idnits 2.17.1 draft-ietf-dime-ovli-08.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 4, 2015) is 3368 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) == Outdated reference: A later version (-05) exists of draft-ietf-dime-e2e-sec-req-01 -- Obsolete informational reference (is this intentional?): RFC 4006 (Obsoleted by RFC 8506) Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Diameter Maintenance and Extensions (DIME) J. Korhonen, Ed. 3 Internet-Draft Broadcom 4 Intended status: Standards Track S. Donovan, Ed. 5 Expires: August 8, 2015 B. Campbell 6 Oracle 7 L. Morand 8 Orange Labs 9 February 4, 2015 11 Diameter Overload Indication Conveyance 12 draft-ietf-dime-ovli-08.txt 14 Abstract 16 This specification defines a base solution for Diameter overload 17 control, referred to as Diameter Overload Indication Conveyance 18 (DOIC). 20 Status of This Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on August 8, 2015. 37 Copyright Notice 39 Copyright (c) 2015 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 55 2. Terminology and Abbreviations . . . . . . . . . . . . . . . . 4 56 3. Conventions Used in This Document . . . . . . . . . . . . . . 5 57 4. Solution Overview . . . . . . . . . . . . . . . . . . . . . . 5 58 4.1. Piggybacking . . . . . . . . . . . . . . . . . . . . . . 7 59 4.2. DOIC Capability Announcement . . . . . . . . . . . . . . 7 60 4.3. DOIC Overload Condition Reporting . . . . . . . . . . . . 9 61 4.4. DOIC Extensibility . . . . . . . . . . . . . . . . . . . 11 62 4.5. Simplified Example Architecture . . . . . . . . . . . . . 11 63 5. Solution Procedures . . . . . . . . . . . . . . . . . . . . . 12 64 5.1. Capability Announcement . . . . . . . . . . . . . . . . . 12 65 5.1.1. Reacting Node Behavior . . . . . . . . . . . . . . . 13 66 5.1.2. Reporting Node Behavior . . . . . . . . . . . . . . . 13 67 5.1.3. Agent Behavior . . . . . . . . . . . . . . . . . . . 14 68 5.2. Overload Report Processing . . . . . . . . . . . . . . . 15 69 5.2.1. Overload Control State . . . . . . . . . . . . . . . 15 70 5.2.2. Reacting Node Behavior . . . . . . . . . . . . . . . 19 71 5.2.3. Reporting Node Behavior . . . . . . . . . . . . . . . 20 72 5.3. Protocol Extensibility . . . . . . . . . . . . . . . . . 22 73 6. Loss Algorithm . . . . . . . . . . . . . . . . . . . . . . . 22 74 6.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 23 75 6.2. Reporting Node Behavior . . . . . . . . . . . . . . . . . 23 76 6.3. Reacting Node Behavior . . . . . . . . . . . . . . . . . 24 77 7. Attribute Value Pairs . . . . . . . . . . . . . . . . . . . . 24 78 7.1. OC-Supported-Features AVP . . . . . . . . . . . . . . . . 25 79 7.2. OC-Feature-Vector AVP . . . . . . . . . . . . . . . . . . 25 80 7.3. OC-OLR AVP . . . . . . . . . . . . . . . . . . . . . . . 25 81 7.4. OC-Sequence-Number AVP . . . . . . . . . . . . . . . . . 26 82 7.5. OC-Validity-Duration AVP . . . . . . . . . . . . . . . . 26 83 7.6. OC-Report-Type AVP . . . . . . . . . . . . . . . . . . . 26 84 7.7. OC-Reduction-Percentage AVP . . . . . . . . . . . . . . . 27 85 7.8. Attribute Value Pair flag rules . . . . . . . . . . . . . 27 86 8. Error Response Codes . . . . . . . . . . . . . . . . . . . . 28 87 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 88 9.1. AVP codes . . . . . . . . . . . . . . . . . . . . . . . . 28 89 9.2. New registries . . . . . . . . . . . . . . . . . . . . . 29 90 10. Security Considerations . . . . . . . . . . . . . . . . . . . 29 91 10.1. Potential Threat Modes . . . . . . . . . . . . . . . . . 30 92 10.2. Denial of Service Attacks . . . . . . . . . . . . . . . 31 93 10.3. Non-Compliant Nodes . . . . . . . . . . . . . . . . . . 31 94 10.4. End-to End-Security Issues . . . . . . . . . . . . . . . 32 95 11. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 33 96 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 33 97 12.1. Normative References . . . . . . . . . . . . . . . . . . 33 98 12.2. Informative References . . . . . . . . . . . . . . . . . 33 99 Appendix A. Issues left for future specifications . . . . . . . 34 100 A.1. Additional traffic abatement algorithms . . . . . . . . . 34 101 A.2. Agent Overload . . . . . . . . . . . . . . . . . . . . . 34 102 A.3. New Error Diagnostic AVP . . . . . . . . . . . . . . . . 34 103 Appendix B. Deployment Considerations . . . . . . . . . . . . . 34 104 Appendix C. Requirements Conformance Analysis . . . . . . . . . 35 105 C.1. Deferred Requirements . . . . . . . . . . . . . . . . . . 35 106 C.2. Detection of non-supporting Intermediaries . . . . . . . 35 107 C.3. Implicit Application Indication . . . . . . . . . . . . . 36 108 C.4. Stateless Operation . . . . . . . . . . . . . . . . . . . 36 109 C.5. No New Vulnerabilities . . . . . . . . . . . . . . . . . 36 110 C.6. Detailed Requirements . . . . . . . . . . . . . . . . . . 36 111 C.6.1. General . . . . . . . . . . . . . . . . . . . . . . . 36 112 C.6.2. Performance . . . . . . . . . . . . . . . . . . . . . 38 113 C.6.3. Heterogeneous Support for Solution . . . . . . . . . 40 114 C.6.4. Granular Control . . . . . . . . . . . . . . . . . . 42 115 C.6.5. Priority and Policy . . . . . . . . . . . . . . . . . 43 116 C.6.6. Security . . . . . . . . . . . . . . . . . . . . . . 43 117 C.6.7. Flexibility and Extensibility . . . . . . . . . . . . 44 118 Appendix D. Considerations for Applications Integrating the DOIC 119 Solution . . . . . . . . . . . . . . . . . . . . . . 46 120 D.1. Application Classification . . . . . . . . . . . . . . . 46 121 D.2. Application Type Overload Implications . . . . . . . . . 47 122 D.3. Request Transaction Classification . . . . . . . . . . . 48 123 D.4. Request Type Overload Implications . . . . . . . . . . . 49 124 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 50 126 1. Introduction 128 This specification defines a base solution for Diameter overload 129 control, referred to as Diameter Overload Indication Conveyance 130 (DOIC), based on the requirements identified in [RFC7068]. 132 This specification addresses Diameter overload control between 133 Diameter nodes that support the DOIC solution. The solution, which 134 is designed to apply to existing and future Diameter applications, 135 requires no changes to the Diameter base protocol [RFC6733] and is 136 deployable in environments where some Diameter nodes do not implement 137 the Diameter overload control solution defined in this specification. 139 A new application specification can incorporate the overload control 140 mechanism specified in this document by making it mandatory to 141 implement for the application and referencing this specification 142 normatively. It is the responsibility of the Diameter application 143 designers to define how overload control mechanisms works on that 144 application. 146 Note that the overload control solution defined in this specification 147 does not address all the requirements listed in [RFC7068]. A number 148 of overload control related features are left for future 149 specifications. See Appendix A for a list of extensions that are 150 currently being considered. See Appendix C for an analysis of 151 conformance to the requirements specified in [RFC7068]. 153 2. Terminology and Abbreviations 155 Abatement 157 Reaction to receipt of an overload report resulting in a reduction 158 in traffic sent to the reporting node. Abatement actions include 159 diversion and throttling. 161 Abatement Algorithm 163 An extensible method requested by reporting nodes and used by 164 reacting nodes to reduce the amount of traffic sent during an 165 occurrence of overload control. 167 Diversion 169 An overload abatement treatment where the reacting node selects 170 alternate destinations or paths for requests. 172 Host-Routed Requests 174 Requests that a reacting node knows will be served by a particular 175 host, either due to the presence of a Destination-Host AVP, or by 176 some other local knowledge on the part of the reacting node. 178 Overload Control State (OCS) 180 Internal state maintained by a reporting or reacting node 181 describing occurrences of overload control. 183 Overload Report (OLR) 185 Overload control information for a particular overload occurrence 186 sent by a reporting node. 188 Reacting Node 190 A Diameter node that acts upon an overload report. 192 Realm-Routed Requests 193 Requests that a reacting node does not know which host will 194 service the request. 196 Reporting Node 198 A Diameter node that generates an overload report. (This may or 199 may not be the overloaded node.) 201 Throttling 203 An abatement treatment that limits the number of requests sent by 204 the DIOC reacting node. Throttling can include a Diameter Client 205 choosing to not send requests, or a Diameter Agent or Server 206 rejecting requests with appropriate error responses. In both 207 cases the result of the throttling is a permanent rejection of the 208 transaction. 210 3. Conventions Used in This Document 212 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 213 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 214 document are to be interpreted as described in RFC 2119 [RFC2119]. 216 RFC 2119 [RFC2119] interpretation does not apply for the above listed 217 words when they are not used in all-caps format. 219 4. Solution Overview 221 The Diameter Overload Information Conveyance (DOIC) solution allows 222 Diameter nodes to request other Diameter nodes to perform overload 223 abatement actions, that is, actions to reduce the load offered to the 224 overloaded node or realm. 226 A Diameter node that supports DOIC is known as a "DOIC node". Any 227 Diameter node can act as a DOIC node, including Diameter Clients, 228 Diameter Servers, and Diameter Agents. DOIC nodes are further 229 divided into "Reporting Nodes" and "Reacting Nodes." A reporting 230 node requests overload abatement by sending Overload Reports (OLR). 232 A reacting node acts upon OLRs, and performs whatever actions are 233 needed to fulfill the abatement requests included in the OLRs. A 234 Reporting node may report overload on its own behalf, or on behalf of 235 other nodes. Likewise, a reacting node may perform overload 236 abatement on its own behalf, or on behalf of other nodes. 238 A Diameter node's role as a DOIC node is independent of its Diameter 239 role. For example, Diameter Agents may act as DOIC nodes, even 240 though they are not endpoints in the Diameter sense. Since Diameter 241 enables bi-directional applications, where Diameter Servers can send 242 requests towards Diameter Clients, a given Diameter node can 243 simultaneously act as both a reporting node and a reacting node. 245 Likewise, a Diameter Agent may act as a reacting node from the 246 perspective of upstream nodes, and a reporting node from the 247 perspective of downstream nodes. 249 DOIC nodes do not generate new messages to carry DOIC related 250 information. Rather, they "piggyback" DOIC information over existing 251 Diameter messages by inserting new AVPs into existing Diameter 252 requests and responses. Nodes indicate support for DOIC, and any 253 needed DOIC parameters, by inserting an OC-Supported-Features AVP 254 (Section 7.2) into existing requests and responses. Reporting nodes 255 send OLRs by inserting OC-OLR AVPs (Section 7.3). 257 A given OLR applies to the Diameter realm and application of the 258 Diameter message that carries it. If a reporting node supports more 259 than one realm and/or application, it reports independently for each 260 combination of realm and application. Similarly, the OC-Supported- 261 Features AVP applies to the realm and application of the enclosing 262 message. This implies that a node may support DOIC for one 263 application and/or realm, but not another, and may indicate different 264 DOIC parameters for each application and realm for which it supports 265 DOIC. 267 Reacting nodes perform overload abatement according to an agreed-upon 268 abatement algorithm. An abatement algorithm defines the meaning of 269 some of the parameters of an OLR and the procedures required for 270 overload abatement. An overload abatement algorithm separates 271 Diameter requests into two sets. The first set contains the requests 272 that are to undergo overload abatement treatment of either throttling 273 or diversion. The second set contains the requests that are to be 274 given normal routing treatment. This document specifies a single 275 must-support algorithm, namely the "loss" algorithm (Section 6). 