idnits 2.17.1 draft-ietf-dime-overload-reqs-10.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords -- however, there's a paragraph with a matching beginning. Boilerplate error? (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (July 29, 2013) is 3924 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- No issues found here. Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group E. McMurry 3 Internet-Draft B. Campbell 4 Intended status: Informational Tekelec 5 Expires: January 30, 2014 July 29, 2013 7 Diameter Overload Control Requirements 8 draft-ietf-dime-overload-reqs-10 10 Abstract 12 When a Diameter server or agent becomes overloaded, it needs to be 13 able to gracefully reduce its load, typically by informing clients to 14 reduce sending traffic for some period of time. Otherwise, it must 15 continue to expend resources parsing and responding to Diameter 16 messages, possibly resulting in congestion collapse. The existing 17 Diameter mechanisms, listed in Section 4 are not sufficient for this 18 purpose. This document describes the limitations of the existing 19 mechanisms in Section 5. Requirements for new overload management 20 mechanisms are provided in Section 7. 22 Status of this Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on January 30, 2014. 39 Copyright Notice 41 Copyright (c) 2013 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 1.1. Documentation Conventions . . . . . . . . . . . . . . . . 3 58 1.2. Causes of Overload . . . . . . . . . . . . . . . . . . . . 4 59 1.3. Effects of Overload . . . . . . . . . . . . . . . . . . . 5 60 1.4. Overload vs. Network Congestion . . . . . . . . . . . . . 5 61 1.5. Diameter Applications in a Broader Network . . . . . . . . 6 62 2. Overload Control Scenarios . . . . . . . . . . . . . . . . . . 6 63 2.1. Peer to Peer Scenarios . . . . . . . . . . . . . . . . . . 7 64 2.2. Agent Scenarios . . . . . . . . . . . . . . . . . . . . . 9 65 2.3. Interconnect Scenario . . . . . . . . . . . . . . . . . . 12 66 3. Diameter Overload Case Studies . . . . . . . . . . . . . . . . 13 67 3.1. Overload in Mobile Data Networks . . . . . . . . . . . . . 13 68 3.2. 3GPP Study on Core Network Overload . . . . . . . . . . . 15 69 4. Existing Mechanisms . . . . . . . . . . . . . . . . . . . . . 15 70 5. Issues with the Current Mechanisms . . . . . . . . . . . . . . 16 71 5.1. Problems with Implicit Mechanism . . . . . . . . . . . . . 17 72 5.2. Problems with Explicit Mechanisms . . . . . . . . . . . . 17 73 6. Extensibility and Application Independence . . . . . . . . . . 18 74 7. Solution Requirements . . . . . . . . . . . . . . . . . . . . 19 75 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 23 76 9. Security Considerations . . . . . . . . . . . . . . . . . . . 23 77 9.1. Access Control . . . . . . . . . . . . . . . . . . . . . . 24 78 9.2. Denial-of-Service Attacks . . . . . . . . . . . . . . . . 24 79 9.3. Replay Attacks . . . . . . . . . . . . . . . . . . . . . . 24 80 9.4. Man-in-the-Middle Attacks . . . . . . . . . . . . . . . . 25 81 9.5. Compromised Hosts . . . . . . . . . . . . . . . . . . . . 25 82 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 25 83 10.1. Normative References . . . . . . . . . . . . . . . . . . . 25 84 10.2. Informative References . . . . . . . . . . . . . . . . . . 26 85 Appendix A. Contributors . . . . . . . . . . . . . . . . . . . . 26 86 Appendix B. Acknowledgements . . . . . . . . . . . . . . . . . . 27 87 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 27 89 1. Introduction 91 A Diameter [RFC6733] node is said to be overloaded when it has 92 insufficient resources to successfully process all of the Diameter 93 requests that it receives. When a node becomes overloaded, it needs 94 to be able to gracefully reduce its load, typically by informing 95 clients to reduce sending traffic for some period of time. 96 Otherwise, it must continue to expend resources parsing and 97 responding to Diameter messages, possibly resulting in congestion 98 collapse. The existing mechanisms provided by Diameter are not 99 sufficient for this purpose. This document describes the limitations 100 of the existing mechanisms, and provides requirements for new 101 overload management mechanisms. 103 This document draws on the work done on SIP overload control 104 ([RFC5390], [RFC6357]) as well as on experience gained via overload 105 handling in Signaling System No. 7 (SS7) networks and studies done by 106 the Third Generation Partnership Project (3GPP) (Section 3). 108 Diameter is not typically an end-user protocol; rather it is 109 generally used as one component in support of some end-user activity. 111 For example, a SIP server might use Diameter to authenticate and 112 authorize user access. Overload in the Diameter backend 113 infrastructure will likely impact the experience observed by the end 114 user in the SIP application. 116 The impact of Diameter overload on the client application (a client 117 application may use the Diameter protocol and other protocols to do 118 its job) is beyond the scope of this document. 120 This document presents non-normative descriptions of causes of 121 overload along with related scenarios and studies. Finally, it 122 offers a set of normative requirements for an improved overload 123 indication mechanism. 125 1.1. Documentation Conventions 127 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 128 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 129 document are to be interpreted as defined in [RFC2119], with the 130 exception that they are not intended for interoperability of 131 implementations. Rather, they are used to describe requirements 132 towards future specifications where the interoperability requirements 133 will be defined. 135 The terms "client", "server", "agent", "node", "peer", "upstream", 136 and "downstream" are used as defined in [RFC6733]. 138 1.2. Causes of Overload 140 Overload occurs when an element, such as a Diameter server or agent, 141 has insufficient resources to successfully process all of the traffic 142 it is receiving. Resources include all of the capabilities of the 143 element used to process a request, including CPU processing, memory, 144 I/O, and disk resources. It can also include external resources such 145 as a database or DNS server, in which case the CPU, processing, 146 memory, I/O, and disk resources of those elements are effectively 147 part of the logical element processing the request. 149 External resources can include upstream Diameter nodes; for example, 150 a Diameter agent can become effectively overloaded if one or more 151 upstream nodes are overloaded. While overload is not the same thing 152 as network congestion, network congestion can reduce a Diameter nodes 153 ability to process and respond to requests, thus contributing to 154 overload. 