416 this is the entity that can offload traffic by redirecting/rejecting 417 new requests. If requests are forwarded in both directions between 418 two SIP servers (i.e., the roles of upstream/downstream neighbors 419 change), there are also responses flowing in both directions. Thus, 420 both SIP servers can exchange overload information. 422 Since overload control protects a SIP server from overload, it is 423 RECOMMENDED that a SIP server use the mechanisms described in this 424 specification. However, if a SIP server wanted to limit its overload 425 control capability for privacy reasons, it MAY decide to perform 426 overload control only for requests that are received on a secure 427 transport channel, such as TLS. This enables a SIP server to protect 428 overload control information and ensure that it is only visible to 429 trusted parties. 431 5.3. Determining the 'oc' Parameter Value 433 The value of the "oc" parameter is determined by the overloaded 434 server using any pertinent information at its disposal. The process 435 by which an overloaded server determines the value of the "oc" 436 parameter is considered out of scope for this document. 438 5.4. Processing the Overload Control Parameters 440 A SIP client compliant to this specification SHOULD remove "oc", "oc- 441 validity" and "oc-seq" parameters from all Via headers of a response 442 received, except for the topmost Via header. This prevents overload 443 control parameters that were accidentally or maliciously inserted 444 into Via headers by a downstream SIP server from traveling upstream. 446 A SIP client maintains the "oc" parameter values received along with 447 the address and port number of the SIP servers from which they were 448 received for the duration specified in the "oc-validity" parameter or 449 the default duration. Each time a SIP client receives a response 450 with an "oc" parameter from a downstream SIP server, it overwrites 451 the "oc" value it has currently stored for this server with the new 452 value received. The SIP client restarts the validity period of an 453 "oc" parameter each time a response with an "oc" parameter is 454 received from this server. A stored "oc" parameter value MUST be 455 discarded once it has reached the end of its validity. 457 5.5. Using the Overload Control Parameter Values 459 A SIP client compliant to this specification MUST honor overload 460 control values it receives from downstream neighbors. The SIP client 461 MUST NOT forward more requests to a SIP server than allowed by the 462 current "oc" parameter value from a particular downstream server. 464 When forwarding a SIP request, a SIP client uses the SIP procedures 465 of [RFC3263] to determine the next hop SIP server. The procedures of 466 [RFC3263] take as input a SIP URI, extract the domain portion of that 467 URI for use as a lookup key, and query the Domain Name Service (DNS) 468 to obtain an ordered set of one or more IP addresses with a port 469 number and transport corresponding to each IP address in this set 470 (the "Expected Output"). 472 After selecting a specific SIP server from the Expected Output, the 473 SIP client MUST determine if it already has overload control 474 parameter values for the server chosen from the Expected Output. If 475 the SIP client has a non-expired "oc" parameter value for the server 476 chosen from the Expected Output, and this chosen server is operating 477 in overload control mode. Thus, the SIP client MUST determine if it 478 can or cannot forward the current request to the SIP server depending 479 on the nature of the request and the prevailing overload conditions. 481 The particular algorithm used to determine whether or not to forward 482 a particular SIP request is a matter of local policy, and may take 483 into account a variety of prioritization factors. However, this 484 local policy SHOULD generate the same number of SIP requests as the 485 default algorithm defined by the overload control scheme being used. 487 5.6. Forwarding the overload control parameters 489 A SIP client MAY forward the content of an "oc" parameter it has 490 received from a downstream neighbor on to its upstream neighbor. 491 However, forwarding the content of the "oc" parameter is generally 492 NOT RECOMMENDED and should only be performed if permitted by the 493 configuration of SIP servers. For example, a SIP server that only 494 relays messages between exactly two SIP servers may forward an "oc" 495 parameter. The "oc" parameter is forwarded by copying it from the 496 Via in which it was received into the next Via header (i.e., the Via 497 header that will be on top after processing the response). If an 498 "oc-validity" parameter is present, MUST be copied along with the 499 "oc" parameter. 501 5.7. Terminating overload control 503 A SIP client stops applying overload control to the number of 504 messages forwarded (i.e., it stops reducing the number of messages 505 forwarded) if one of the following events occur: 507 1. The "oc" parameter is set to a value that allows the client to 508 forward all traffic; 509 2. The "oc-validity" period negotiated to put the server and client 510 in overload state expires; 511 3. The client is explicitly told by the server to stop performing 512 overload control using the "oc-validity=0" parameter. 514 A SIP server can decide to terminate overload control by explicitly 515 signaling the client. To do so, the SIP server MUST set the value of 516 the "oc-validity" parameter to 0. The SIP server MUST increment the 517 value of "oc-seq", and SHOULD set the value of the "oc" parameter to 518 0. 520 Note that the loss-based overload control scheme (Section 6) can 521 effectively stop overload control by setting the value of the "oc" 522 parameter to 0. However, the rate-based scheme 523 ([I-D.noel-soc-overload-rate-control]) needs an additional piece 524 of information in the form of "oc-validity=0". 