276 Future specifications may introduce new algorithms. 278 Overload conditions may vary in scope. For example, a single 279 Diameter node may be overloaded, in which case reacting nodes may 280 attempt to send requests to other destinations. On the other hand, 281 an entire Diameter realm may be overloaded, in which case such 282 attempts would do harm. DOIC OLRs have a concept of "report type" 283 (Section 7.6), where the type defines such behaviors. Report types 284 are extensible. This document defines report types for overload of a 285 specific host, and for overload of an entire realm. 287 DOIC works through non supporting Diameter Agents that properly pass 288 unknown AVPs unchanged. 290 4.1. Piggybacking 292 There is no new Diameter application defined to carry overload 293 related AVPs. The overload control AVPs defined in this 294 specification have been designed to be piggybacked on top of existing 295 application messages. This is made possible by adding the optional 296 overload control AVPs OC-OLR and OC-Supported-Features into existing 297 commands. 299 Reacting nodes indicate support for DOIC by including the OC- 300 Supported-Features AVP in all request messages originated or relayed 301 by the reacting node. 303 Reporting nodes indicate support for DOIC by including the OC- 304 Supported-Features AVP in all answer messages originated or relayed 305 by the reporting node that are in response to a request that 306 contained the OC-Supported-Features AVP. Reporting nodes may include 307 overload reports using the OC-OLR AVP in answer messages. 309 Note that the overload control solution does not have fixed server 310 and client roles. The DOIC node role is determined based on the 311 message type: whether the message is a request (i.e. sent by a 312 "reacting node") or an answer (i.e. sent by a "reporting node"). 313 Therefore, in a typical "client-server" deployment, the Diameter 314 Client may report its overload condition to the Diameter Server for 315 any Diameter Server initiated message exchange. An example of such 316 is the Diameter Server requesting a re-authentication from a Diameter 317 Client. 319 4.2. DOIC Capability Announcement 321 The DOIC solution supports the ability for Diameter nodes to 322 determine if other nodes in the path of a request support the 323 solution. This capability is referred to as DOIC Capability 324 Announcement (DCA) and is separate from Diameter Capability Exchange. 326 The DCA mechanism uses the OC-Supported-Features AVPs to indicate the 327 Diameter overload features supported. 329 The first node in the path of a Diameter request that supports the 330 DOIC solution inserts the OC-Supported-Features AVP in the request 331 message. 333 The individual features supported by the DOIC nodes are indicated in 334 the OC-Feature-Vector AVP. Any semantics associated with the 335 features will be defined in extension specifications that introduce 336 the features. 338 Note: As discussed elsewhere in the document, agents in the path 339 of the request can modify the OC-Supported-Features AVP. 341 Note: The DOIC solution must support deployments where Diameter 342 Clients and/or Diameter Servers do not support the DOIC solution. 343 In this scenario, Diameter Agents that support the DOIC solution 344 may handle overload abatement for the non supporting Diameter 345 nodes. In this case the DOIC agent will insert the OC-Supported- 346 Features AVP in requests that do not already contain one, telling 347 the reporting node that there is a DOIC node that will handle 348 overload abatement. For transactions where there was an OC- 349 Supporting-Features AVP in the request, the agent will insert the 350 OC-Supported-Features AVP in answers, telling the reacting node 351 that there is a reporting node. 353 The OC-Feature-Vector AVP will always contain an indication of 354 support for the loss overload abatement algorithm defined in this 355 specification (see Section 6). This ensures that a reporting node 356 always supports at least one of the advertized abatement algorithms 357 received in a request messages. 359 The reporting node inserts the OC-Supported-Features AVP in all 360 answer messages to requests that contained the OC-Supported-Features 361 AVP. The contents of the reporting node's OC-Supported-Features AVP 362 indicate the set of Diameter overload features supported by the 363 reporting node. This specification defines one exception - the 364 reporting node only includes an indication of support for one 365 overload abatement algorithm, independent of the number of overload 366 abatement algorithms actually supported by the reacting node. The 367 overload abatement algorithm indicated is the algorithm that the 368 reporting node intends to use should it enter an overload condition. 369 Reacting nodes can use the indicated overload abatement algorithm to 370 prepare for possible overload reports and must use the indicated 371 overload abatement algorithm if traffic reduction is actually 372 requested. 374 Note that the loss algorithm defined in this document is a 375 stateless abatement algorithm. As a result it does not require 376 any actions by reacting nodes prior to the receipt of an overload 377 report. Stateful abatement algorithms that base the abatement 378 logic on a history of request messages sent might require reacting 379 nodes to maintain state in advance of receiving an overload report 380 to ensure that the overload reports can be properly handled. 382 The DCA mechanism must also allow the scenario where the set of 383 features supported by the sender of a request and by agents in the 384 path of a request differ. In this case, the agent can update the OC- 385 Supported-Features AVP to reflect the mixture of the two sets of 386 supported features. 388 Note: The logic to determine if the content of the OC-Supported- 389 Features AVP should be changed is out-of-scope for this document, 390 as is the logic to determine the content of a modified OC- 391 Supported-Features AVP. These are left to implementation 392 decisions. Care must be taken not to introduce interoperability 393 issues for downstream or upstream DOIC nodes. 395 4.3. DOIC Overload Condition Reporting 397 As with DOIC capability announcement, overload condition reporting 398 uses new AVPs (Section 7.3) to indicate an overload condition. 400 The OC-OLR AVP is referred to as an overload report. The OC-OLR AVP 401 includes the type of report, a sequence number, the length of time 402 that the report is valid and abatement algorithm specific AVPs. 404 Two types of overload reports are defined in this document: host 405 reports and realm reports. 407 A report of type "HOST_REPORT" is sent to indicate the overload of a 408 specific host, identified by the Origin-Host AVP of the message 409 containing the OLR, for the application-id indicated in the 410 transaction. When receiving an OLR of type "HOST_REPORT", a reacting 411 node applies overload abatement treatment to the host-routed requests 412 identified by the overload abatement algorithm (see definition in 413 Section 2) sent for this application to the overloaded host. 415 A report of type "REALM_REPORT" is sent to indicate the overload of a 416 realm for the application-id indicated in the transaction. The 417 overloaded realm is identified by the Destination-Realm AVP of the 418 message containing the OLR. When receiving an OLR of type 419 "REALM_REPORT", a reacting node applies overload abatement treatment 420 to realm-routed requests identified by the overload abatement 421 algorithm (see definition in Section 2) sent for this application to 422 the overloaded realm. 424 This document assumes that there is a single source for realm-reports 425 for a given realm, or that if multiple nodes can send realm reports, 426 that each such node has full knowledge of the overload state of the 427 entire realm. A reacting node cannot distinguish between receiving 428 realm-reports from a single node, or from multiple nodes. 430 Note: Known issues exist if multiple sources for overload reports 431 which apply to the same Diameter entity exist. Reacting nodes 432 have no way of determining the source and, as such, will treat 433 them as coming from a single source. Variance in sequence numbers 434 between the two sources can then cause incorrect overload 435 abatement treatment to be applied for indeterminate periods of 436 time. 438 Reporting nodes are responsible for determining the need for a 439 reduction of traffic. The method for making this determination is 440 implementation specific and depends on the type of overload report 441 being generated. A host-report might be generated by tracking use of 442 resources required by the host to handle transactions for the 443 Diameter application. A realm-report generally impacts the traffic 444 sent to multiple hosts and, as such, requires tracking the capacity 445 of all servers able to handle realm-routed requests for the 446 application and realm. 448 Once a reporting node determines the need for a reduction in traffic, 449 it uses the DOIC defined AVPs to report on the condition. These AVPs 450 are included in answer messages sent or relayed by the reporting 451 node. The reporting node indicates the overload abatement algorithm 452 that is to be used to handle the traffic reduction in the OC- 453 Supported-Features AVP. The OC-OLR AVP is used to communicate 454 information about the requested reduction. 456 Reacting nodes, upon receipt of an overload report, apply the 457 overload abatement algorithm to traffic impacted by the overload 458 report. The method used to determine the requests that are to 459 receive overload abatement treatment is dependent on the abatement 460 algorithm. The loss abatement algorithm is defined in this document 461 (Section 6). Other abatement algorithms can be defined in extensions 462 to the DOIC solution. 464 Two types of overload abatement treatment are defined, diversion and 465 throttling. Reacting nodes are responsible for determining which 466 treatment is appropriate for individual requests. 468 As the conditions that lead to the generation of the overload report 469 change the reporting node can send new overload reports requesting 470 greater reduction if the condition gets worse or less reduction if 471 the condition improves. The reporting node sends an overload report 472 with a duration of zero to indicate that the overload condition has 473 ended and abatement is no longer needed. 475 The reacting node also determines when the overload report expires 476 based on the OC-Validity-Duration AVP in the overload report and 477 stops applying the abatement algorithm when the report expires. 479 4.4. DOIC Extensibility 481 The DOIC solution is designed to be extensible. This extensibility 482 is based on existing Diameter based extensibility mechanisms, along 483 with the DOIC capability announcement mechanism. 485 There are multiple categories of extensions that are expected. This 486 includes the definition of new overload abatement algorithms, the 487 definition of new report types and the definition of new scopes of 488 messages impacted by an overload report. 490 A DOIC node communicates supported features by including them in the 491 OC-Feature-Vector AVP, as a sub-AVP of OC-Supported-Features. Any 492 non-backwards compatible DOIC extensions define new values for the 493 OC-Feature-Vector AVP. DOIC extensions also have the ability to add 494 new AVPs to the OC-Supported-Features AVP, if additional information 495 about the new feature is required. 