156 A Diameter node can become overloaded due to request levels that 157 exceed its capacity, a reduction of available resources ( for 158 example, a local or upstream hardware failure) or a combination of 159 the two. 161 Overload can occur for many reasons, including: 163 Inadequate capacity: When designing Diameter networks, that is, 164 application layer multi-node Diameter deployments, it can be very 165 difficult to predict all scenarios that may cause elevated 166 traffic. It may also be more costly to implement support for some 167 scenarios than a network operator may deem worthwhile. This 168 results in the likelihood that a Diameter network will not have 169 adequate capacity to handle all situations. 171 Dependency failures: A Diameter node can become overloaded because a 172 resource on which it is dependent has failed or become overloaded, 173 greatly reducing the logical capacity of the node. In these 174 cases, even minimal traffic might cause the node to go into 175 overload. Examples of such dependency overloads include DNS 176 servers, databases, disks, and network interfaces. 178 Component failures: A Diameter node can become overloaded when it is 179 a member of a cluster of servers that each share the load of 180 traffic, and one or more of the other members in the cluster fail. 181 In this case, the remaining nodes take over the work of the failed 182 nodes. Normally, capacity planning takes such failures into 183 account, and servers are typically run with enough spare capacity 184 to handle failure of another node. However, unusual failure 185 conditions can cause many nodes to fail at once. This is often 186 the case with software failures, where a bad packet or bad 187 database entry hits the same bug in a set of nodes in a cluster. 189 Network Initiated Traffic Flood: Issues with the radio access 190 network in a mobile network such as radio overlays with frequent 191 handovers, and operational changes are examples of network events 192 that can precipitate a flood of Diameter signaling traffic, such 193 as an avalanche restart. Failure of a Diameter proxy may also 194 result in a large amount of signaling as connections and sessions 195 are reestablished. 197 Subscriber Initiated Traffic Flood: Large gatherings of subscribers 198 or events that result in many subscribers interacting with the 199 network in close time proximity can result in Diameter signaling 200 traffic floods. For example, the finale of a large fireworks show 201 could be immediately followed by many subscribers posting 202 messages, pictures, and videos concentrated on one portion of a 203 network. Subscriber devices, such as smartphones, may use 204 aggressive registration strategies that generate unusually high 205 Diameter traffic loads. 207 DoS attacks: An attacker, wishing to disrupt service in the network, 208 can cause a large amount of traffic to be launched at a target 209 element. This can be done from a central source of traffic or 210 through a distributed DoS attack. In all cases, the volume of 211 traffic well exceeds the capacity of the element, sending the 212 system into overload. 214 1.3. Effects of Overload 216 Modern Diameter networks, comprised of application layer multi-node 217 deployments of Diameter elements, may operate at very large 218 transaction volumes. If a Diameter node becomes overloaded, or even 219 worse, fails completely, a large number of messages may be lost very 220 quickly. Even with redundant servers, many messages can be lost in 221 the time it takes for failover to complete. While a Diameter client 222 or agent should be able to retry such requests, an overloaded peer 223 may cause a sudden large increase in the number of transaction 224 transactions needing to be retried, rapidly filling local queues or 225 otherwise contributing to local overload. Therefore Diameter devices 226 need to be able to shed load before critical failures can occur. 228 1.4. Overload vs. Network Congestion 230 This document uses the term "overload" to refer to application-layer 231 overload at Diameter nodes. This is distinct from "network 232 congestion", that is, congestion that occurs at the lower networking 233 layers that may impact the delivery of Diameter messages between 234 nodes. The authors recognize that element overload and network 235 congestion are interrelated, and that overload can contribute to 236 network congestion and vice versa. 238 Network congestion issues are better handled by the transport 239 protocols. Diameter uses TCP and SCTP, both of which include 240 congestion management features. Analysis of whether those features 241 are sufficient for transport level congestion between Diameter nodes, 242 and any work to further mitigate network congestion is out of scope 243 both for this document, and for the work proposed by this document. 245 1.5. Diameter Applications in a Broader Network 247 Most elements using Diameter applications do not use Diameter 248 exclusively. It is important to realize that overload of an element 249 can be caused by a number of factors that may be unrelated to the 250 processing of Diameter or Diameter applications. 252 An element that doesn't use Diameter exclusively needs to be able to 253 signal to Diameter peers that it is experiencing overload regardless 254 of the cause of the overload, since the overload will affect that 255 element's ability to process Diameter transactions. If the element 256 communicates with other protocols than Diameter, it may also need to 257 signal the overload situation on these protocols depending on its 258 function and the architecture of the network and application it is 259 providing services for. Whether that is necessary can only be 260 decided within the context of that architecture and use cases. A 261 mechanism for signaling overload with Diameter, which this 262 specification details the requirements for, provides Diameter nodes 263 the ability to signal their Diameter peers of overload, mitigating 264 that part of the issue. Diameter nodes may need to use this, as well 265 as other mechanisms, to solve their broader overload issues. 266 Indicating overload on protocols other than Diameter is out of scope 267 for this document, and for the work proposed by this document. 269 2. Overload Control Scenarios 271 Several Diameter deployment scenarios exist that may impact overload 272 management. The following scenarios help motivate the requirements 273 for an overload management mechanism. 275 These scenarios are by no means exhaustive, and are in general 276 simplified for the sake of clarity. In particular, the authors 277 assume for the sake of clarity that the client sends Diameter 278 requests to the server, and the server sends responses to client, 279 even though Diameter supports bidirectional applications. Each 280 direction in such an application can be modeled separately. 282 In a large scale deployment, many of the nodes represented in these 283 scenarios would be deployed as clusters of servers. The authors 284 assume that such a cluster is responsible for managing its own 285 internal load balancing and overload management so that it appears as 286 a single Diameter node. That is, other Diameter nodes can treat it 287 as single, monolithic node for the purposes of overload management. 