526 When the client receives a response with a higher "oc-seq" number 527 than the one it currently is processing, it checks the "oc-validity" 528 parameter. If the value of the "oc-validity" parameter is 0, the 529 client MUST stop performing overload control of messages destined to 530 the server and the traffic should flow without any reduction. 531 Furthermore, when the value of the "oc-validity" parameter is 0, the 532 client SHOULD disregard the value in the "oc" parameter. 534 5.8. Stabilizing overload control 536 Realities of deployments of SIP necessitate that the overload control 537 algorithm be renegotiated upon a system reboot or a software upgrade. 538 However, frequent renegotiation of the overload control algorithm 539 MUST be avoided. A rapid renegotiation of the overload control 540 algorithm will not benefit the client or the server as such flapping 541 does not allow the chosen algorithm to measure and fine tune its 542 behavior over a period of time. Renegotiation, when desired, is 543 simply accomplished by the SIP client sending a complete list of 544 overload control algorithms it supports in a "oc-algo" parameter in a 545 request going downstream. The downstream server, as before, MUST 546 choose one algorithm from the list and MUST pare the list down to 547 include the one chosen algorithm. The pared down list consisting of 548 the chosen algorithm MUST be returned to the upstream SIP client in 549 the response and stays in effect until the next renegotiation. 551 Once the client and server agree on an overload control algorithm, it 552 MUST remain in effect for at least 3600 seconds (1 hour) before 553 renegotiation occurs. 555 One way to accomplish this involves the client saving the time of 556 the last negotiation in a lookup table, indexed by the server's 557 network identifiers. Renegotiation is only done when the time of 558 the last negotiation has surpassed 3600 seconds. 560 5.9. Self-Limiting 562 In some cases, a SIP client may not receive a response from a server 563 after sending a request. RFC3261 [RFC3261] defines that when a 564 timeout error is received from the transaction layer, it MUST be 565 treated as if a 408 (Request Timeout) status code has been received. 566 If a fatal transport error is reported by the transport layer, it 567 MUST be treated as a 503 (Service Unavailable) status code. 569 In the event of repeated timeouts or fatal transport errors, the SIP 570 client MUST stop sending requests to this server. The SIP client 571 SHOULD occasionally forward a single request to probe if the 572 downstream server is alive. Once a SIP client has successfully 573 transmitted a request to the downstream server, the SIP client can 574 resume normal traffic rates. It should, of course, honor any "oc" 575 parameters it may receive subsequent to resuming normal traffic 576 rates. 578 5.10. Responding to an Overload Indication 580 A SIP client can receive overload control feedback indicating that it 581 needs to reduce the traffic it sends to its downstream server. The 582 client can accomplish this task by sending some of the requests that 583 would have gone to the overloaded element to a different destination. 584 It needs to ensure, however, that this destination is not in overload 585 and capable of processing the extra load. A client can also buffer 586 requests in the hope that the overload condition will resolve quickly 587 and the requests still can be forwarded in time. In many cases, 588 however, it will need to reject these requests. 590 5.10.1. Message prioritization at the hop before the overloaded server 592 During an overload condition, a SIP client needs to prioritize 593 requests and select those requests that need to be rejected or 594 redirected. While this selection is largely a matter of local 595 policy, certain heuristics can be suggested. One, during overload 596 control, the SIP client should preserve existing dialogs as much as 597 possible. This suggests that mid-dialog requests MAY be given 598 preferential treatment. Similarly, requests that result in releasing 599 resources (such as a BYE) MAY also be given preferential treatment. 601 A SIP client SHOULD honor the local policy for prioritizing SIP 602 requests such as policies based on the content of the Resource- 603 Priority header (RPH, RFC4412 [RFC4412]). Specific (namespace.value) 604 RPH contents may indicate high priority requests that should be 605 preserved as much as possible during overload. The RPH contents can 606 also indicate a low-priority request that is eligible to be dropped 607 during times of overload. Other indicators, such as the SOS URN 608 [RFC5031] indicating an emergency request, may also be used for 609 prioritization. 611 Local policy could also include giving precedence to mid-dialog SIP 612 requests (re-INVITEs, UPDATEs, BYEs etc.) in times of overload. A 613 local policy can be expected to combine both the SIP request type and 614 the prioritization markings, and SHOULD be honored when overload 615 conditions prevail. 617 A SIP client SHOULD honor user-level load control filters installed 618 by signaling neighbors [I-D.ietf-soc-load-control-event-package] by 619 sending the SIP messages that matched the filter downstream. 621 5.10.2. Rejecting requests at an overloaded server 623 If the upstream SIP client to the overloaded server does not support 624 overload control, it will continue to direct requests to the 625 overloaded server. Thus, the overloaded server must bear the cost of 626 rejecting some session requests as well as the cost of processing 627 other requests to completion. It would be fair to devote the same 628 amount of processing at the overloaded server to the combination of 629 rejection and processing as the overloaded server would devote to 630 processing requests from an upstream SIP client that supported 631 overload control. This is to ensure that SIP servers that do not 632 support this specification don't receive an unfair advantage over 633 those that do. 635 A SIP server that is under overload and has started to throttle 636 incoming traffic MUST reject this request with a "503 (Service 637 Unavailable)" response without Retry-After header to reject a 638 fraction of requests from upstream neighbors that do not support 639 overload control. 