497 Overload reports can also be extended by adding new sub-AVPs to the 498 OC-OLR AVP, allowing reporting nodes to communicate additional 499 information about handling an overload condition. 501 If necessary, new extensions can also define new AVPs that are not 502 part of the OC-Supported-Features and OC-OLR group AVPs. It is, 503 however, recommended that DOIC extensions use the OC-Supported- 504 Features AVP and OC-OLR AVP to carry all DOIC related AVPs. 506 4.5. Simplified Example Architecture 508 Figure 1 illustrates the simplified architecture for Diameter 509 overload information conveyance. 511 Realm X Same or other Realms 512 <--------------------------------------> <----------------------> 514 +--------+ : (optional) : 515 |Diameter| : : 516 |Server A|--+ .--. : +--------+ : .--. 517 +--------+ | _( `. : |Diameter| : _( `. +--------+ 518 +--( )--:-| Agent |-:--( )--|Diameter| 519 +--------+ | ( ` . ) ) : +--------+ : ( ` . ) ) | Client | 520 |Diameter|--+ `--(___.-' : : `--(___.-' +--------+ 521 |Server B| : : 522 +--------+ : : 524 End-to-end Overload Indication 525 1) <-----------------------------------------------> 526 Diameter Application Y 528 Overload Indication A Overload Indication A' 529 2) <----------------------> <----------------------> 530 Diameter Application Y Diameter Application Y 532 Figure 1: Simplified architecture choices for overload indication 533 delivery 535 In Figure 1, the Diameter overload indication can be conveyed (1) 536 end-to-end between servers and clients or (2) between servers and 537 Diameter agent inside the realm and then between the Diameter agent 538 and the clients. 540 5. Solution Procedures 542 This section outlines the normative behavior for the DOIC solution. 544 5.1. Capability Announcement 546 This section defines DOIC Capability Announcement (DCA) behavior. 548 Note: This specification assumes that changes in DOIC node 549 capabilities are relatively rare events that occur as a result of 550 administrative action. Reacting nodes ought to minimize changes 551 that force the reporting node to change the features being used, 552 especially during active overload conditions. But even if 553 reacting nodes avoid such changes, reporting nodes still have to 554 be prepared for them to occur. For example, differing 555 capabilities between multiple reacting nodes may still force a 556 reporting node to select different features on a per-transaction 557 basis. 559 5.1.1. Reacting Node Behavior 561 A reacting node MUST include the OC-Supported-Features AVP in all 562 requests. It MAY include the OC-Feature-Vector AVP, as a sub-avp of 563 OC-Supported-Features. If it does so, it MUST indicate support for 564 the "loss" algorithm. If the reacting node is configured to support 565 features (including other algorithms) in addition to the loss 566 algorithm, it MUST indicate such support in an OC-Feature-Vector AVP. 568 An OC-Supported-Features AVP in answer messages indicates there is a 569 reporting node for the transaction. The reacting node MAY take 570 action, for example creating state for some stateful abatement 571 algorithm, based on the features indicated in the OC-Feature-Vector 572 AVP. 574 Note: The loss abatement algorithm does not require stateful 575 behavior when there is no active overload report. 577 Reacting nodes need to be prepared for the reporting node to change 578 selected algorithms. This can happen at any time, including when the 579 reporting node has sent an active overload report. The reacting node 580 can minimize the potential for changes by modifying the advertised 581 abatement algorithms sent to an overloaded reporting node to the 582 currently selected algorithm and loss (or just loss if it is the 583 currently selected algorithm). This has the effect of limiting the 584 potential change in abatement algorithm from the currently selected 585 algorithm to loss, avoiding changes to more complex abatement 586 algorithms that require state to operate properly. 588 5.1.2. Reporting Node Behavior 590 Upon receipt of a request message, a reporting node determines if 591 there is a reacting node for the transaction based on the presence of 592 the OC-Supported-Features AVP in the request message. 594 If the request message contains an OC-Supported-Features AVP then a 595 reporting node MUST include the OC-Supported-Features AVP in the 596 answer message for that transaction. 598 Note: Capability announcement is done on a per transaction basis. 599 The reporting node cannot assume that the capabilities announced 600 by a reacting node will be the same between transactions. 602 A reporting node MUST NOT include the OC-Supported-Features AVP, OC- 603 OLR AVP or any other overload control AVPs defined in extension 604 drafts in response messages for transactions where the request 605 message does not include the OC-Supported-Features AVP. Lack of the 606 OC-Supported-Features AVP in the request message indicates that there 607 is no reacting node for the transaction. 609 A reporting node knows what overload control functionality is 610 supported by the reacting node based on the content or absence of the 611 OC-Feature-Vector AVP within the OC-Supported-Features AVP in the 612 request message. 614 A reporting node MUST indicate support for one and only one abatement 615 algorithm in the OC-Feature-Vector AVP. The abatement algorithm 616 selected MUST indicate the abatement algorithm the reporting node 617 wants the reacting node to use when the reporting node enters an 618 overload condition. 620 The abatement algorithm selected MUST be from the set of abatement 621 algorithms contained in the request message's OC-Feature-Vector AVP. 623 A reporting node that selects the loss algorithm may do so by 624 including the OC-Feature-Vector AVP with an explicit indication of 625 the loss algorithm, or it MAY omit OC-Feature-Vector. If it selects 626 a different algorithm, it MUST include the OC-Feature-Vector AVP with 627 an explicit indication of the selected algorithm. 629 The reporting node SHOULD indicate support for other DOIC features 630 defined in extension drafts that it supports and that apply to the 631 transaction. It does so using the OC-Feature-Vector AVP. 633 Note: Not all DOIC features will apply to all Diameter 634 applications or deployment scenarios. The features included in 635 the OC-Feature-Vector AVP are based on local reporting node 636 policy. 638 5.1.3. Agent Behavior 640 Diameter Agents that support DOIC can ensure that all messages 641 relayed by the agent contain the OC-Supported-Features AVP. 643 A Diameter Agent MAY take on reacting node behavior for Diameter 644 endpoints that do not support the DOIC solution. A Diameter Agent 645 detects that a Diameter endpoint does not support DOIC reacting node 646 behavior when there is no OC-Supported-Features AVP in a request 647 message. 649 For a Diameter Agent to be a reacting node for a non supporting 650 Diameter endpoint, the Diameter Agent MUST include the OC-Supported- 651 Features AVP in request messages it relays that do not contain the 652 OC-Supported-Features AVP. 654 A Diameter Agent MAY take on reporting node behavior for Diameter 655 endpoints that do not support the DOIC solution. The Diameter Agent 656 MUST have visibility to all traffic destined for the non supporting 657 host in order to become the reporting node for the Diameter endpoint. 658 A Diameter Agent detects that a Diameter endpoint does not support 659 DOIC reporting node behavior when there is no OC-Supported-Features 660 AVP in an answer message for a transaction that contained the OC- 661 Supported-Features AVP in the request message. 663 If a request already has the OC-Supported-Features AVP, a Diameter 664 agent MAY modify it to reflect the features appropriate for the 665 transaction. Otherwise, the agent relays the OC-Supported-Features 666 AVP without change. 668 For instance, if the agent supports a superset of the features 669 reported by the reacting node then the agent might choose, based 670 on local policy, to advertise that superset of features to the 671 reporting node. 673 If the Diameter Agent changes the OC-Supported-Features AVP in a 674 request message then it is likely it will also need to modify the OC- 675 Supported-Features AVP in the answer message for the transaction. A 676 Diameter Agent MAY modify the OC-Supported-Features AVP carried in 677 answer messages. 679 When making changes to the OC-Supported-Features or OC-OLR AVPs, the 680 Diameter Agent needs to ensure consistency in its behavior with both 681 upstream and downstream DOIC nodes. 683 5.2. Overload Report Processing 685 5.2.1. Overload Control State 687 Both reacting and reporting nodes maintain Overload Control State 688 (OCS) for active overload conditions. The following sections define 689 behavior associated with that OCS. 691 The contents of the OCS in the reporting node and in the reacting 692 node represent logical constructs. The actual internal physical 693 structure of the state included in the OCS is an implementation 694 decision. 696 5.2.1.1. Overload Control State for Reacting Nodes 698 A reacting node maintains the following OCS per supported Diameter 699 application: 701 o A host-type OCS entry for each Destination-Host to which it sends 702 host-type requests and 704 o A realm-type OCS entry for each Destination-Realm to which it 705 sends realm-type requests. 707 A host-type OCS entry is identified by the pair of application-id and 708 the node's DiameterIdentity. 710 A realm-type OCS entry is identified by the pair of application-id 711 and realm. 713 The host-type and realm-type OCS entries include the following 714 information (the actual information stored is an implementation 715 decision): 717 o Sequence number (as received in OC-OLR) 719 o Time of expiry (derived from OC-Validity-Duration AVP received in 720 the OC-OLR AVP and time of reception of the message carrying OC- 721 OLR AVP) 723 o Selected Abatement Algorithm (as received in the OC-Supported- 724 Features AVP) 726 o Abatement Algorithm specific input data (as received in the OC-OLR 727 AVP, for example, OC-Reduction-Percentage for the Loss abatement 728 algorithm) 730 5.2.1.2. Overload Control State for Reporting Nodes 732 A reporting node maintains OCS entries per supported Diameter 733 application, per supported (and eventually selected) Abatement 734 Algorithm and per report-type. 736 An OCS entry is identified by the tuple of Application-Id, Report- 737 Type and Abatement Algorithm and includes the following information 738 (the actual information stored is an implementation decision): 740 o Sequence number 742 o Validity Duration 743 o Expiration Time 745 o Algorithm specific input data (for example, the Reduction 746 Percentage for the Loss Abatement Algorithm) 748 5.2.1.3. Reacting Node Maintenance of Overload Control State 750 When a reacting node receives an OC-OLR AVP, it MUST determine if it 751 is for an existing or new overload condition. 753 Note: For the remainder of this section the term OLR refers to the 754 combination of the contents of the received OC-OLR AVP and the 755 abatement algorithm indicated in the received OC-Supported- 756 Features AVP. 758 When receiving an answer message with multiple OLRs of different 759 supported report types, a reacting node MUST process each received 760 OLR. 762 The OLR is for an existing overload condition if a reacting node has 763 an OCS that matches the received OLR. 765 For a host-report this means it matches the application-id and the 766 host's DiameterIdentity in an existing host OCS entry. 768 For a realm-report this means it matches the application-id and the 769 realm in an existing realm OCS entry. 771 If the OLR is for an existing overload condition then a reacting node 772 MUST determine if the OLR is a retransmission or an update to the 773 existing OLR. 775 If the sequence number for the received OLR is greater than the 776 sequence number stored in the matching OCS entry then a reacting node 777 MUST update the matching OCS entry. 