289 These scenarios do not illustrate the client application. As 290 mentioned in Section 1, Diameter is not typically an end-user 291 protocol; rather it is generally used in support of some other client 292 application. These scenarios do not consider the impact of Diameter 293 overload on the client application. 295 2.1. Peer to Peer Scenarios 297 This section describes Diameter peer-to-peer scenarios. That is, 298 scenarios where a Diameter client talks directly with a Diameter 299 server, without the use of a Diameter agent. 301 Figure 1 illustrates the simplest possible Diameter relationship. 302 The client and server share a one-to-one peer-to-peer relationship. 303 If the server becomes overloaded, either because the client exceeds 304 the server's capacity, or because the server's capacity is reduced 305 due to some resource dependency, the client needs to reduce the 306 amount of Diameter traffic it sends to the server. Since the client 307 cannot forward requests to another server, it must either queue 308 requests until the server recovers, or itself become overloaded in 309 the context of the client application and other protocols it may also 310 use. 312 +------------------+ 313 | | 314 | | 315 | Server | 316 | | 317 +--------+---------+ 318 | 319 | 320 +--------+---------+ 321 | | 322 | | 323 | Client | 324 | | 325 +------------------+ 327 Figure 1: Basic Peer to Peer Scenario 329 Figure 2 shows a similar scenario, except in this case the client has 330 multiple servers that can handle work for a specific realm and 331 application. If server 1 becomes overloaded, the client can forward 332 traffic to server 2. Assuming server 2 has sufficient reserve 333 capacity to handle the forwarded traffic, the client should be able 334 to continue serving client application protocol users. If server 1 335 is approaching overload, but can still handle some number of new 336 request, it needs to be able to instruct the client to forward a 337 subset of its traffic to server 2. 339 +------------------+ +------------------+ 340 | | | | 341 | | | | 342 | Server 1 | | Server 2 | 343 | | | | 344 +--------+-`.------+ +------.'+---------+ 345 `. .' 346 `. .' 347 `. .' 348 `. .' 349 +-------`.'--------+ 350 | | 351 | | 352 | Client | 353 | | 354 +------------------+ 356 Figure 2: Multiple Server Peer to Peer Scenario 358 Figure 3 illustrates a peer-to-peer scenario with multiple Diameter 359 realm and application combinations. In this example, server 2 can 360 handle work for both applications. Each application might have 361 different resource dependencies. For example, a server might need to 362 access one database for application A, and another for application B. 363 This creates a possibility that Server 2 could become overloaded for 364 application A but not for application B, in which case the client 365 would need to divert some part of its application A requests to 366 server 1, but should not divert any application B requests. This 367 requires server 2 to be able to distinguish between applications when 368 it indicates an overload condition to the client. 370 On the other hand, it's possible that the servers host many 371 applications. If server 2 becomes overloaded for all applications, 372 it would be undesirable for it to have to notify the client 373 separately for each application. Therefore it also needs a way to 374 indicate that it is overloaded for all possible applications. 376 +---------------------------------------------+ 377 | Application A +----------------------+----------------------+ 378 |+------------------+ | +----------------+ | +------------------+| 379 || | | | | | | || 380 || | | | | | | || 381 || Server 1 | | | Server 2 | | | Server 3 || 382 || | | | | | | || 383 |+--------+---------+ | +-------+--------+ | +-+----------------+| 384 | | | | | | | 385 +---------+-----------+----------+-----------+ | | 386 | | | | | 387 | | | | Application B | 388 | +----------+----------------+-----------------+ 389 ``-.._ | | 390 `-..__ | _.-'' 391 `--._ | _.-'' 392 ``-._ | _.-'' 393 +-----`-.-''-----+ 394 | | 395 | | 396 | Client | 397 | | 398 +----------------+ 400 Figure 3: Multiple Application Peer to Peer Scenario 402 2.2. Agent Scenarios 404 This section describes scenarios that include a Diameter agent, 405 either in the form of a Diameter relay or Diameter proxy. These 406 scenarios do not consider Diameter redirect agents, since they are 407 more readily modeled as end-servers. 409 Figure 4 illustrates a simple Diameter agent scenario with a single 410 client, agent, and server. In this case, overload can occur at the 411 server, at the agent, or both. But in most cases, client behavior is 412 the same whether overload occurs at the server or at the agent. From 413 the client's perspective, server overload and agent overload is the 414 same thing. 416 +------------------+ 417 | | 418 | | 419 | Server | 420 | | 421 +--------+---------+ 422 | 423 | 424 +--------+---------+ 425 | | 426 | | 427 | Agent | 428 | | 429 +--------+---------+ 430 | 431 | 432 +--------+---------+ 433 | | 434 | | 435 | Client | 436 | | 437 +------------------+ 439 Figure 4: Basic Agent Scenario 441 Figure 5 shows an agent scenario with multiple servers. If server 1 442 becomes overloaded, but server 2 has sufficient reserve capacity, the 443 agent may be able to transparently divert some or all Diameter 444 requests originally bound for server 1 to server 2. 446 In most cases, the client does not have detailed knowledge of the 447 Diameter topology upstream of the agent. If the agent uses dynamic 448 discovery to find eligible servers, the set of eligible servers may 449 not be enumerable from the perspective of the client. Therefore, in 450 most cases the agent needs to deal with any upstream overload issues 451 in a way that is transparent to the client. If one server notifies 452 the agent that it has become overloaded, the notification should not 453 be passed back to the client in a way that the client could 454 mistakenly perceive the agent itself as being overloaded. If the set 455 of all possible destinations upstream of the agent no longer has 456 sufficient capacity for incoming load, the agent itself becomes 457 effectively overloaded. 459 On the other hand, there are cases where the client needs to be able 460 to select a particular server from behind an agent. For example, if 461 a Diameter request is part of a multiple-round-trip authentication, 462 or is otherwise part of a Diameter "session", it may have a 463 DestinationHost AVP that requires the request to be served by server 464 1. Therefore the agent may need to inform a client that a particular 465 upstream server is overloaded or otherwise unavailable. Note that 466 there can be many ways a server can be specified, which may have 467 different implications (e.g. by IP address, by host name, etc). 469 +------------------+ +------------------+ 470 | | | | 471 | | | | 472 | Server 1 | | Server 2 | 473 | | | | 474 +--------+-`.------+ +------.'+---------+ 475 `. .' 476 `. .' 477 `. .' 478 `. .' 479 +-------`.'