641 5.11. 100-Trying provisional response and overload control parameters 643 The overload control information sent from a SIP server to a client 644 is transported in the responses. While implementations can insert 645 overload control information in any response, special attention 646 should be accorded to overload control information transported in a 647 100-Trying response. 649 Traditionally, the 100-Trying response has been used in SIP to quench 650 retransmissions. In some implementations, the 100-Trying message may 651 not be generated by the transaction user (TU) nor consumed by the TU. 652 In these implementations, the 100-Trying response is generated at the 653 transaction layer and sent to the upstream SIP client. At the 654 receiving SIP client, the 100-Trying is consumed at the transaction 655 layer by inhibiting the retransmission of the corresponding request. 656 Consequently, implementations that insert overload control 657 information in the 100-Trying cannot assume that the upstream SIP 658 client passed the overload control information in the 100-Trying to 659 their corresponding TU. For this reason, implementations that insert 660 overload control information in the 100-Trying MUST re-insert the 661 same (or updated) overload control information in the first non-100 662 response being sent to the upstream SIP client. 664 6. The loss-based overload control scheme 666 A loss percentage enables a SIP server to ask an upstream neighbor to 667 reduce the number of requests it would normally forward to this 668 server by X%. For example, a SIP server can ask an upstream neighbor 669 to reduce the number of requests this neighbor would normally send by 670 10%. The upstream neighbor then redirects or rejects 10% of the 671 traffic that is destined for this server. 673 This section specifies the semantics of the overload control 674 parameters associated with the loss-based overload control scheme. 675 The general behaviour of SIP clients and servers is specified in 676 Section 5 and is applicable to SIP clients and servers that implement 677 loss-based overload control. 679 6.1. Special parameter values for loss-based overload control 681 The loss-based overload control scheme is identified using the token 682 "loss". This token MUST appear in the "oc-algo" parameter. 684 A SIP server, upon entering the overload state, will assign a value 685 to the "oc" parameter. This value MUST be restricted in the range of 686 [0, 100], inclusive. This value MUST be interpreted as a percentage, 687 and the SIP client MUST reduce the number of requests being forwarded 688 to the overloaded server by that amount. The SIP client may use any 689 algorithm that reduces the traffic arriving at the overloaded server 690 by the amount indicated. Such an algorithm SHOULD honor the message 691 prioritization discussion of Section 5.10.1. While a particular 692 algorithm is not subject to standardization, for completeness a 693 default algorithm for loss-based overload control is provided in 694 Section 6.3. 696 When a SIP server receives a request from a client with an "oc" 697 parameter but without a value, and the SIP server is not experiencing 698 overload, it MUST assign a value of 0 to the "oc" parameter in the 699 response. Assigning such a value lets the client know that the 700 server supports overload control and is not currently experiencing 701 overload. 703 When the "oc-validity" parameter is used to signify overload control 704 termination (Section 5.7), the server MUST insert a value of 0 in the 705 "oc-validity" parameter. The server MUST insert a value of 0 in the 706 "oc" parameter as well. When a client receives a response whose "oc- 707 validity" parameter contains a 0, it MUST treat any non-zero value in 708 the "oc" parameter as if it had received a value of 0 in that 709 parameter. 711 6.2. Example 713 Consider a SIP client, P1, which is sending requests to another 714 downstream SIP server, P2. The following snippets of SIP messages 715 demonstrate how the overload control parameters work. 717 INVITE sips:user@example.com SIP/2.0 718 Via: SIP/2.0/TLS p1.example.net; 719 branch=z9hG4bK2d4790.1;received=192.0.2.111;oc; 720 oc-algo="loss,A" 721 ... 723 SIP/2.0 100 Trying 724 Via: SIP/2.0/TLS p1.example.net; 725 branch=z9hG4bK2d4790.1;received=192.0.2.111; 726 oc=0;oc-algo="loss"; 727 ... 729 In the messages above, the first line is sent by P1 to P2. This line 730 is a SIP request; because P1 supports overload control, it inserts 731 the "oc" parameter in the topmost Via header that it created. P1 732 supports two overload control algorithms: loss and some algorithm 733 called "A". 735 The second line --- a SIP response --- shows the topmost Via header 736 amended by P2 according to this specification and sent to P1. 737 Because P2 also supports overload control, it chooses the "loss" 738 based scheme and sends that back to P1 in the "oc-algo" parameter. 739 It also sets the value of "oc" parameter to 0. 741 Had P2 not supported overload control, it would have left the "oc" 742 and "oc-algo" parameters unchanged, thus allowing the client to know 743 that it did not support overload control. 745 At some later time, P2 starts to experience overload. It sends the 746 following SIP message indicating that P1 should decrease the messages 747 arriving to P2 by 20% for 1s. 749 SIP/2.0 180 Ringing 750 Via: SIP/2.0/TLS p1.example.net; 751 branch=z9hG4bK2d4790.3;received=192.0.2.111; 752 oc=20;oc-algo="loss";oc-validity=1000; 753 oc-seq=1282321615.782 754 ... 756 After 500ms, the overload condition at P2 subsides. It then sends 757 out the message below to allow P1 to send all messages destined to 758 P2. 760 SIP/2.0 183 Queued 761 Via: SIP/2.0/TLS p1.example.net; 762 branch=z9hG4bK2d4790.4;received=192.0.2.111; 763 oc=0;oc-algo="loss";oc-validity=0;oc-seq=1282321887.783 764 ... 