779 If the sequence number for the received OLR is less than or equal to 780 the sequence number in the matching OCS entry then a reacting node 781 MUST silently ignore the received OLR. The matching OCS MUST NOT be 782 updated in this case. 784 If the received OLR is for a new overload condition then a reacting 785 node MUST generate a new OCS entry for the overload condition. 787 For a host-report this means a reacting node creates on OCS entry 788 with the application-id in the received message and DiameterIdentity 789 of the Origin-Host in the received message. 791 Note: This solution assumes that the Origin-Host AVP in the answer 792 message included by the reporting node is not changed along the 793 path to the reacting node. 795 For a realm-report this means a reacting node creates on OCS entry 796 with the application-id in the received message and realm of the 797 Origin-Realm in the received message. 799 If the received OLR contains a validity duration of zero ("0") then a 800 reacting node MUST update the OCS entry as being expired. 802 Note: It is not necessarily appropriate to delete the OCS entry, 803 as there is recommended behavior that the reacting node slowly 804 returns to full traffic when ending an overload abatement period. 806 The reacting node does not delete an OCS when receiving an answer 807 message that does not contain an OC-OLR AVP (i.e. absence of OLR 808 means "no change"). 810 5.2.1.4. Reporting Node Maintenance of Overload Control State 812 A reporting node SHOULD create a new OCS entry when entering an 813 overload condition. 815 Note: If a reporting node knows through absence of the OC- 816 Supported-Features AVP in received messages that there are no 817 reacting nodes supporting DOIC then the reporting node can choose 818 to not create OCS entries. 820 When generating a new OCS entry the sequence number SHOULD be set to 821 zero ("0"). 823 When generating sequence numbers for new overload conditions, the new 824 sequence number MUST be greater than any sequence number in an active 825 (unexpired) overload report for the same application and report-type 826 previously sent by the reporting node. This property MUST hold over 827 a reboot of the reporting node. 829 Note: One way of addressing this over a reboot of a reporting node 830 is to use a time stamp for the first overload condition that 831 occurs after the report and to start using sequences beginning 832 with zero for subsequent overload conditions. 834 A reporting node MUST update an OCS entry when it needs to adjust the 835 validity duration of the overload condition at reacting nodes. 837 For instance, if a reporting node wishes to instruct reacting 838 nodes to continue overload abatement for a longer period of time 839 than originally communicated. This also applies if the reporting 840 node wishes to shorten the period of time that overload abatement 841 is to continue. 843 A reporting node MUST update an OCS entry when it wishes to adjust 844 any abatement algorithm specific parameters, including, for example, 845 the reduction percentage used for the Loss abatement algorithm. 847 For instance, if a reporting node wishes to change the reduction 848 percentage either higher, if the overload condition has worsened, 849 or lower, if the overload condition has improved, then the 850 reporting node would update the appropriate OCS entry. 852 A reporting node MUST increment the sequence number associated with 853 the OCS entry anytime the contents of the OCS entry are changed. 854 This will result in a new sequence number being sent to reacting 855 nodes, instructing reacting nodes to process the OC-OLR AVP. 857 A reporting node SHOULD update an OCS entry with a validity duration 858 of zero ("0") when the overload condition ends. 860 Note: If a reporting node knows that the OCS entries in the 861 reacting nodes are near expiration then the reporting node might 862 decide not to send an OLR with a validity duration of zero. 864 A reporting node MUST keep an OCS entry with a validity duration of 865 zero ("0") for a period of time long enough to ensure that any non- 866 expired reacting node's OCS entry created as a result of the overload 867 condition in the reporting node is deleted. 869 5.2.2. Reacting Node Behavior 871 When a reacting node sends a request it MUST determine if that 872 request matches an active OCS. 874 If the request matches an active OCS then the reacting node MUST use 875 the overload abatement algorithm indicated in the OCS to determine if 876 the request is to receive overload abatement treatment. 878 For the Loss abatement algorithm defined in this specification, see 879 Section 6 for the overload abatement algorithm logic applied. 881 If the overload abatement algorithm selects the request for overload 882 abatement treatment then the reacting node MUST apply overload 883 abatement treatment on the request. The abatement treatment applied 884 depends on the context of the request. 886 If diversion abatement treatment is possible (i.e. a different path 887 for the request can be selected where the overloaded node is not part 888 of the different path), then the reacting node SHOULD apply diversion 889 abatement treatment to the request. The reacting node MUST apply 890 throttling abatement treatment to requests identified for abatement 891 treatment when diversion treatment is not possible or was not 892 applied. 894 Note: This only addresses the case where there are two defined 895 abatement treatments, diversion and throttling. Any extension 896 that defines a new abatement treatment must also defined the 897 interaction of the new abatement treatment with existing 898 treatments. 900 If the overload abatement treatment results in throttling of the 901 request and if the reacting node is an agent then the agent MUST send 902 an appropriate error as defined in Section 8. 904 Diameter endpoints that throttle requests need to do so according to 905 the rules of the client application. Those rules will vary by 906 application, and are beyond the scope of this document. 908 In the case that the OCS entry indicated no traffic was to be sent to 909 the overloaded entity and the validity duration expires then overload 910 abatement associated with the overload report MUST be ended in a 911 controlled fashion. 913 5.2.3. Reporting Node Behavior 915 If there is an active OCS entry then a reporting node SHOULD include 916 the OC-OLR AVP in all answers to requests that contain the OC- 917 Supported-Features AVP and that match the active OCS entry. 919 Note: A request matches if the application-id in the request 920 matches the application-id in any active OCS entry and if the 921 report-type in the OCS entry matches a report-type supported by 922 the reporting node as indicated in the OC-Supported-Features AVP. 924 The contents of the OC-OLR AVP depend on the selected algorithm. 926 A reporting node MAY choose to not resend an overload report to a 927 reacting node if it can guarantee that this overload report is 928 already active in the reacting node. 930 Note: In some cases (e.g. when there are one or more agents in the 931 path between reporting and reacting nodes, or when overload 932 reports are discarded by reacting nodes) a reporting node may not 933 be able to guarantee that the reacting node has received the 934 report. 936 A reporting node MUST NOT send overload reports of a type that has 937 not been advertised as supported by the reacting node. 939 Note: A reacting node implicitly advertises support for the host 940 and realm report types by including the OC-Supported-Features AVP 941 in the request. Support for other report types will be explicitly 942 indicated by new feature bits in the OC-Feature-Vector AVP. 944 A reporting node SHOULD explicitly indicate the end of an overload 945 occurrence by sending a new OLR with OC-Validity-Duration set to a 946 value of zero ("0"). The reporting node SHOULD ensure that all 947 reacting nodes receive the updated overload report. 949 A reporting node MAY rely on the OC-Validity-Duration AVP values for 950 the implicit overload control state cleanup on the reacting node. 952 Note: All OLRs sent have an expiration time calculated by adding 953 the validity-duration contained in the OLR to the time the message 954 was sent. Transit time for the OLR can be safely ignored. The 955 reporting node can ensure that all reacting nodes have received 956 the OLR by continuing to send it in answer messages until the 957 expiration time for all OLRs sent for that overload condition have 958 expired. 960 When a reporting node sends an OLR, it effectively delegates any 961 necessary throttling to downstream nodes. If the reporting node also 962 locally throttles the same set of messages, the overall number of 963 throttled requests may be higher than intended. Therefore, before 964 applying local message throttling, a reporting node needs to check if 965 these messages match existing OCS entries, indicating that these 966 messages have survived throttling applied by downstream nodes that 967 have received the related OLR. 969 However, even if the set of messages match existing OCS entries, the 970 reporting node can still apply other abatement methods such as 971 diversion. The reporting node might also need to throttle requests 972 for reasons other than overload. For example, an agent or server 973 might have a configured rate limit for each client, and throttle 974 requests that exceed that limit, even if such requests had already 975 been candidates for throttling by downstream nodes. The reporting 976 node also has the option to send new OLRs requesting greater 977 reductions in traffic, reducing the need for local throttling. 979 A reporting node SHOULD decrease requested overload abatement 980 treatment in a controlled fashion to avoid oscillations in traffic. 982 For example, it might wait some period of time after overload ends 983 before terminating the OLR, or it might send a series of OLRs 984 indicating progressively less overload severity. 986 5.3. Protocol Extensibility 988 The DOIC solution can be extended. Types of potential extensions 989 include new traffic abatement algorithms, new report types or other 990 new functionality. 992 When defining a new extension that requires new normative behavior, 993 the specification MUST define a new feature for the OC-Feature- 994 Vector. This feature bit is used to communicate support for the new 995 feature. 997 The extension MAY define new AVPs for use in DOIC Capability 998 Announcement and for use in DOIC Overload reporting. These new AVPs 999 SHOULD be defined to be extensions to the OC-Supported-Features or 1000 OC-OLR AVPs defined in this document. 1002 [RFC6733] defined Grouped AVP extension mechanisms apply. This 1003 allows, for example, defining a new feature that is mandatory to be 1004 understood even when piggybacked on an existing application. 1006 When defining new report type values, the corresponding specification 1007 MUST define the semantics of the new report types and how they affect 1008 the OC-OLR AVP handling. 1010 The OC-Supported-Feature and OC-OLR AVPs can be expanded with 1011 optional sub-AVPs only if a legacy DOIC implementation can safely 1012 ignore them without breaking backward compatibility for the given OC- 1013 Report-Type AVP value. Any new sub-AVPs MUST NOT require that the 1014 M-bit be set. 1016 Documents that introduce new report types MUST describe any 1017 limitations on their use across non-supporting agents. 1019 As with any Diameter specification, RFC6733 requires all new AVPs to 1020 be registered with IANA. See Section 9 for the required procedures. 1021 New features (feature bits in the OC-Feature-Vector AVP) and report 1022 types (in the OC-Report-Type AVP) MUST be registered with IANA. 1024 6. Loss Algorithm 1026 This section documents the Diameter overload loss abatement 1027 algorithm. 