--------+ 480 | | 481 | | 482 | Agent | 483 | | 484 +--------+---------+ 485 | 486 | 487 | 488 +--------+---------+ 489 | | 490 | | 491 | Client | 492 | | 493 +------------------+ 495 Figure 5: Multiple Server Agent Scenario 497 Figure 6 shows a scenario where an agent routes requests to a set of 498 servers for more than one Diameter realm and application. In this 499 scenario, if server 1 becomes overloaded or unavailable while server 500 2 still has available capacity, the agent may effectively operate at 501 reduced capacity for application A, but at full capacity for 502 application B. Therefore, the agent needs to be able to report that 503 it is overloaded for one application, but not for another. 505 +--------------------------------------------+ 506 | Application A +----------------------+----------------------+ 507 |+------------------+ | +----------------+ | +------------------+| 508 || | | | | | | || 509 || | | | | | | || 510 || Server 1 | | | Server 2 | | | Server 3 || 511 || | | | | | | || 512 |+---------+--------+ | +-------+--------+ | +--+---------------+| 513 | | | | | | | 514 +----------+----------+----------+-----------+ | | 515 | | | | | 516 | | | | Application B | 517 | +----------+-----------------+----------------+ 518 | | | 519 ``--.__ | _. 520 ``-.__ | __.--'' 521 `--.._ | _..--' 522 +----``-+.''-----+ 523 | | 524 | | 525 | Agent | 526 | | 527 +-------+--------+ 528 | 529 | 530 +-------+--------+ 531 | | 532 | | 533 | Client | 534 | | 535 +----------------+ 537 Figure 6: Multiple Application Agent Scenario 539 2.3. Interconnect Scenario 541 Another scenario to consider when looking at Diameter overload is 542 that of multiple network operators using Diameter components 543 connected through an interconnect service, e.g. using IPX. IPX (IP 544 eXchange) [IR.34] is an Inter-Operator IP Backbone that provides 545 roaming interconnection network between mobile operators and service 546 providers. The IPX is also used to transport Diameter signaling 547 between operators [IR.88]. Figure 7 shows two network operators with 548 an interconnect network in-between. There could be any number of 549 these networks between any two network operator's networks. 551 +-------------------------------------------+ 552 | Interconnect | 553 | | 554 | +--------------+ +--------------+ | 555 | | Server 3 |------| Server 4 | | 556 | +--------------+ +--------------+ | 557 | .' `. | 558 +------.-'--------------------------`.------+ 559 .' `. 560 .-' `. 561 ------------.'-----+ +----`.------------- 562 +----------+ | | +----------+ 563 | Server 1 | | | | Server 2 | 564 +----------+ | | +----------+ 565 | | 566 Network Operator 1 | | Network Operator 2 567 -------------------+ +------------------- 569 Figure 7: Two Network Interconnect Scenario 571 The characteristics of the information that an operator would want to 572 share over such a connection are different from the information 573 shared between components within a network operator's network. 574 Network operators may not want to convey topology or operational 575 information, which limits how much overload and loading information 576 can be sent. For the interconnect scenario shown, Server 2 may want 577 to signal overload to Server 1, to affect traffic coming from Network 578 Operator 1. 580 This case is distinct from those internal to a network operator's 581 network, where there may be many more elements in a more complicated 582 topology. Also, the elements in the interconnect network may not 583 support Diameter overload control, and the network operators may not 584 want the interconnect network to use overload or loading information. 585 They may only want the information to pass through the interconnect 586 network without further processing or action by the interconnect 587 network even if the elements in the interconnect network do support 588 Diameter overload control. 590 3. Diameter Overload Case Studies 592 3.1. Overload in Mobile Data Networks 594 As the number of Third Generation (3G) and Long Term Evolution (LTE) 595 enabled smartphone devices continue to expand in mobile networks, 596 there have been situations where high signaling traffic load led to 597 overload events at the Diameter-based Home Location Registries (HLR) 598 and/or Home Subscriber Servers (HSS) [TR23.843]. The root causes of 599 the HLR congestion events were manifold but included hardware failure 600 and procedural errors. The result was high signaling traffic load on 601 the HLR and HSS. 603 The 3GPP architecture [TS23.002] makes extensive use of Diameter. It 604 is used for mobility management [TS29.272] (and others), (IP 605 Multimedia Subsystem) IMS [TS29.228] (and others), policy and 606 charging control [TS29.212] (and others) as well as other functions. 607 The details of the architecture are out of scope for this document, 608 but it is worth noting that there are quite a few Diameter 609 applications, some with quite large amounts of Diameter signaling in 610 deployed networks. 612 The 3GPP specifications do not currently address overload for 613 Diameter applications or provide an equivalent load control mechanism 614 to those provided in the more traditional SS7 elements in (Global 615 System for Mobile Communications) GSM [TS29.002]. The capabilities 616 specified in the 3GPP standards do not adequately address the 617 abnormal condition where excessively high signaling traffic load 618 situations are experienced. 620 Smartphones, an increasingly large percentage of mobile devices, 621 contribute much more heavily, relative to non-smartphones, to the 622 continuation of a registration surge due to their very aggressive 623 registration algorithms. Smartphone behavior contributes to network 624 loading and can contribute to overload conditions. The aggressive 625 smartphone logic is designed to: 627 a. always have voice and data registration, and 629 b. constantly try to be on 3G or LTE data (and thus on 3G voice or 630 VoLTE [IR.92]) for their added benefits. 632 Non-smartphones typically have logic to wait for a time period after 633 registering successfully on voice and data. 635 The smartphone aggressive registration is problematic in two ways: 637 o first by generating excessive signaling load towards the HSS that 638 is ten times that from a non-smartphone, 640 o and second by causing continual registration attempts when a 641 network failure affects registrations through the 3G data network. 643 3.2. 3GPP Study on Core Network Overload 645 A study in 3GPP SA2 on core network overload has produced the 646 technical report [TR23.843]. This enumerates several causes of 647 overload in mobile core networks including portions that are signaled 648 using Diameter. This document is a work in progress and is not 649 complete. However, it is useful for pointing out scenarios and the 650 general need for an overload control mechanism for Diameter. 