766 6.3. Default algorithm for loss-based overload control 768 This section describes a default algorithm that a SIP client can to 769 throttle SIP traffic going downstream by the percentage loss value 770 specified in the "oc" parameter. 772 The client maintains two categories of requests; the first category 773 will include requests that are candidates for reduction, and the 774 second category will include requests that are not subject to 775 reduction (except under extenuating circumstances when there aren't 776 any messages in the first category that can be reduced). Section 777 Section 5.10.1 contains normative directives on how to prioritize 778 messages for inclusion in the second category. The remaining 779 messages can be allocated to the first category. 781 The client determines the mix of requests falling into the first 782 category and those falling into the second category. For example, 783 40% of the requests may be eligible for reduction and 60% not 784 eligible (and therefore, must be sent downstream). 786 Under overload condition, the client converts the value of the "oc" 787 parameter to a value that it applies to requests in the first 788 category. As a simple example, if "oc=10" and 40% of the requests 789 should be included in the first category, then: 791 10 / 40 * 100 = 25 793 Or, 25% of the requests in the first category can be reduced to get 794 an overall reduction of 10%. The client uses random discard to 795 achieve the 25% reduction of messages in the first category. 796 Messages in the second category proceed downstream unscathed. To 797 affect the 25% reduction rate from the first category, the client 798 draws a random number between 1 and 100 for the request picked from 799 the first category. If the random number is less than or equal to 800 converted value of the "oc" parameter, the request is not forwarded; 801 otherwise the request is forwarded. 803 A reference algorithm is shown below. 805 cat1 := 40.0 // Category 1 --- subject to reduction 806 cat2 := 100.0 - cat1 // Category 2 --- Not subject to 807 // reduction. 40/60 mix. 808 in_oc := false // Not operating under overload 810 while (true) { 811 sip_msg := get_sip_msg() 812 if (is_response(sip_msg)) { 813 process_msg(sip_msg) 815 } 816 else if (is_request(sip_msg)) { 818 // Determine if server wants to enter overload or is 819 // in overload 820 in_oc := extract_in_oc(sip_msg) 822 // Get validity value 823 oc_validity := extract_oc_validity(sip_msg) 825 // Get oc parameter value 826 oc_value := extract_oc_value(sip_msg) 828 pct_to_reduce := oc_value / cat1 * 100 829 // Example: if oc=10, 830 // server uses 10 / 40 * 100 = 25 or 25% of messages in 831 // Category 1 can be reduced. 833 if (in_oc == false) { 834 process_msg(sip_msg) 835 } 836 else { 838 // Either Category 1 or Category 2 839 assign_msg_to_category(sip_msg) 841 if (oc_validity is in effect) { 842 process_msg(get_msg_from_cat2()) 843 sip_msg := get_msg_from_cat1() 845 // Draw a random number between 1 and 100 846 r := random() 848 if (r <= pct_to_reduce) { 849 // Do not send to server, handle locally by 850 // generating a final response 851 } 852 else { 853 process_msg(sip_msg) 854 } 855 } 856 } 857 } 858 } 860 Note that in the event that there are not enough messages in the 861 first category to reduce, the client may use local policies to target 862 messages in the second category. 864 7. Relationship with other IETF SIP load control efforts 866 The overload control mechanism described in this document is reactive 867 in nature and apart from message prioritization directives listed in 868 Section 5.10.1 the mechanisms described in this draft will not 869 discriminate requests based on user identity, filtering action and 870 arrival time. SIP networks that require pro-active overload control 871 mechanisms can upload user-level load control filters as described in 872 [I-D.ietf-soc-load-control-event-package]. 874 8. Syntax 876 This specification extends the existing definition of the Via header 877 field parameters of [RFC3261] as follows: 879 via-params = via-ttl / via-maddr 880 / via-received / via-branch 881 / oc / oc-validity 882 / oc-seq / oc-algo / via-extension 884 oc = "oc" [EQUAL oc-num] 885 oc-num = 1*DIGIT 886 oc-validity = "oc-validity" [EQUAL delta-ms] 887 oc-seq = "oc-seq" EQUAL 1*12DIGIT "." 1*5DIGIT 888 oc-algo = "oc-algo" EQUAL DQUOTE algo-list *(COMMA algo-list) 889 DQUOTE 890 algo-list = "loss" / *(other-algo) 891 other-algo = %x41-5A / %x61-7A / %x30-39 893 9. Design Considerations 895 This section discusses specific design considerations for the 896 mechanism described in this document. General design considerations 897 for SIP overload control can be found in [RFC6357]. 899 9.1. SIP Mechanism 901 A SIP mechanism is needed to convey overload feedback from the 902 receiving to the sending SIP entity. A number of different 903 alternatives exist to implement such a mechanism. 905 9.1.1. SIP Response Header 907 Overload control information can be transmitted using a new Via 908 header field parameter for overload control. A SIP server can add 909 this header parameter to the responses it is sending upstream to 910 provide overload control feedback to its upstream neighbors. This 911 approach has the following characteristics: 913 o A Via header parameter is light-weight and creates very little 914 overhead. It does not require the transmission of additional 915 messages for overload control and does not increase traffic or 916 processing burdens in an overload situation. 917 o Overload control status can frequently be reported to upstream 918 neighbors since it is a part of a SIP response. This enables the 919 use of this mechanism in scenarios where the overload status needs 920 to be adjusted frequently. It also enables the use of overload 921 control mechanisms that use regular feedback such as window-based 922 overload control. 