1029 6.1. Overview 1031 The DOIC specification supports the ability for multiple overload 1032 abatement algorithms to be specified. The abatement algorithm used 1033 for any instance of overload is determined by the Diameter Overload 1034 Capability Announcement process documented in Section 5.1. 1036 The loss algorithm described in this section is the default algorithm 1037 that must be supported by all Diameter nodes that support DOIC. 1039 The loss algorithm is designed to be a straightforward and stateless 1040 overload abatement algorithm. It is used by reporting nodes to 1041 request a percentage reduction in the amount of traffic sent. The 1042 traffic impacted by the requested reduction depends on the type of 1043 overload report. 1045 Reporting nodes request the stateless reduction of the number of 1046 requests by an indicated percentage. This percentage reduction is in 1047 comparison to the number of messages the node otherwise would send, 1048 regardless of how many requests the node might have sent in the past. 1050 From a conceptual level, the logic at the reacting node could be 1051 outlined as follows. 1053 1. An overload report is received and the associated OCS is either 1054 saved or updated (if required) by the reacting node. 1056 2. A new Diameter request is generated by the application running on 1057 the reacting node. 1059 3. The reacting node determines that an active overload report 1060 applies to the request, as indicated by the corresponding OCS 1061 entry. 1063 4. The reacting node determines if overload abatement treatment 1064 should be applied to the request. One approach that could be 1065 taken for each request is to select a random number between 1 and 1066 100. If the random number is less than or equal to the indicated 1067 reduction percentage then the request is given abatement 1068 treatment, otherwise the request is given normal routing 1069 treatment. 1071 6.2. Reporting Node Behavior 1073 The method a reporting node uses to determine the amount of traffic 1074 reduction required to address an overload condition is an 1075 implementation decision. 1077 When a reporting node that has selected the loss abatement algorithm 1078 determines the need to request a reduction in traffic, it includes an 1079 OC-OLR AVP in answer messages as described in Section 5.2.3. 1081 When sending the OC-OLR AVP, the reporting node MUST indicate a 1082 percentage reduction in the OC-Reduction-Percentage AVP. 1084 The reporting node MAY change the reduction percentage in subsequent 1085 overload reports. When doing so the reporting node must conform to 1086 overload report handing specified in Section 5.2.3. 1088 6.3. Reacting Node Behavior 1090 The method a reacting node uses to determine which request messages 1091 are given abatement treatment is an implementation decision. 1093 When receiving an OC-OLR in an answer message where the algorithm 1094 indicated in the OC-Supported-Features AVP is the loss algorithm, the 1095 reacting node MUST apply abatement treatment to the requested 1096 percentage of request messages sent. 1098 Note: The loss algorithm is a stateless algorithm. As a result, 1099 the reacting node does not guarantee that there will be an 1100 absolute reduction in traffic sent. Rather, it guarantees that 1101 the requested percentage of new requests will be given abatement 1102 treatment. 1104 If reacting node comes out of the 100 percent traffic reduction as a 1105 result of the overload report timing out, the reacting node sending 1106 the traffic SHOULD be conservative and, for example, first send 1107 "probe" messages to learn the overload condition of the overloaded 1108 node before converging to any traffic amount/rate decided by the 1109 sender. Similar concerns apply in all cases when the overload report 1110 times out unless the previous overload report stated 0 percent 1111 reduction. 1113 The goal of this behavior is to reduce the probability of overload 1114 condition thrashing where an immediate transition from 100% 1115 reduction to 0% reduction results in the reporting node moving 1116 quickly back into an overload condition. 1118 7. Attribute Value Pairs 1120 This section describes the encoding and semantics of the Diameter 1121 Overload Indication Attribute Value Pairs (AVPs) defined in this 1122 document. 1124 7.1. OC-Supported-Features AVP 1126 The OC-Supported-Features AVP (AVP code TBD1) is of type Grouped and 1127 serves two purposes. First, it announces a node's support for the 1128 DOIC solution in general. Second, it contains the description of the 1129 supported DOIC features of the sending node. The OC-Supported- 1130 Features AVP MUST be included in every Diameter request message a 1131 DOIC supporting node sends. 1133 OC-Supported-Features ::= < AVP Header: TBD1 > 1134 [ OC-Feature-Vector ] 1135 * [ AVP ] 1137 7.2. OC-Feature-Vector AVP 1139 The OC-Feature-Vector AVP (AVP code TBD2) is of type Unsigned64 and 1140 contains a 64 bit flags field of announced capabilities of a DOIC 1141 node. The value of zero (0) is reserved. 1143 The OC-Feature-Vector sub-AVP is used to announce the DOIC features 1144 supported by the DOIC node, in the form of a flag-bits field in which 1145 each bit announces one feature or capability supported by the node. 1146 The absence of the OC-Feature-Vector AVP in request messages 1147 indicates that only the default traffic abatement algorithm described 1148 in this specification is supported. The absence of the OC- Feature- 1149 Vector AVP in answer messages indicates that the default traffic 1150 abatement algorithm described in this specification is selected 1151 (while other traffic abatement algorithms may be supported), and no 1152 features other than abatement algorithms are supported. 1154 The following capabilities are defined in this document: 1156 OLR_DEFAULT_ALGO (0x0000000000000001) 1158 When this flag is set by the a DOIC reacting node it means that 1159 the default traffic abatement (loss) algorithm is supported. When 1160 this flag is set by a DOIC reporting node it means that the loss 1161 algorithm will be used for requested overload abatement. 1163 7.3. OC-OLR AVP 1165 The OC-OLR AVP (AVP code TBD3) is of type Grouped and contains the 1166 information necessary to convey an overload report on an overload 1167 condition at the reporting node. The application the OC-OLR AVP 1168 applies to is the same as the Application-Id found in the Diameter 1169 message header. The host or realm the OC-OLR AVP concerns is 1170 determined from the Origin-Host AVP and/or Origin-Realm AVP found in 1171 the encapsulating Diameter command. The OC-OLR AVP is intended to be 1172 sent only by a reporting node. 1174 OC-OLR ::= < AVP Header: TBD2 > 1175 < OC-Sequence-Number > 1176 < OC-Report-Type > 1177 [ OC-Reduction-Percentage ] 1178 [ OC-Validity-Duration ] 1179 * [ AVP ] 1181 7.4. OC-Sequence-Number AVP 1183 The OC-Sequence-Number AVP (AVP code TBD4) is of type Unsigned64. 1184 Its usage in the context of overload control is described in 1185 Section 5.2. 1187 From the functionality point of view, the OC-Sequence-Number AVP is 1188 used as a non-volatile increasing counter for a sequence of overload 1189 reports between two DOIC nodes for the same overload occurrence. 1190 Sequence numbers are treated in a uni-directional manner, i.e. two 1191 sequence numbers on each direction between two DOIC nodes are not 1192 related or correlated. 1194 7.5. OC-Validity-Duration AVP 1196 The OC-Validity-Duration AVP (AVP code TBD5) is of type Unsigned32 1197 and indicates in seconds the validity time of the overload report. 1198 The number of seconds is measured after reception of the first OC-OLR 1199 AVP with a given value of OC-Sequence-Number AVP. The default value 1200 for the OC-Validity-Duration AVP is 30 seconds. When the OC- 1201 Validity-Duration AVP is not present in the OC-OLR AVP, the default 1202 value applies. The maximum value for the OC-Validity-Duration AVP is 1203 86,400 seconds (24 hours). 1205 7.6. OC-Report-Type AVP 1207 The OC-Report-Type AVP (AVP code TBD6) is of type Enumerated. The 1208 value of the AVP describes what the overload report concerns. The 1209 following values are initially defined: 1211 HOST_REPORT 0 The overload report is for a host. Overload abatement 1212 treatment applies to host-routed requests. 1214 REALM_REPORT 1 The overload report is for a realm. Overload 1215 abatement treatment applies to realm-routed requests. 1217 7.7. OC-Reduction-Percentage AVP 1219 The OC-Reduction-Percentage AVP (AVP code TBD7) is of type Unsigned32 1220 and describes the percentage of the traffic that the sender is 1221 requested to reduce, compared to what it otherwise would send. The 1222 OC-Reduction-Percentage AVP applies to the default (loss) algorithm 1223 specified in this specification. However, the AVP can be reused for 1224 future abatement algorithms, if its semantics fit into the new 1225 algorithm. 1227 The value of the Reduction-Percentage AVP is between zero (0) and one 1228 hundred (100). Values greater than 100 are ignored. The value of 1229 100 means that all traffic is to be throttled, i.e. the reporting 1230 node is under a severe load and ceases to process any new messages. 1231 The value of 0 means that the reporting node is in a stable state and 1232 has no need for the reacting node to apply any traffic abatement. 1234 7.8. Attribute Value Pair flag rules 1236 +---------+ 1237 |AVP flag | 1238 |rules | 1239 +----+----+ 1240 AVP Section | |MUST| 1241 Attribute Name Code Defined Value Type |MUST| NOT| 1242 +--------------------------------------------------+----+----+ 1243 |OC-Supported-Features TBD1 7.1 Grouped | | V | 1244 +--------------------------------------------------+----+----+ 1245 |OC-Feature-Vector TBD2 7.2 Unsigned64 | | V | 1246 +--------------------------------------------------+----+----+ 1247 |OC-OLR TBD3 7.3 Grouped | | V | 1248 +--------------------------------------------------+----+----+ 1249 |OC-Sequence-Number TBD4 7.4 Unsigned64 | | V | 1250 +--------------------------------------------------+----+----+ 1251 |OC-Validity-Duration TBD5 7.5 Unsigned32 | | V | 1252 +--------------------------------------------------+----+----+ 1253 |OC-Report-Type TBD6 7.6 Enumerated | | V | 1254 +--------------------------------------------------+----+----+ 1255 |OC-Reduction | | | 1256 | -Percentage TBD7 7.7 Unsigned32 | | V | 1257 +--------------------------------------------------+----+----+ 1259 As described in the Diameter base protocol [RFC6733], the M-bit usage 1260 for a given AVP in a given command may be defined by the application. 1262 8. Error Response Codes 1264 When a DOIC node rejects a Diameter request due to overload, the DOIC 1265 node MUST select an appropriate error response code. This 1266 determination is made based on the probability of the request 1267 succeeding if retried on a different path. 1269 Note: This only applies for DOIC nodes that are not the originator 1270 of the request. 1272 A reporting node rejecting a Diameter request due to an overload 1273 condition SHOULD send a DIAMETER_TOO_BUSY error response, if it can 1274 assume that the same request may succeed on a different path. 1276 If a reporting node knows or assumes that the same request will not 1277 succeed on a different path, DIAMETER_UNABLE_TO_COMPLY error response 1278 SHOULD be used. Retrying would consume valuable resources during an 1279 occurrence of overload. 