652 It is common for mobile networks to employ more than one radio 653 technology and to do so in an overlay fashion with multiple 654 technologies present in the same location (such as 2nd or 3rd 655 generation mobile technologies along with LTE). This presents 656 opportunities for traffic storms when issues occur on one overlay and 657 not another as all devices that had been on the overlay with issues 658 switch. This causes a large amount of Diameter traffic as locations 659 and policies are updated. 661 Another scenario called out by this study is a flood of registration 662 and mobility management events caused by some element in the core 663 network failing. This flood of traffic from end nodes falls under 664 the network initiated traffic flood category. There is likely to 665 also be traffic resulting directly from the component failure in this 666 case. A similar flood can occur when elements or components recover 667 as well. 669 Subscriber initiated traffic floods are also indicated in this study 670 as an overload mechanism where a large number of mobile devices 671 attempting to access services at the same time, such as in response 672 to an entertainment event or a catastrophic event. 674 While this 3GPP study is concerned with the broader effects of these 675 scenarios on wireless networks and their elements, they have 676 implications specifically for Diameter signaling. One of the goals 677 of this document is to provide guidance for a core mechanism that can 678 be used to mitigate the scenarios called out by this study. 680 4. Existing Mechanisms 682 Diameter offers both implicit and explicit mechanisms for a Diameter 683 node to learn that a peer is overloaded or unreachable. The implicit 684 mechanism is simply the lack of responses to requests. If a client 685 fails to receive a response in a certain time period, it assumes the 686 upstream peer is unavailable, or overloaded to the point of effective 687 unavailability. The watchdog mechanism [RFC3539] ensures that a 688 certain rate of transaction responses occur even when there is 689 otherwise little or no other Diameter traffic. 691 The explicit mechanism can involve specific protocol error responses, 692 where an agent or server tells a downstream peer that it is either 693 too busy to handle a request (DIAMETER_TOO_BUSY) or unable to route a 694 request to an upstream destination (DIAMETER_UNABLE_TO_DELIVER), 695 perhaps because that destination itself is overloaded to the point of 696 unavailability. 698 Another explicit mechanism, a DPR (Disconnect-Peer-Request) message, 699 can be sent with a Disconnect-Cause of BUSY. This signals the 700 sender's intent to close the transport connection, and requests the 701 client not to reconnect. 703 Once a Diameter node learns that an upstream peer has become 704 overloaded via one of these mechanisms, it can then attempt to take 705 action to reduce the load. This usually means forwarding traffic to 706 an alternate destination, if available. If no alternate destination 707 is available, the node must either reduce the number of messages it 708 originates (in the case of a client) or inform the client to reduce 709 traffic (in the case of an agent.) 711 Diameter requires the use of a congestion-managed transport layer, 712 currently TCP or SCTP, to mitigate network congestion. It is 713 expected that these transports manage network congestion and that 714 issues with transport (e.g. congestion propagation and window 715 management) are managed at that level. But even with a congestion- 716 managed transport, a Diameter node can become overloaded at the 717 Diameter protocol or application layers due to the causes described 718 in Section 1.2 and congestion managed transports do not provide 719 facilities (and are at the wrong level) to handle server overload. 720 Transport level congestion management is also not sufficient to 721 address overload in cases of multi-hop and multi-destination 722 signaling. 724 5. Issues with the Current Mechanisms 726 The currently available Diameter mechanisms for indicating an 727 overload condition are not adequate to avoid service outages due to 728 overload. This inadequacy may, in turn, contribute to broader 729 congestion collapse due to unresponsive Diameter nodes causing 730 application or transport layer retransmissions. In particular, they 731 do not allow a Diameter agent or server to shed load as it approaches 732 overload. At best, a node can only indicate that it needs to 733 entirely stop receiving requests, i.e. that it has effectively 734 failed. Even that is problematic due to the inability to indicate 735 durational validity on the transient errors available in the base 736 Diameter protocol. Diameter offers no mechanism to allow a node to 737 indicate different overload states for different categories of 738 messages, for example, if it is overloaded for one Diameter 739 application but not another. 741 5.1. Problems with Implicit Mechanism 743 The implicit mechanism doesn't allow an agent or server to inform the 744 client of a problem until it is effectively too late to do anything 745 about it. The client does not know to take action until the upstream 746 node has effectively failed. A Diameter node has no opportunity to 747 shed load early to avoid collapse in the first place. 749 Additionally, the implicit mechanism cannot distinguish between 750 overload of a Diameter node and network congestion. Diameter treats 751 the failure to receive an answer as a transport failure. 753 5.2. Problems with Explicit Mechanisms 755 The Diameter specification is ambiguous on how a client should handle 756 receipt of a DIAMETER_TOO_BUSY response. The base specification 757 [RFC6733] indicates that the sending client should attempt to send 758 the request to a different peer. It makes no suggestion that the 759 receipt of a DIAMETER_TOO_BUSY response should affect future Diameter 760 messages in any way. 762 The Authentication, Authorization, and Accounting (AAA) Transport 763 Profile [RFC3539] recommends that a AAA node that receives a "Busy" 764 response failover all remaining requests to a different agent or 765 server. But while the Diameter base specification explicitly depends 766 on RFC3539 to define transport behavior, it does not refer to RFC3539 767 in the description of behavior on receipt of DIAMETER_TOO_BUSY. 768 There's a strong likelihood that at least some implementations will 769 continue to send Diameter requests to an upstream peer even after 770 receiving a DIAMETER_TOO_BUSY error. 772 BCP 41 [RFC2914] describes, among other things, how end-to-end 773 application behavior can help avoid congestion collapse. In 774 particular, an application should avoid sending messages that will 775 never be delivered or processed. The DIAMETER_TOO_BUSY behavior as 776 described in the Diameter base specification fails at this, since if 777 an upstream node becomes overloaded, a client attempts each request, 778 and does not discover the need to failover the request until the 779 initial attempt fails. 