923 o With a Via header parameter, overload control status is inherent 924 in SIP signaling and is automatically conveyed to all relevant 925 upstream neighbors, i.e., neighbors that are currently 926 contributing traffic. There is no need for a SIP server to 927 specifically track and manage the set of current upstream or 928 downstream neighbors with which it should exchange overload 929 feedback. 930 o Overload status is not conveyed to inactive senders. This avoids 931 the transmission of overload feedback to inactive senders, which 932 do not contribute traffic. If an inactive sender starts to 933 transmit while the receiver is in overload it will receive 934 overload feedback in the first response and can adjust the amount 935 of traffic forwarded accordingly. 936 o A SIP server can limit the distribution of overload control 937 information by only inserting it into responses to known upstream 938 neighbors. A SIP server can use transport level authentication 939 (e.g., via TLS) with its upstream neighbors. 941 9.1.2. SIP Event Package 943 Overload control information can also be conveyed from a receiver to 944 a sender using a new event package. Such an event package enables a 945 sending entity to subscribe to the overload status of its downstream 946 neighbors and receive notifications of overload control status 947 changes in NOTIFY requests. This approach has the following 948 characteristics: 950 o Overload control information is conveyed decoupled from SIP 951 signaling. It enables an overload control manager, which is a 952 separate entity, to monitor the load on other servers and provide 953 overload control feedback to all SIP servers that have set up 954 subscriptions with the controller. 956 o With an event package, a receiver can send updates to senders that 957 are currently inactive. Inactive senders will receive a 958 notification about the overload and can refrain from sending 959 traffic to this neighbor until the overload condition is resolved. 960 The receiver can also notify all potential senders once they are 961 permitted to send traffic again. However, these notifications do 962 generate additional traffic, which adds to the overall load. 963 o A SIP entity needs to set up and maintain overload control 964 subscriptions with all upstream and downstream neighbors. A new 965 subscription needs to be set up before/while a request is 966 transmitted to a new downstream neighbor. Servers can be 967 configured to subscribe at boot time. However, this would require 968 additional protection to avoid the avalanche restart problem for 969 overload control. Subscriptions need to be terminated when they 970 are not needed any more, which can be done, for example, using a 971 timeout mechanism. 972 o A receiver needs to send NOTIFY messages to all subscribed 973 upstream neighbors in a timely manner when the control algorithm 974 requires a change in the control variable (e.g., when a SIP server 975 is in an overload condition). This includes active as well as 976 inactive neighbors. These NOTIFYs add to the amount of traffic 977 that needs to be processed. To ensure that these requests will 978 not be dropped due to overload, a priority mechanism needs to be 979 implemented in all servers these request will pass through. 980 o As overload feedback is sent to all senders in separate messages, 981 this mechanism is not suitable when frequent overload control 982 feedback is needed. 983 o A SIP server can limit the set of senders that can receive 984 overload control information by authenticating subscriptions to 985 this event package. 986 o This approach requires each proxy to implement user agent 987 functionality (UAS and UAC) to manage the subscriptions. 989 9.2. Backwards Compatibility 991 An new overload control mechanism needs to be backwards compatible so 992 that it can be gradually introduced into a network and functions 993 properly if only a fraction of the servers support it. 995 Hop-by-hop overload control (see [RFC6357]) has the advantage that it 996 does not require that all SIP entities in a network support it. It 997 can be used effectively between two adjacent SIP servers if both 998 servers support overload control and does not depend on the support 999 from any other server or user agent. The more SIP servers in a 1000 network support hop-by-hop overload control, the better protected the 1001 network is against occurrences of overload. 1003 A SIP server may have multiple upstream neighbors from which only 1004 some may support overload control. If a server would simply use this 1005 overload control mechanism, only those that support it would reduce 1006 traffic. Others would keep sending at the full rate and benefit from 1007 the throttling by the servers that support overload control. In 1008 other words, upstream neighbors that do not support overload control 1009 would be better off than those that do. 1011 A SIP server should therefore use 5xx responses towards upstream 1012 neighbors that do not support overload control. The server should 1013 reject the same amount of requests with 5xx responses that would be 1014 otherwise be rejected/redirected by the upstream neighbor if it would 1015 support overload control. If the load condition on the server does 1016 not permit the creation of 5xx responses, the server should drop all 1017 requests from servers that do not support overload control. 1019 10. Security Considerations 1021 Overload control mechanisms can be used by an attacker to conduct a 1022 denial-of-service attack on a SIP entity if the attacker can pretend 1023 that the SIP entity is overloaded. When such a forged overload 1024 indication is received by an upstream SIP client, it will stop 1025 sending traffic to the victim. Thus, the victim is subject to a 1026 denial-of-service attack. 1028 An attacker can create forged overload feedback by inserting itself 1029 into the communication between the victim and its upstream neighbors. 