1281 For instance, if the request arrived at the reporting node without 1282 a Destination-Host AVP then the reporting node might determine 1283 that there is an alternative Diameter node that could successfully 1284 process the request and that retrying the transaction would not 1285 negatively impact the reporting node. DIAMETER_TOO_BUSY would be 1286 sent in this case. 1288 If the request arrived at the reporting node with a Destination- 1289 Host AVP populated with its own Diameter identity then the 1290 reporting node can assume that retrying the request would result 1291 in it coming to the same reporting node. 1292 DIAMETER_UNABLE_TO_COMPLY would be sent in this case. 1294 A second example is when an agent that supports the DOIC solution 1295 is performing the role of a reacting node for a non supporting 1296 client. Requests that are rejected as a result of DOIC throttling 1297 by the agent in this scenario would generally be rejected with a 1298 DIAMETER_UNABLE_TO_COMPLY response code. 1300 9. IANA Considerations 1302 9.1. AVP codes 1304 New AVPs defined by this specification are listed in Section 7. All 1305 AVP codes are allocated from the 'Authentication, Authorization, and 1306 Accounting (AAA) Parameters' AVP Codes registry. 1308 9.2. New registries 1310 Two new registries are needed under the 'Authentication, 1311 Authorization, and Accounting (AAA) Parameters' registry. 1313 A new "Overload Control Feature Vector" registry is required. The 1314 registry must contain the following: 1316 Feature Vector Value 1318 Specification - the specification that defines the new value. 1320 See Section 7.2 for the initial Feature Vector Value in the registry. 1321 This specification is the specification defining the value. New 1322 values can be added into the registry using the Specification 1323 Required policy. [RFC5226]. 1325 A new "Overload Report Type" registry is required. The registry must 1326 contain the following: 1328 Report Type Value 1330 Specification - the specification that defines the new value. 1332 See Section 7.6 for the initial assignment in the registry. New 1333 types can be added using the Specification Required policy [RFC5226]. 1335 10. Security Considerations 1337 DOIC gives Diameter nodes the ability to request that downstream 1338 nodes send fewer Diameter requests. Nodes do this by exchanging 1339 overload reports that directly effect this reduction. This exchange 1340 is potentially subject to multiple methods of attack, and has the 1341 potential to be used as a Denial-of-Service (DoS) attack vector. 1343 Overload reports may contain information about the topology and 1344 current status of a Diameter network. This information is 1345 potentially sensitive. Network operators may wish to control 1346 disclosure of overload reports to unauthorized parties to avoid its 1347 use for competitive intelligence or to target attacks. 1349 Diameter does not include features to provide end-to-end 1350 authentication, integrity protection, or confidentiality. This may 1351 cause complications when sending overload reports between non- 1352 adjacent nodes. 1354 10.1. Potential Threat Modes 1356 The Diameter protocol involves transactions in the form of requests 1357 and answers exchanged between clients and servers. These clients and 1358 servers may be peers, that is, they may share a direct transport 1359 (e.g. TCP or SCTP) connection, or the messages may traverse one or 1360 more intermediaries, known as Diameter Agents. Diameter nodes use 1361 TLS, DTLS, or IPsec to authenticate peers, and to provide 1362 confidentiality and integrity protection of traffic between peers. 1363 Nodes can make authorization decisions based on the peer identities 1364 authenticated at the transport layer. 1366 When agents are involved, this presents an effectively transitive 1367 trust model. That is, a Diameter client or server can authorize an 1368 agent for certain actions, but it must trust that agent to make 1369 appropriate authorization decisions about its peers, and so on. 1370 Since confidentiality and integrity protection occurs at the 1371 transport layer, agents can read, and perhaps modify, any part of a 1372 Diameter message, including an overload report. 1374 There are several ways an attacker might attempt to exploit the 1375 overload control mechanism. An unauthorized third party might inject 1376 an overload report into the network. If this third party is upstream 1377 of an agent, and that agent fails to apply proper authorization 1378 policies, downstream nodes may mistakenly trust the report. This 1379 attack is at least partially mitigated by the assumption that nodes 1380 include overload reports in Diameter answers but not in requests. 1381 This requires an attacker to have knowledge of the original request 1382 in order to construct an answer. Such an answer would also need to 1383 arrive at a Diameter node via a protected transport connection. 1384 Therefore, implementations MUST validate that an answer containing an 1385 overload report is a properly constructed response to a pending 1386 request prior to acting on the overload report, and that the answer 1387 was received via an appropriate transport connection. 1389 A similar attack involves a compromised but otherwise authorized node 1390 that sends an inappropriate overload report. For example, a server 1391 for the realm "example.com" might send an overload report indicating 1392 that a competitor's realm "example.net" is overloaded. If other 1393 nodes act on the report, they may falsely believe that "example.net" 1394 is overloaded, effectively reducing that realm's capacity. 1395 Therefore, it's critical that nodes validate that an overload report 1396 received from a peer actually falls within that peer's responsibility 1397 before acting on the report or forwarding the report to other peers. 1398 For example, an overload report from a peer that applies to a realm 1399 not handled by that peer is suspect. 1401 This attack is partially mitigated by the fact that the 1402 application, as well as host and realm, for a given OLR is 1403 determined implicitly by respective AVPs in the enclosing answer. 1404 If a reporting node modifies any of those AVPs, the enclosing 1405 transaction will also be affected. 1407 10.2. Denial of Service Attacks 1409 Diameter overload reports, especially realm-reports, can cause a node 1410 to cease sending some or all Diameter requests for an extended 1411 period. This makes them a tempting vector for DoS attacks. 1412 Furthermore, since Diameter is almost always used in support of other 1413 protocols, a DoS attack on Diameter is likely to impact those 1414 protocols as well. Therefore, Diameter nodes MUST NOT honor or 1415 forward OLRs received from peers that are not trusted to send them. 1417 An attacker might use the information in an OLR to assist in DoS 1418 attacks. For example, an attacker could use information about 1419 current overload conditions to time an attack for maximum effect, or 1420 use subsequent overload reports as a feedback mechanism to learn the 1421 results of a previous or ongoing attack. Operators need the ability 1422 to ensure that OLRs are not leaked to untrusted parties. 1424 10.3. Non-Compliant Nodes 1426 In the absence of an overload control mechanism, Diameter nodes need 1427 to implement strategies to protect themselves from floods of 1428 requests, and to make sure that a disproportionate load from one 1429 source does not prevent other sources from receiving service. For 1430 example, a Diameter server might throttle a certain percentage of 1431 requests from sources that exceed certain limits. Overload control 1432 can be thought of as an optimization for such strategies, where 1433 downstream nodes never send the excess requests in the first place. 1434 However, the presence of an overload control mechanism does not 1435 remove the need for these other protection strategies. 1437 When a Diameter node sends an overload report, it cannot assume that 1438 all nodes will comply, even if they indicate support for DOIC. A 1439 non-compliant node might continue to send requests with no reduction 1440 in load. Such non-compliance could be done accidentally, or 1441 maliciously to gain an unfair advantage over compliant nodes. 1442 Requirement 28 [RFC7068] indicates that the overload control solution 1443 cannot assume that all Diameter nodes in a network are trusted, and 1444 that malicious nodes not be allowed to take advantage of the overload 1445 control mechanism to get more than their fair share of service. 1447 10.4. End-to End-Security Issues 1449 The lack of end-to-end integrity features makes it difficult to 1450 establish trust in overload reports received from non-adjacent nodes. 1451 Any agents in the message path may insert or modify overload reports. 1452 Nodes must trust that their adjacent peers perform proper checks on 1453 overload reports from their peers, and so on, creating a transitive- 1454 trust requirement extending for potentially long chains of nodes. 1455 Network operators must determine if this transitive trust requirement 1456 is acceptable for their deployments. Nodes supporting Diameter 1457 overload control MUST give operators the ability to select which 1458 peers are trusted to deliver overload reports, and whether they are 1459 trusted to forward overload reports from non-adjacent nodes. DOIC 1460 nodes MUST strip DOIC AVPs from messages received from peers that are 1461 not trusted for DOIC purposes. 1463 The lack of end-to-end confidentiality protection means that any 1464 Diameter agent in the path of an overload report can view the 1465 contents of that report. In addition to the requirement to select 1466 which peers are trusted to send overload reports, operators MUST be 1467 able to select which peers are authorized to receive reports. A node 1468 MUST NOT send an overload report to a peer not authorized to receive 1469 it. Furthermore, an agent MUST remove any overload reports that 1470 might have been inserted by other nodes before forwarding a Diameter 1471 message to a peer that is not authorized to receive overload reports. 1473 A DOIC node cannot always automatically detect that a peer also 1474 supports DOIC. For example, a node might have a peer that is a 1475 non-supporting agent. If nodes on the other side of that agent 1476 send OC-Supported-Features AVPs, the agent is likely to forward 1477 them as unknown AVPs. Messages received across the non-supporting 1478 agent may be indistinguishable from messages received across a 1479 DOIC supporting agent, giving the false impression that the non- 1480 supporting agent actually supports DOIC. This complicates the 1481 transitive-trust nature of DOIC. Operators need to be careful to 1482 avoid situations where a non-supporting agent is mistakenly 1483 trusted to enforce DOIC related authorization policies. 1485 At the time of this writing, the DIME working group is studying 1486 requirements for adding end-to-end security features 1487 [I-D.ietf-dime-e2e-sec-req] to Diameter. These features, when they 1488 become available, might make it easier to establish trust in non- 1489 adjacent nodes for overload control purposes. Readers should be 1490 reminded, however, that the overload control mechanism encourages 1491 Diameter agents to modify AVPs in, or insert additional AVPs into, 1492 existing messages that are originated by other nodes. If end-to-end 1493 security is enabled, there is a risk that such modification could 1494 violate integrity protection. The details of using any future 1495 Diameter end-to-end security mechanism with overload control will 1496 require careful consideration, and are beyond the scope of this 1497 document. 1499 11. Contributors 1501 The following people contributed substantial ideas, feedback, and 1502 discussion to this document: 1504 o Eric McMurry 1506 o Hannes Tschofenig 1508 o Ulrich Wiehe 1510 o Jean-Jacques Trottin 1512 o Maria Cruz Bartolome 1514 o Martin Dolly 1516 o Nirav Salot 1518 o Susan Shishufeng 1520 12. References 1522 12.1. Normative References 1524 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1525 Requirement Levels", BCP 14, RFC 2119, March 1997. 1527 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 1528 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 1529 May 2008. 1531 [RFC6733] Fajardo, V., Arkko, J., Loughney, J., and G. Zorn, 1532 "Diameter Base Protocol", RFC 6733, October 2012. 1534 12.2. Informative References 1536 [Cx] 3GPP, , "ETSI TS 129 229 V11.4.0", August 2013. 