781 The situation is improved if implementations follow the [RFC3539] 782 recommendation and keep state about upstream peer overload. But even 783 then, the Diameter specification offers no guidance on how long a 784 client should wait before retrying the overloaded destination. If an 785 agent or server supports multiple realms and/or applications, 786 DIAMETER_TOO_BUSY offers no way to indicate that it is overloaded for 787 one application but not another. A DIAMETER_TOO_BUSY error can only 788 indicate overload at a "whole server" scope. 790 Agent processing of a DIAMETER_TOO_BUSY response is also problematic 791 as described in the base specification. DIAMETER_TOO_BUSY is defined 792 as a protocol error. If an agent receives a protocol error, it may 793 either handle it locally or it may forward the response back towards 794 the downstream peer. If a downstream peer receives the 795 DIAMETER_TOO_BUSY response, it may stop sending all requests to the 796 agent for some period of time, even though the agent may still be 797 able to deliver requests to other upstream peers. 799 DIAMETER_UNABLE_TO_DELIVER, or using DPR with cause code BUSY also 800 have no mechanisms for specifying the scope or cause of the failure, 801 or the durational validity. 803 The issues with error responses in [RFC6733] extend beyond the 804 particular issues for overload control and have been addressed in an 805 ad hoc fashion by various implementations. Addressing these in a 806 standard way would be a useful exercise, but it us beyond the scope 807 of this document. 809 6. Extensibility and Application Independence 811 Given the variety of scenarios Diameter elements can be deployed in, 812 and the variety of roles they can fulfill with Diameter and other 813 technologies, a single algorithm for handling overload may not be 814 sufficient. For purposes of this discussion, algorithm is inclusive 815 of behavior for control of overload, but does not encompass the 816 general mechanism or transport of control information. This effort 817 cannot anticipate all possible future scenarios and roles. 818 Extensibility, particularly of algorithms used to deal with overload, 819 will be important to cover these cases. 821 Similarly, the scopes that overload information may apply to may 822 include cases that have not yet been considered. Extensibility in 823 this area will also be important. 825 The basic mechanism is intended to be application-independent, that 826 is, a Diameter node can use it across any existing and future 827 Diameter applications and expect reasonable results. Certain 828 Diameter applications might, however, benefit from application- 829 specific behavior over and above the mechanism's defaults. For 830 example, an application specification might specify relative 831 priorities of messages or selection of a specific overload control 832 algorithm. 834 7. Solution Requirements 836 This section proposes requirements for an improved mechanism to 837 control Diameter overload, with the goals of improving the issues 838 described in Section 5 and supporting the scenarios described in 839 Section 2 841 REQ 1: The solution MUST provide a communication method for 842 Diameter nodes to exchange load and overload information. 844 REQ 2: The solution MUST allow Diameter nodes to support overload 845 control regardless of which Diameter applications they 846 support. Diameter clients and agents must be able to use 847 the received load and overload information to support 848 graceful behavior during an overload condition. Graceful 849 behavior under overload conditions is best described by REQ 850 3. 852 REQ 3: The solution MUST limit the impact of overload on the 853 overall useful throughput of a Diameter server, even when 854 the incoming load on the network is far in excess of its 855 capacity. The overall useful throughput under load is the 856 ultimate measure of the value of a solution. 858 REQ 4: Diameter allows requests to be sent from either side of a 859 connection and either side of a connection may have need to 860 provide its overload status. The solution MUST allow each 861 side of a connection to independently inform the other of 862 its overload status. 864 REQ 5: Diameter allows nodes to determine their peers via dynamic 865 discovery or manual configuration. The solution MUST work 866 consistently without regard to how peers are determined. 868 REQ 6: The solution designers SHOULD seek to minimize the amount of 869 new configuration required in order to work. For example, 870 it is better to allow peers to advertise or negotiate 871 support for the solution, rather than to require this 872 knowledge to be configured at each node. 874 REQ 7: The solution and any associated default algorithm(s) MUST 875 ensure that the system remains stable. At some point after 876 an overload condition has ended, the solution MUST enable 877 capacity to stabilize and become equal to what it would be 878 in the absence of an overload condition. Note that this 879 also requires that the solution MUST allow nodes to shed 880 load without introducing non converging oscillations during 881 or after an overload condition. 883 REQ 8: Supporting nodes MUST be able to distinguish current 884 overload information from stale information. 886 REQ 9: The solution MUST function across fully loaded as well as 887 quiescent transport connections. This is partially derived 888 from the requirement for stability in REQ 7. 890 REQ 10: Consumers of overload information MUST be able to determine 891 when the overload condition improves or ends. 893 REQ 11: The solution MUST be able to operate in networks of 894 different sizes. 896 REQ 12: When a single network node fails, goes into overload, or 897 suffers from reduced processing capacity, the solution MUST 898 make it possible to limit the impact of this on other nodes 899 in the network. This helps to prevent a small-scale failure 900 from becoming a widespread outage. 902 REQ 13: The solution MUST NOT introduce substantial additional work 903 for node in an overloaded state. For example, a requirement 904 for an overloaded node to send overload information every 905 time it received a new request would introduce substantial 906 work. Existing messaging is likely to have the 907 characteristic of increasing as an overload condition 908 approaches, allowing for the possibility of increased 909 feedback for information piggybacked on it. 911 REQ 14: Some scenarios that result in overload involve a rapid 912 increase of traffic with little time between normal levels 913 and overload inducing levels. The solution SHOULD provide 914 for rapid feedback when traffic levels increase. 916 REQ 15: The solution MUST NOT interfere with the congestion control 917 mechanisms of underlying transport protocols. For example, 918 a solution that opened additional TCP connections when the 919 network is congested would reduce the effectiveness of the 920 underlying congestion control mechanisms. 922 REQ 16: The solution is likely to be deployed incrementally. The 923 solution MUST support a mixed environment where some, but 924 not all, nodes implement it. 926 REQ 17: In a mixed environment with nodes that support the solution 927 and that do not, the solution MUST NOT result in materially 928 less useful throughput as would have resulted if the 929 solution were not present. It SHOULD result in less severe 930 congestion in this environment. 