1030 The attacker would need to add overload feedback indicating a high 1031 load to the responses passed from the victim to its upstream 1032 neighbor. Proxies can prevent this attack by communicating via TLS. 1033 Since overload feedback has no meaning beyond the next hop, there is 1034 no need to secure the communication over multiple hops. 1036 Another way to conduct an attack is to send a message containing a 1037 high overload feedback value through a proxy that does not support 1038 this extension. If this feedback is added to the second Via headers 1039 (or all Via headers), it will reach the next upstream proxy. If the 1040 attacker can make the recipient believe that the overload status was 1041 created by its direct downstream neighbor (and not by the attacker 1042 further downstream) the recipient stops sending traffic to the 1043 victim. A precondition for this attack is that the victim proxy does 1044 not support this extension since it would not pass through overload 1045 control feedback otherwise. 1047 A malicious SIP entity could gain an advantage by pretending to 1048 support this specification but never reducing the amount of traffic 1049 it forwards to the downstream neighbor. If its downstream neighbor 1050 receives traffic from multiple sources which correctly implement 1051 overload control, the malicious SIP entity would benefit since all 1052 other sources to its downstream neighbor would reduce load. 1054 The solution to this problem depends on the overload control 1055 method. For rate-based and window-based overload control, it is 1056 very easy for a downstream entity to monitor if the upstream 1057 neighbor throttles traffic forwarded as directed. For percentage 1058 throttling this is not always obvious since the load forwarded 1059 depends on the load received by the upstream neighbor. 1061 11. IANA Considerations 1063 This specification defines four new Via header parameters as detailed 1064 below in the "Header Field Parameter and Parameter Values" sub- 1065 registry as per the registry created by [RFC3968]. The required 1066 information is: 1068 Header Field Parameter Name Predefined Values Reference 1069 __________________________________________________________ 1070 Via oc Yes RFCXXXX 1071 Via oc-validity Yes RFCXXXX 1072 Via oc-seq Yes RFCXXXX 1073 Via oc-algo Yes RFCXXXX 1075 RFC XXXX [NOTE TO RFC-EDITOR: Please replace with final RFC 1076 number of this specification.] 1078 NOTE: Do we need to do anything special to register "loss" 1079 as a value for "oc-algo" parameter? 1081 12. References 1083 12.1. Normative References 1085 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1086 Requirement Levels", BCP 14, RFC 2119, March 1997. 1088 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 1089 A., Peterson, J., Sparks, R., Handley, M., and E. 1090 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 1091 June 2002. 1093 [RFC3263] Rosenberg, J. and H. Schulzrinne, "Session Initiation 1094 Protocol (SIP): Locating SIP Servers", RFC 3263, 1095 June 2002. 1097 [RFC3968] Camarillo, G., "The Internet Assigned Number Authority 1098 (IANA) Header Field Parameter Registry for the Session 1099 Initiation Protocol (SIP)", BCP 98, RFC 3968, 1100 December 2004. 1102 [RFC4412] Schulzrinne, H. and J. Polk, "Communications Resource 1103 Priority for the Session Initiation Protocol (SIP)", 1104 RFC 4412, February 2006. 1106 12.2. Informative References 1108 [I-D.ietf-soc-load-control-event-package] 1109 Shen, C., Schulzrinne, H., and A. Koike, "A Session 1110 Initiation Protocol (SIP) Load Control Event Package", 1111 draft-ietf-soc-load-control-event-package-01 (work in 1112 progress), July 2011. 1114 [I-D.noel-soc-overload-rate-control] 1115 Noel, E., Williams, P., and J. Gunn, "Session Initiation 1116 Protocol (SIP) Rate Control", 1117 draft-noel-soc-overload-rate-control-01 (work in 1118 progress), September 2011. 1120 [RFC5031] Schulzrinne, H., "A Uniform Resource Name (URN) for 1121 Emergency and Other Well-Known Services", RFC 5031, 1122 January 2008. 1124 [RFC5390] Rosenberg, J., "Requirements for Management of Overload in 1125 the Session Initiation Protocol", RFC 5390, December 2008. 1127 [RFC6357] Hilt, V., Noel, E., Shen, C., and A. Abdelal, "Design 1128 Considerations for Session Initiation Protocol (SIP) 1129 Overload Control", RFC 6357, August 2011. 1131 Appendix A. Acknowledgements 1133 Many thanks to Bruno Chatras, Keith Drage, Janet Gunn, Rich Terpstra, 1134 Daryl Malas, R. Parthasarathi, Antoine Roly, Jonathan Rosenberg, 1135 Charles Shen, Rahul Srivastava, Padma Valluri, Shaun Bharrat, and 1136 Paul Kyzivat for their contributions to this specification. 1138 Appendix B. RFC5390 requirements 1140 Table 1 provides a summary how this specification fulfills the 1141 requirements of [RFC5390]. A more detailed view on how each 1142 requirements is fulfilled is provided after the table. 1144 +-------------+-------------------+ 1145 | Requirement | Meets requirement | 1146 +-------------+-------------------+ 1147 | REQ 1 | Yes | 1148 | REQ 2 | Yes | 1149 | REQ 3 | Partially | 1150 | REQ 4 | Partially | 1151 | REQ 5 | Partially | 1152 | REQ 6 | Not applicable | 1153 | REQ 7 | Yes | 1154 | REQ 8 | Partially | 1155 | REQ 9 | Yes | 1156 | REQ 10 | Yes | 1157 | REQ 11 | Yes | 1158 | REQ 12 | Yes | 1159 | REQ 13 | Yes | 1160 | REQ 14 | Yes | 1161 | REQ 15 | Yes | 1162 | REQ 16 | Yes | 1163 | REQ 17 | Partially | 1164 | REQ 18 | Yes | 1165 | REQ 19 | Yes | 1166 | REQ 20 | Yes | 1167 | REQ 21 | Yes | 1168 | REQ 22 | Yes | 1169 | REQ 23 | Yes | 1170 +-------------+-------------------+ 1172 Summary of meeting requirements in RFC5390 1174 Table 1 1176 REQ 1: The overload mechanism shall strive to maintain the overall 1177 useful throughput (taking into consideration the quality-of-service 1178 needs of the using applications) of a SIP server at reasonable 1179 levels, even when the incoming load on the network is far in excess 1180 of its capacity. The overall throughput under load is the ultimate 1181 measure of the value of an overload control mechanism. 1183 Meeting REQ 1: Yes, the overload control mechanism allows an 1184 overloaded SIP server to maintain a reasonable level of throughput as 1185 it enters into congestion mode by requesting the upstream clients to 1186 reduce traffic destined downstream. 