1538 [I-D.ietf-dime-e2e-sec-req] 1539 Tschofenig, H., Korhonen, J., Zorn, G., and K. Pillay, 1540 "Diameter AVP Level Security: Scenarios and Requirements", 1541 draft-ietf-dime-e2e-sec-req-01 (work in progress), October 1542 2013. 1544 [PCC] 3GPP, , "ETSI TS 123 203 V11.12.0", December 2013. 1546 [RFC4006] Hakala, H., Mattila, L., Koskinen, J-P., Stura, M., and J. 1547 Loughney, "Diameter Credit-Control Application", RFC 4006, 1548 August 2005. 1550 [RFC7068] McMurry, E. and B. Campbell, "Diameter Overload Control 1551 Requirements", RFC 7068, November 2013. 1553 [S13] 3GPP, , "ETSI TS 129 272 V11.9.0", December 2012. 1555 Appendix A. Issues left for future specifications 1557 The base solution for the overload control does not cover all 1558 possible use cases. A number of solution aspects were intentionally 1559 left for future specification and protocol work. The following sub- 1560 sections define some of the potential extensions to the DOIC 1561 solution. 1563 A.1. Additional traffic abatement algorithms 1565 This specification describes only means for a simple loss based 1566 algorithm. Future algorithms can be added using the designed 1567 solution extension mechanism. The new algorithms need to be 1568 registered with IANA. See Sections 7.1 and 9 for the required IANA 1569 steps. 1571 A.2. Agent Overload 1573 This specification focuses on Diameter endpoint (server or client) 1574 overload. A separate extension will be required to outline the 1575 handling of the case of agent overload. 1577 A.3. New Error Diagnostic AVP 1579 This specification indicates the use of existing error messages when 1580 nodes reject requests due to overload. The DIME working group is 1581 considering defining additional error codes or AVPs to indicate that 1582 overload was the reason for the rejection of the message. 1584 Appendix B. Deployment Considerations 1586 Non Supporting Agents 1588 Due to the way that realm-routed requests are handled in Diameter 1589 networks with the server selection for the request done by an 1590 agent, network operators should enable DOIC at agents that perform 1591 server selection first. 1593 Topology Hiding Interactions 1595 There exist proxies that implement what is referred to as Topology 1596 Hiding. This can include cases where the agent modifies the 1597 Origin-Host in answer messages. The behavior of the DOIC solution 1598 is not well understood when this happens. As such, the DOIC 1599 solution does not address this scenario. 1601 Appendix C. Requirements Conformance Analysis 1603 This section contains the result of an analysis of the DOIC solutions 1604 conformance to the requirements defined in [RFC7068]. 1606 C.1. Deferred Requirements 1608 The 3GPP has adopted an early version of this document as normative 1609 references in various Diameter related specifications to support the 1610 overload control mechanism in their release 12 framework. The DIME 1611 working group has therefore decided to defer certain requirements in 1612 order to complete the design of an extensible, generic solution 1613 before the deadline scheduled by the 3GPP for the completion of the 1614 release 12 protocol work by the end of 2014. The deferred work 1615 includes the following: 1617 o Agent Overload - The ability for an agent to report an overload 1618 condition of the agent itself. 1620 o Load Information - The ability for a node to report its load level 1621 when not overloaded. 1623 At the time of this writing, DIME has begun separate work efforts for 1624 these requirements. 1626 C.2. Detection of non-supporting Intermediaries 1628 The DOIC mechanism as currently defined does not allow supporting 1629 nodes to automatically determine whether OC-Supported-Features or OC- 1630 OLR AVPs are originated by a peer node, or by a non-peer node and 1631 sent across a non-supporting peer. This makes it impossible to 1632 detect the presence of non-supporting nodes between supporting nodes, 1633 except by configuration. The working group determined that such a 1634 configuration requirement is acceptable. 1636 This limits full compliance with certain requirements related to the 1637 limitation of new configuration, deployment in environments with 1638 mixed support, operating across non-supporting agents, and 1639 authorization. 1641 C.3. Implicit Application Indication 1643 The working group elected to determine the application for an 1644 overload report from that of the enclosing message. This prevents 1645 sending an OLR for an application when there are no transactions for 1646 that application. 1648 As a consequence, DOIC does not comply with the requirement to be 1649 able to report overload information across quiescent connections. 1650 DOIC does not fully comply with requirements to operate on up-to-date 1651 information, since if an OLR causes all transactions to stop for an 1652 application, the only way traffic will resume is for the OLR to 1653 expire. 1655 C.4. Stateless Operation 1657 RFC7068 explicitly discourages the sending of OLRs in every answer 1658 message, as part of the requirement to avoid additional work for 1659 overloaded nodes. DOIC recommends exactly that behavior during 1660 active overload conditions. The working group determined that doing 1661 otherwise would reduce reliability and increase statefulness. (Note 1662 that DOIC does allow nodes to avoid sending OLRs in every answer if 1663 they have some other method of ensuring that OLRs get to all relevant 1664 reacting nodes.) 1666 C.5. No New Vulnerabilities 1668 The working group believes that DOIC is compliant with the 1669 requirement to avoid introducing new vulnerabilities. However, this 1670 requirement may warrant an early security expert review. 1672 C.6. Detailed Requirements 1674 [RFC Editor: Please remove this section and subsections prior to 1675 publication as an RFC.] 1677 C.6.1. General 1679 REQ 1: The solution MUST provide a communication method for Diameter 1680 nodes to exchange load and overload information. 1682 *Partially Compliant*. The mechanism uses new AVPs 1683 piggybacked on existing Diameter messages to exchange 1684 overload information. It does not currently support "load" 1685 information or the ability to report overload of an agent. 1686 These have been left for future extensions. 1688 REQ 2: The solution MUST allow Diameter nodes to support overload 1689 control regardless of which Diameter applications they 1690 support. Diameter clients and agents must be able to use the 1691 received load and overload information to support graceful 1692 behavior during an overload condition. Graceful behavior 1693 under overload conditions is best described by REQ 3. 1695 *Partially Compliant*. The DOIC AVPs can be used in any 1696 application that allows the extension of AVPs. However, 1697 "load" information is not currently supported. 1699 REQ 3: The solution MUST limit the impact of overload on the overall 1700 useful throughput of a Diameter server, even when the 1701 incoming load on the network is far in excess of its 1702 capacity. The overall useful throughput under load is the 1703 ultimate measure of the value of a solution. 1705 *Compliant*. DOIC provides information that nodes can use to 1706 reduce the impact of overload. 1708 REQ 4: Diameter allows requests to be sent from either side of a 1709 connection, and either side of a connection may have need to 1710 provide its overload status. The solution MUST allow each 1711 side of a connection to independently inform the other of its 1712 overload status. 1714 *Compliant*. DOIC AVPs can be included regardless of 1715 transaction "direction" 1717 REQ 5: Diameter allows nodes to determine their peers via dynamic 1718 discovery or manual configuration. The solution MUST work 1719 consistently without regard to how peers are determined. 1721 *Compliant*. DOIC contains no assumptions about how peers are 1722 discovered. 1724 REQ 6: The solution designers SHOULD seek to minimize the amount of 1725 new configuration required in order to work. For example, it 1726 is better to allow peers to advertise or negotiate support 1727 for the solution, rather than to require that this knowledge 1728 to be configured at each node. 1730 *Partially Compliant*. Most DOIC parameters are advertised 1731 using the DOIC capability announcement mechanism. However, 1732 there are some situations where configuration is required. 1733 For example, a DOIC node detect the fact that a peer may not 1734 support DOIC when nodes on the other side of the non- 1735 supporting node do support DOIC without configuration. 1737 C.6.2. Performance 1739 REQ 7: The solution and any associated default algorithm(s) MUST 1740 ensure that the system remains stable. At some point after 1741 an overload condition has ended, the solution MUST enable 1742 capacity to stabilize and become equal to what it would be in 1743 the absence of an overload condition. Note that this also 1744 requires that the solution MUST allow nodes to shed load 1745 without introducing non-converging oscillations during or 1746 after an overload condition. 1748 *Compliant*. The specification offers guidance that 1749 implementations should apply hysteresis when recovering from 1750 overload, and avoid sudden ramp ups in offered load when 1751 recovering. 1753 REQ 8: Supporting nodes MUST be able to distinguish current overload 1754 information from stale information. 1756 *Partially Compliant*. DOIC overload reports are "soft 1757 state", that is they expire after an indicated period. DOIC 1758 nodes may also send reports that end existing overload 1759 conditions. DOIC requires reporting nodes to ensure that all 1760 relevant reacting nodes receive overload reports. 1762 However, since DOIC does not allow reporting nodes to send 1763 OLRs in watchdog messages, if an overload condition results 1764 in zero offered load, the reporting node cannot update the 1765 condition until the expiration of the original OLR. 1767 REQ 9: The solution MUST function across fully loaded as well as 1768 quiescent transport connections. This is partially derived 1769 from the requirement for stability in REQ 7. 1771 *Not Compliant*. DOIC does not allow OLRs to be sent over 1772 quiescent transport connections. This is due to the fact 1773 that OLRs cannot be sent outside of the application to which 1774 they apply. 1776 REQ 10: Consumers of overload information MUST be able to determine 1777 when the overload condition improves or ends. 1779 *Partially Compliant*. (See response to previous two 1780 requirements.) 1782 REQ 11: The solution MUST be able to operate in networks of different 1783 sizes. 1785 *Compliant*. DOIC makes no assumptions about the size of the 1786 network. DOIC can operate purely between clients and 1787 servers, or across agents. 1789 REQ 12: When a single network node fails, goes into overload, or 1790 suffers from reduced processing capacity, the solution MUST 1791 make it possible to limit the impact of the affected node on 1792 other nodes in the network. This helps to prevent a small- 1793 scale failure from becoming a widespread outage. 1795 *Partially Compliant*. DOIC allows overload reports for an 1796 entire realm, where abated traffic will not be redirected 1797 towards another server. But in situations where nodes choose 1798 to divert traffic to other nodes, DOIC offers no way of 1799 knowing whether the new recipients can handle the traffic if 1800 they have not already indicated overload. This may be 1801 mitigated with the use of a future "load" extension, or with 1802 the use of proprietary dynamic load-balancing mechanisms. 1804 REQ 13: The solution MUST NOT introduce substantial additional work 1805 for a node in an overloaded state. For example, a 1806 requirement for an overloaded node to send overload 1807 information every time it received a new request would 1808 introduce substantial work. 1810 *Not Compliant*. DOIC does in fact encourage an overloaded 1811 node to send an OLR in every response. The working group 1812 that other mechanisms to ensure that every relevant node 1813 receives an OLR would create even more work. [Note: This 1814 needs discussion.] 