932 REQ 18: In a mixed environment of nodes that support the solution 933 and that do not, the solution MUST NOT preclude elements 934 that support overload control from treating elements that do 935 not support overload control in a equitable fashion relative 936 to those that do. Users and operators of nodes that do not 937 support the solution MUST NOT unfairly benefit from the 938 solution. The solution specification SHOULD provide 939 guidance to implementors for dealing with elements not 940 supporting overload control. 942 REQ 19: It MUST be possible to use the solution between nodes in 943 different realms and in different administrative domains. 945 REQ 20: Any explicit overload indication MUST be clearly 946 distinguishable from other errors reported via Diameter. 948 REQ 21: In cases where a network node fails, is so overloaded that 949 it cannot process messages, or cannot communicate due to a 950 network failure, it may not be able to provide explicit 951 indications of the nature of the failure or its levels of 952 congestion. The solution MUST result in at least as much 953 useful throughput as would have resulted if the solution was 954 not in place. 956 REQ 22: The solution MUST provide a way for a node to throttle the 957 amount of traffic it receives from a peer node. This 958 throttling SHOULD be graded so that it can be applied 959 gradually as offered load increases. Overload is not a 960 binary state; there may be degrees of overload. 962 REQ 23: The solution MUST provide sufficient information to enable a 963 load balancing node to divert messages that are rejected or 964 otherwise throttled by an overloaded upstream node to other 965 upstream nodes that are the most likely to have sufficient 966 capacity to process them. 968 REQ 24: The solution MUST provide a mechanism for indicating load 969 levels even when not in an overloaded condition, to assist 970 nodes making decisions to prevent overload conditions from 971 occurring. 973 REQ 25: The base specification for the solution SHOULD offer general 974 guidance on which message types might be desirable to send 975 or process over others during times of overload, based on 976 application-specific considerations. For example, it may be 977 more beneficial to process messages for existing sessions 978 ahead of new sessions. Some networks may have a requirement 979 to give priority to requests associated with emergency 980 sessions. Any normative or otherwise detailed definition of 981 the relative priorities of message types during an overload 982 condition will be the responsibility of the application 983 specification. 985 REQ 26: The solution MUST NOT prevent a node from prioritizing 986 requests based on any local policy, so that certain requests 987 are given preferential treatment, given additional 988 retransmission, not throttled, or processed ahead of others. 990 REQ 27: The solution MUST NOT provide new vulnerabilities to 991 malicious attack, or increase the severity of any existing 992 vulnerabilities. This includes vulnerabilities to DoS and 993 DDoS attacks as well as replay and man-in-the middle 994 attacks. Note that the Diameter base specification 995 [RFC6733] lacks end to end security and this must be 996 considered. 998 REQ 28: The solution MUST NOT depend on being deployed in 999 environments where all Diameter nodes are completely 1000 trusted. It SHOULD operate as effectively as possible in 1001 environments where other nodes are malicious; this includes 1002 preventing malicious nodes from obtaining more than a fair 1003 share of service. Note that this does not imply any 1004 responsibility on the solution to detect, or take 1005 countermeasures against, malicious nodes. 1007 REQ 29: It MUST be possible for a supporting node to make 1008 authorization decisions about what information will be sent 1009 to peer nodes based on the identity of those nodes. This 1010 allows a domain administrator who considers the load of 1011 their nodes to be sensitive information to restrict access 1012 to that information. Of course, in such cases, there is no 1013 expectation that the solution itself will help prevent 1014 overload from that peer node. 1016 REQ 30: The solution MUST NOT interfere with any Diameter compliant 1017 method that a node may use to protect itself from overload 1018 from non-supporting nodes, or from denial of service 1019 attacks. 1021 REQ 31: There are multiple situations where a Diameter node may be 1022 overloaded for some purposes but not others. For example, 1023 this can happen to an agent or server that supports multiple 1024 applications, or when a server depends on multiple external 1025 resources, some of which may become overloaded while others 1026 are fully available. The solution MUST allow Diameter nodes 1027 to indicate overload with sufficient granularity to allow 1028 clients to take action based on the overloaded resources 1029 without unreasonably forcing available capacity to go 1030 unused. The solution MUST support specification of overload 1031 information with granularities of at least "Diameter node", 1032 "realm", and "Diameter application", and MUST allow 1033 extensibility for others to be added in the future. 1035 REQ 32: The solution MUST provide a method for extending the 1036 information communicated and the algorithms used for 1037 overload control. 1039 REQ 33: The solution MUST provide a default algorithm that is 1040 mandatory to implement. 1042 REQ 34: The solution SHOULD provide a method for exchanging overload 1043 and load information between elements that are connected by 1044 intermediaries that do not support the solution. 1046 8. IANA Considerations 1048 This document makes no requests of IANA. 1050 9. Security Considerations 1052 A Diameter overload control mechanism is primarily concerned with the 1053 load and overload related behavior of nodes in a Diameter network, 1054 and the information used to affect that behavior. Load and overload 1055 information is shared between nodes and directly affects the behavior 1056 and thus is potentially vulnerable to a number of methods of attack. 1058 Load and overload information may also be sensitive from both 1059 business and network protection viewpoints. Operators of Diameter 1060 equipment want to control visibility to load and overload information 1061 to keep it from being used for competitive intelligence or for 1062 targeting attacks. It is also important that the Diameter overload 1063 control mechanism not introduce any way in which any other 1064 information carried by Diameter is sent inappropriately. 1066 Note that the Diameter base specification [RFC6733] lacks end to end 1067 security, making verifying the authenticity and ownership of load and 1068 overload information difficult for non-adjacent nodes. 