1188 REQ 2: When a single network element fails, goes into overload, or 1189 suffers from reduced processing capacity, the mechanism should strive 1190 to limit the impact of this on other elements in the network. This 1191 helps to prevent a small-scale failure from becoming a widespread 1192 outage. 1194 Meeting REQ 2: Yes. When a SIP server enters overload mode, it will 1195 request the upstream clients to throttle the traffic destined to it. 1196 As a consequence of this, the overloaded SIP server will itself 1197 generate proportionally less downstream traffic, thereby limiting the 1198 impact on other elements in the network. 1200 REQ 3: The mechanism should seek to minimize the amount of 1201 configuration required in order to work. For example, it is better 1202 to avoid needing to configure a server with its SIP message 1203 throughput, as these kinds of quantities are hard to determine. 1205 Meeting REQ 3: Partially. On the server side, the overload condition 1206 is determined monitoring S (c.f., Section 4 of [RFC6357]) and 1207 reporting a load feedback F as a value to the "oc" parameter. On the 1208 client side, a throttle T is applied to requests going downstream 1209 based on F. This specification does not prescribe any value for S, 1210 nor a particular value for F. The "oc-algo" parameter allows for 1211 automatic convergence to a particular class of overload control 1212 algorithm. There are suggested default values for the "oc-validity" 1213 parameter. 1215 REQ 4: The mechanism must be capable of dealing with elements that do 1216 not support it, so that a network can consist of a mix of elements 1217 that do and don't support it. In other words, the mechanism should 1218 not work only in environments where all elements support it. It is 1219 reasonable to assume that it works better in such environments, of 1220 course. Ideally, there should be incremental improvements in overall 1221 network throughput as increasing numbers of elements in the network 1222 support the mechanism. 1224 Meeting REQ 4: Partially. The mechanism is designed to reduce 1225 congestion when a pair of communicating entities support it. If a 1226 downstream overloaded SIP server does not respond to a request in 1227 time, a SIP client conformant to this specification will attempt to 1228 reduce traffic destined towards the non-responsive server as outlined 1229 in Section 5.9. 1231 REQ 5: The mechanism should not assume that it will only be deployed 1232 in environments with completely trusted elements. It should seek to 1233 operate as effectively as possible in environments where other 1234 elements are malicious; this includes preventing malicious elements 1235 from obtaining more than a fair share of service. 1237 Meeting REQ 5: Partially. Since overload control information is 1238 shared between a pair of communicating entities, a confidential and 1239 authenticated channel can be used for this communication. However, 1240 if such a channel is not available, then the security ramifications 1241 outlined in Section 10 apply. 1243 REQ 6: When overload is signaled by means of a specific message, the 1244 message must clearly indicate that it is being sent because of 1245 overload, as opposed to other, non overload-based failure conditions. 1246 This requirement is meant to avoid some of the problems that have 1247 arisen from the reuse of the 503 response code for multiple purposes. 1248 Of course, overload is also signaled by lack of response to requests. 1249 This requirement applies only to explicit overload signals. 1251 Meeting REQ 6: Not applicable. Overload control information is 1252 signaled as part of the Via header and not in a new header. 1254 REQ 7: The mechanism shall provide a way for an element to throttle 1255 the amount of traffic it receives from an upstream element. This 1256 throttling shall be graded so that it is not all- or-nothing as with 1257 the current 503 mechanism. This recognizes the fact that "overload" 1258 is not a binary state and that there are degrees of overload. 1260 Meeting REQ 7: Yes, please see Section 5.5 and Section 5.10. 1262 REQ 8: The mechanism shall ensure that, when a request was not 1263 processed successfully due to overload (or failure) of a downstream 1264 element, the request will not be retried on another element that is 1265 also overloaded or whose status is unknown. This requirement derives 1266 from REQ 1. 1268 Meeting REQ 8: Partially. A SIP client that has overload information 1269 from multiple downstream servers will not retry the request on 1270 another element. However, if a SIP client does not know the overload 1271 status of a downstream server, it may send the request to that 1272 server. 1274 REQ 9: That a request has been rejected from an overloaded element 1275 shall not unduly restrict the ability of that request to be submitted 1276 to and processed by an element that is not overloaded. This 1277 requirement derives from REQ 1. 1279 Meeting REQ 9: Yes, a SIP client conformant to this specification 1280 will send the request to a different element. 1282 REQ 10: The mechanism should support servers that receive requests 1283 from a large number of different upstream elements, where the set of 1284 upstream elements is not enumerable. 1286 Meeting REQ 10: Yes, there are no constraints on the number of 1287 upstream clients. 1289 REQ 11: The mechanism should support servers that receive requests 1290 from a finite set of upstream elements, where the set of upstream 1291 elements is enumerable. 1293 Meeting REQ 11: Yes, there are no constraints on the number of 1294 upstream clients. 1296 REQ 12: The mechanism should work between servers in different 1297 domains. 1299 Meeting REQ 12: Yes, there are no inherent limitations on using 1300 overload control between domains. 