1069 Authentication of load and overload information helps to alleviate 1070 several of the security issues listed in this section. 1072 This document includes requirements intended to mitigate the effects 1073 of attacks and to protect the information used by the mechanism. 1074 This section discusses potential security considerations for overload 1075 control solutions. This discussion provides the motivation for 1076 several normative requirements described in Section 7. The 1077 discussion includes specific references to the normative requirements 1078 that apply for each issue. 1080 9.1. Access Control 1082 To control the visibility of load and overload information, sending 1083 should be subject to some form of authentication and authorization of 1084 the receiver. It is also important to the receivers that they are 1085 confident the load and overload information they receive is from a 1086 legitimate source. REQ 28 requires the solution to work without 1087 assuming that all Diameter nodes in a network are trusted for the 1088 purposes of exchanging overload and load information. REQ 29 1089 requires the solution to let nodes restrict unauthorized parties from 1090 seeing overload information. Note that this implies a certain amount 1091 of configurability on the nodes supporting the Diameter overload 1092 control mechanism. 1094 9.2. Denial-of-Service Attacks 1096 An overload control mechanism provides a very attractive target for 1097 denial-of-service attacks. A small number of messages may affect a 1098 large service disruption by falsely reporting overload conditions. 1099 Alternately, attacking servers nearing, or in, overload may also be 1100 facilitated by disrupting their overload indications, potentially 1101 preventing them from mitigating their overload condition. 1103 A design goal for the Diameter overload control mechanism is to 1104 minimize or eliminate the possibility of using the mechanism for this 1105 type of attack. More strongly, REQ 27 forbids the solution from 1106 introducing new vulnerabilities to malicious attack. Additionally, 1107 REQ 30 stipulates that the solution not interfere with other 1108 mechanisms used for protection against denial of service attacks. 1110 As the intent of some denial-of-service attacks is to induce overload 1111 conditions, an effective overload control mechanism should help to 1112 mitigate the effects of an such an attack. 1114 9.3. Replay Attacks 1116 An attacker that has managed to obtain some messages from the 1117 overload control mechanism may attempt to affect the behavior of 1118 nodes supporting the mechanism by sending those messages at 1119 potentially inopportune times. In addition to time shifting, replay 1120 attacks may send messages to other nodes as well (target shifting). 1122 A design goal for the Diameter overload control solution is to 1123 minimize or eliminate the possibility of causing disruption by using 1124 a replay attack on the Diameter overload control mechanism. 1125 (Allowing a replay attack using the overload control solution would 1126 violate REQ 27.) 1128 9.4. Man-in-the-Middle Attacks 1130 By inserting themselves in between two nodes supporting the Diameter 1131 overload control mechanism, an attacker may potentially both access 1132 and alter the information sent between those nodes. This can be used 1133 for information gathering for business intelligence and attack 1134 targeting, as well as direct attacks. 1136 REQs 27, 28, and 29 imply a need to prevent man-in-the-middle attacks 1137 on the overload control solution. A transport using TLS and/or IPSEC 1138 may be desirable for this purpose. 1140 9.5. Compromised Hosts 1142 A compromised host that supports the Diameter overload control 1143 mechanism could be used for information gathering as well as for 1144 sending malicious information to any Diameter node that would 1145 normally accept information from it. While is is beyond the scope of 1146 the Diameter overload control mechanism to mitigate any operational 1147 interruption to the compromised host, REQs 28 and 29 imply a need to 1148 minimize the impact that a compromised host can have on other nodes 1149 through the use of the Diameter overload control mechanism. Of 1150 course, a compromised host could be used to cause damage in a number 1151 of other ways. This is out of scope for a Diameter overload control 1152 mechanism. 1154 10. References 1156 10.1. Normative References 1158 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1159 Requirement Levels", BCP 14, RFC 2119, March 1997. 1161 [RFC6733] Fajardo, V., Arkko, J., Loughney, J., and G. Zorn, 1162 "Diameter Base Protocol", RFC 6733, October 2012. 1164 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 1165 RFC 2914, September 2000. 1167 [RFC3539] Aboba, B. and J. Wood, "Authentication, Authorization and 1168 Accounting (AAA) Transport Profile", RFC 3539, June 2003. 1170 10.2. Informative References 1172 [RFC5390] Rosenberg, J., "Requirements for Management of Overload in 1173 the Session Initiation Protocol", RFC 5390, December 2008. 1175 [RFC6357] Hilt, V., Noel, E., Shen, C., and A. Abdelal, "Design 1176 Considerations for Session Initiation Protocol (SIP) 1177 Overload Control", RFC 6357, August 2011. 1179 [TR23.843] 1180 3GPP, "Study on Core Network Overload Solutions", 1181 TR 23.843 0.6.0, October 2012. 1183 [IR.34] GSMA, "Inter-Service Provider IP Backbone Guidelines", 1184 IR 34 7.0, January 2012. 1186 [IR.88] GSMA, "LTE Roaming Guidelines", IR 88 7.0, January 2012. 1188 [IR.92] GSMA, "IMS Profile for Voice and SMS", IR 92 7.0, 1189 March 2013. 1191 [TS23.002] 1192 3GPP, "Network Architecture", TS 23.002 12.0.0, 1193 September 2012. 1195 [TS29.272] 1196 3GPP, "Evolved Packet System (EPS); Mobility Management 1197 Entity (MME) and Serving GPRS Support Node (SGSN) related 1198 interfaces based on Diameter protocol", TS 29.272 11.4.0, 1199 September 2012. 1201 [TS29.212] 1202 3GPP, "Policy and Charging Control (PCC) over Gx/Sd 1203 reference point", TS 29.212 11.6.0, September 2012. 1205 [TS29.228] 1206 3GPP, "IP Multimedia (IM) Subsystem Cx and Dx interfaces; 1207 Signalling flows and message contents", TS 29.228 11.5.0, 1208 September 2012. 1210 [TS29.002] 1211 3GPP, "Mobile Application Part (MAP) specification", 1212 TS 29.002 11.4.0, September 2012. 1214 Appendix A. Contributors 1216 Significant contributions to this document were made by Adam Roach 1217 and Eric Noel. 1219 Appendix B. Acknowledgements 1221 Review of, and contributions to, this specification by Martin Dolly, 1222 Carolyn Johnson, Jianrong Wang, Imtiaz Shaikh, Jouni Korhonen, Robert 1223 Sparks, Dieter Jacobsohn, Janet Gunn, Jean-Jacques Trottin, Laurent 1224 Thiebaut, Andrew Booth, and Lionel Morand were most appreciated. We 1225 would like to thank them for their time and expertise. 1227 Authors' Addresses 1229 Eric McMurry 1230 Tekelec 1231 17210 Campbell Rd. 1232 Suite 250 1233 Dallas, TX 75252 1234 US 1236 Email: emcmurry@computer.org 1238 Ben Campbell 1239 Tekelec 1240 17210 Campbell Rd. 1241 Suite 250 1242 Dallas, TX 75252 1243 US 1245 Email: ben@nostrum.com