1302 REQ 13: The mechanism must not dictate a specific algorithm for 1303 prioritizing the processing of work within a proxy during times of 1304 overload. It must permit a proxy to prioritize requests based on any 1305 local policy, so that certain ones (such as a call for emergency 1306 services or a call with a specific value of the Resource-Priority 1307 header field [RFC4412]) are given preferential treatment, such as not 1308 being dropped, being given additional retransmission, or being 1309 processed ahead of others. 1311 Meeting REQ 13: Yes, please see Section 5.10. 1313 REQ 14: REQ 14: The mechanism should provide unambiguous directions 1314 to clients on when they should retry a request and when they should 1315 not. This especially applies to TCP connection establishment and SIP 1316 registrations, in order to mitigate against avalanche restart. 1318 Meeting REQ 14: Yes, Section 5.9 provides normative behavior on when 1319 to retry a request after repeated timeouts and fatal transport errors 1320 resulting from communications with a non-responsive downstream SIP 1321 server. 1323 REQ 15: In cases where a network element fails, is so overloaded that 1324 it cannot process messages, or cannot communicate due to a network 1325 failure or network partition, it will not be able to provide explicit 1326 indications of the nature of the failure or its levels of congestion. 1327 The mechanism must properly function in these cases. 1329 Meeting REQ 15: Yes, Section 5.9 provides normative behavior on when 1330 to retry a request after repeated timeouts and fatal transport errors 1331 resulting from communications with a non-responsive downstream SIP 1332 server. 1334 REQ 16: The mechanism should attempt to minimize the overhead of the 1335 overload control messaging. 1337 Meeting REQ 16: Yes, overload control messages are sent in the 1338 topmost Via header, which is always processed by the SIP elements. 1340 REQ 17: The overload mechanism must not provide an avenue for 1341 malicious attack, including DoS and DDoS attacks. 1343 Meeting REQ 17: Partially. Since overload control information is 1344 shared between a pair of communicating entities, a confidential and 1345 authenticated channel can be used for this communication. However, 1346 if such a channel is not available, then the security ramifications 1347 outlined in Section 10 apply. 1349 REQ 18: The overload mechanism should be unambiguous about whether a 1350 load indication applies to a specific IP address, host, or URI, so 1351 that an upstream element can determine the load of the entity to 1352 which a request is to be sent. 1354 Meeting REQ 18: Yes, please see discussion in Section 5.5. 1356 REQ 19: The specification for the overload mechanism should give 1357 guidance on which message types might be desirable to process over 1358 others during times of overload, based on SIP-specific 1359 considerations. For example, it may be more beneficial to process a 1360 SUBSCRIBE refresh with Expires of zero than a SUBSCRIBE refresh with 1361 a non-zero expiration (since the former reduces the overall amount of 1362 load on the element), or to process re-INVITEs over new INVITEs. 1364 Meeting REQ 19: Yes, please see Section 5.10. 1366 REQ 20: In a mixed environment of elements that do and do not 1367 implement the overload mechanism, no disproportionate benefit shall 1368 accrue to the users or operators of the elements that do not 1369 implement the mechanism. 1371 Meeting REQ 20: Yes, an element that does not implement overload 1372 control does not receive any measure of extra benefit. 1374 REQ 21: The overload mechanism should ensure that the system remains 1375 stable. When the offered load drops from above the overall capacity 1376 of the network to below the overall capacity, the throughput should 1377 stabilize and become equal to the offered load. 1379 Meeting REQ 21: Yes, the overload control mechanism described in this 1380 draft ensures the stability of the system. 1382 REQ 22: It must be possible to disable the reporting of load 1383 information towards upstream targets based on the identity of those 1384 targets. This allows a domain administrator who considers the load 1385 of their elements to be sensitive information, to restrict access to 1386 that information. Of course, in such cases, there is no expectation 1387 that the overload mechanism itself will help prevent overload from 1388 that upstream target. 1390 Meeting REQ 22: Yes, an operator of a SIP server can configure the 1391 SIP server to only report overload control information for requests 1392 received over a confidential channel, for example. However, note 1393 that this requirement is in conflict with REQ 3, as it introduces a 1394 modicum of extra configuration. 1396 REQ 23: It must be possible for the overload mechanism to work in 1397 cases where there is a load balancer in front of a farm of proxies. 1399 Meeting REQ 23: Yes. Depending on the type of load balancer, this 1400 requirement is met. A load balancer fronting a farm of SIP proxies 1401 could be a SIP-aware load balancer or one that is not SIP-aware. If 1402 the load balancer is SIP-aware, it can make conscious decisions on 1403 throttling outgoing traffic towards the individual server in the farm 1404 based on the overload control parameters returned by the server. On 1405 the other hand, if the load balancer is not SIP-aware, then there are 1406 other strategies to perform overload control. Section 6 of [RFC6357] 1407 documents some of these strategies in more detail (see discussion 1408 related to Figure 3(a) in Section 6). 1410 Authors' Addresses 1412 Vijay K. Gurbani (editor) 1413 Bell Laboratories, Alcatel-Lucent 1414 1960 Lucent Lane, Rm 9C-533 1415 Naperville, IL 60563 1416 USA 1418 Email: vkg@bell-labs.com 1420 Volker Hilt 1421 Bell Labs/Alcatel-Lucent 1422 791 Holmdel-Keyport Rd 1423 Holmdel, NJ 07733 1424 USA 1426 Email: volkerh@bell-labs.com 1427 Henning Schulzrinne 1428 Columbia University/Department of Computer Science 1429 450 Computer Science Building 1430 New York, NY 10027 1431 USA 1433 Phone: +1 212 939 7004 1434 Email: hgs@cs.columbia.edu 1435 URI: http://www.cs.columbia.edu