1816 REQ 14: Some scenarios that result in overload involve a rapid 1817 increase of traffic with little time between normal levels 1818 and levels that induce overload. The solution SHOULD provide 1819 for rapid feedback when traffic levels increase. 1821 *Compliant*. The piggyback mechanism allows OLRs to be sent 1822 at the same rate as application traffic. 1824 REQ 15: The solution MUST NOT interfere with the congestion control 1825 mechanisms of underlying transport protocols. For example, a 1826 solution that opened additional TCP connections when the 1827 network is congested would reduce the effectiveness of the 1828 underlying congestion control mechanisms. 1830 *Compliant*. DOIC does not require or recommend changes in 1831 the handling of transport protocols or connections. 1833 C.6.3. Heterogeneous Support for Solution 1835 REQ 16: The solution is likely to be deployed incrementally. The 1836 solution MUST support a mixed environment where some, but not 1837 all, nodes implement it. 1839 *Partially Compliant*. DOIC works with most mixed-deployment 1840 scenarios. However, it cannot work across a non-supporting 1841 proxy that modifies Origin-Host AVPs in answer messages. 1842 DOIC will have limited impact in networks where the nodes 1843 that perform server selections do not support the mechanism. 1845 REQ 17: In a mixed environment with nodes that support the solution 1846 and nodes that do not, the solution MUST NOT result in 1847 materially less useful throughput during overload as would 1848 have resulted if the solution were not present. It SHOULD 1849 result in less severe overload in this environment. 1851 *Compliant*. In most mixed-support deployment, DOIC will 1852 offer at least some value, and will not make things worse. 1854 REQ 18: In a mixed environment of nodes that support the solution and 1855 nodes that do not, the solution MUST NOT preclude elements 1856 that support overload control from treating elements that do 1857 not support overload control in an equitable fashion relative 1858 to those that do. Users and operators of nodes that do not 1859 support the solution MUST NOT unfairly benefit from the 1860 solution. The solution specification SHOULD provide guidance 1861 to implementers for dealing with elements not supporting 1862 overload control. 1864 *Compliant*. DOIC provides mechanisms to abate load from non- 1865 supporting sources. Furthermore, it recommends that 1866 reporting nodes will still need to be able to apply whatever 1867 protections they would ordinarily apply if DOIC were not in 1868 use. 1870 REQ 19: It MUST be possible to use the solution between nodes in 1871 different realms and in different administrative domains. 1873 *Partially Compliant*. DOIC allows sending OLRs across 1874 administrative domains, and potentially to nodes in other 1875 realms. However, an OLR cannot indicate overload for realms 1876 other than the one in the Origin-Realm AVP of the containing 1877 answer. 1879 REQ 20: Any explicit overload indication MUST be clearly 1880 distinguishable from other errors reported via Diameter. 1882 *Compliant*. DOIC sends explicit overload indication in 1883 overload reports. It does not depend on error result codes. 1885 REQ 21: In cases where a network node fails, is so overloaded that it 1886 cannot process messages, or cannot communicate due to a 1887 network failure, it may not be able to provide explicit 1888 indications of the nature of the failure or its levels of 1889 overload. The solution MUST result in at least as much 1890 useful throughput as would have resulted if the solution were 1891 not in place. 1893 *Compliant*. DOIC overload reports have the primary effect of 1894 suppressing message retries in overload conditions. DOIC 1895 recommends that messages never be silently dropped if at all 1896 possible. 1898 C.6.4. Granular Control 1900 REQ 22: The solution MUST provide a way for a node to throttle the 1901 amount of traffic it receives from a peer node. This 1902 throttling SHOULD be graded so that it can be applied 1903 gradually as offered load increases. Overload is not a 1904 binary state; there may be degrees of overload. 1906 *Compliant*. The "loss" algorithm expresses a percentage 1907 reduction. 1909 REQ 23: The solution MUST provide sufficient information to enable a 1910 load-balancing node to divert messages that are rejected or 1911 otherwise throttled by an overloaded upstream node to other 1912 upstream nodes that are the most likely to have sufficient 1913 capacity to process them. 1915 *Not Compliant*. DOIC provides no built in mechanism to 1916 determine the best place to divert messages that would 1917 otherwise be throttled. This can be accomplished with a 1918 future "load" extension, or with proprietary load balancing 1919 mechanisms. 1921 REQ 24: The solution MUST provide a mechanism for indicating load 1922 levels, even when not in an overload condition, to assist 1923 nodes in making decisions to prevent overload conditions from 1924 occurring. 1926 *Not Compliant*. "Load" information has been left for a 1927 future extension. 1929 C.6.5. Priority and Policy 1931 REQ 25: The base specification for the solution SHOULD offer general 1932 guidance on which message types might be desirable to send or 1933 process over others during times of overload, based on 1934 application-specific considerations. For example, it may be 1935 more beneficial to process messages for existing sessions 1936 ahead of new sessions. Some networks may have a requirement 1937 to give priority to requests associated with emergency 1938 sessions. Any normative or otherwise detailed definition of 1939 the relative priorities of message types during an overload 1940 condition will be the responsibility of the application 1941 specification. 1943 *Compliant*. The specification offers guidance on how 1944 requests might be prioritized for different types of 1945 applications. 1947 REQ 26: The solution MUST NOT prevent a node from prioritizing 1948 requests based on any local policy, so that certain requests 1949 are given preferential treatment, given additional 1950 retransmission, not throttled, or processed ahead of others. 1952 *Compliant*. Nothing in the specification prevents 1953 application-specific, implementation-specific, or local 1954 policies. 1956 C.6.6. Security 1958 REQ 27: The solution MUST NOT provide new vulnerabilities to 1959 malicious attack or increase the severity of any existing 1960 vulnerabilities. This includes vulnerabilities to DoS and 1961 DDoS attacks as well as replay and man-in-the-middle attacks. 1962 Note that the Diameter base specification [RFC6733] lacks 1963 end-to-end security and this must be considered (see the 1964 Security Considerations in [RFC7068]). Note that this 1965 requirement was expressed at a high level so as to not 1966 preclude any particular solution. It is expected that the 1967 solution will address this in more detail. 1969 *Compliant*. The working group is not aware of any such 1970 vulnerabilities. [This may need further analysis.] 1972 REQ 28: The solution MUST NOT depend on being deployed in 1973 environments where all Diameter nodes are completely trusted. 1974 It SHOULD operate as effectively as possible in environments 1975 where other nodes are malicious; this includes preventing 1976 malicious nodes from obtaining more than a fair share of 1977 service. Note that this does not imply any responsibility on 1978 the solution to detect, or take countermeasures against, 1979 malicious nodes. 1981 *Partially Compliant*. Since all Diameter security is 1982 currently at the transport layer, nodes must trust immediate 1983 peers to enforce trust policies. However, there are 1984 situations where a DOIC node cannot determine if an immediate 1985 peer supports DOIC. The authors recommend an expert security 1986 review. 1988 REQ 29: It MUST be possible for a supporting node to make 1989 authorization decisions about what information will be sent 1990 to peer nodes based on the identity of those nodes. This 1991 allows a domain administrator who considers the load of their 1992 nodes to be sensitive information to restrict access to that 1993 information. Of course, in such cases, there is no 1994 expectation that the solution itself will help prevent 1995 overload from that peer node. 1997 *Partially Compliant*. (See response to previous 1998 requirement.) 2000 REQ 30: The solution MUST NOT interfere with any Diameter-compliant 2001 method that a node may use to protect itself from overload 2002 from non-supporting nodes or from denial-of-service attacks. 2004 *Compliant*. The specification recommends that any such 2005 protection mechanism needed without DOIC should continue to 2006 be employed with DOIC. 2008 C.6.7. Flexibility and Extensibility 2010 REQ 31: There are multiple situations where a Diameter node may be 2011 overloaded for some purposes but not others. For example, 2012 this can happen to an agent or server that supports multiple 2013 applications, or when a server depends on multiple external 2014 resources, some of which may become overloaded while others 2015 are fully available. The solution MUST allow Diameter nodes 2016 to indicate overload with sufficient granularity to allow 2017 clients to take action based on the overloaded resources 2018 without unreasonably forcing available capacity to go unused. 2019 The solution MUST support specification of overload 2020 information with granularities of at least "Diameter node", 2021 "realm", and "Diameter application" and MUST allow 2022 extensibility for others to be added in the future. 2024 *Partially Compliant*. All DOIC overload reports are scoped 2025 to the specific application and realm. Inside that scope, 2026 overload can be reported at the specific server or whole 2027 realm scope. As currently specified, DOIC cannot indicate 2028 local overload for an agent. At the time of this writing, 2029 the DIME working group has plans to work on an agent-overload 2030 extension. 2032 DOIC allows new "scopes" through the use of extended report 2033 types. 2035 REQ 32: The solution MUST provide a method for extending the 2036 information communicated and the algorithms used for overload 2037 control. 2039 *Compliant*. DOIC allows new report types and abatement 2040 algorithms to be created. These may be indicated using the 2041 OC-Supported-Features AVP. 2043 REQ 33: The solution MUST provide a default algorithm that is 2044 mandatory to implement. 2046 *Compliant*. The "loss" algorithm is mandatory to implement. 2048 REQ 34: The solution SHOULD provide a method for exchanging overload 2049 and load information between elements that are connected by 2050 intermediaries that do not support the solution. 2052 *Partially Compliant*. DOIC information can traverse non- 2053 supporting agents, as long as those agents do not modify 2054 certain AVPs. (e.g., Origin-Host). DOIC does not provide a 2055 way for supporting nodes to detect such modification. 2057 Appendix D. Considerations for Applications Integrating the DOIC 2058 Solution 2060 This section outlines considerations to be taken into account when 2061 integrating the DOIC solution into Diameter applications. 2063 D.1. Application Classification 2065 The following is a classification of Diameter applications and 2066 request types. This discussion is meant to document factors that 2067 play into decisions made by the Diameter identity responsible for 2068 handling overload reports. 2070 Section 8.1 of [RFC6733] defines two state machines that imply two 2071 types of applications, session-less and session-based applications. 2072 The primary difference between these types of applications is the 2073 lifetime of Session-Ids. 2075 For session-based applications, the Session-Id is used to tie 2076 multiple requests into a single session. 2078 The Credit-Control application defined in [RFC4006] is an example of 2079 a Diameter session-based application. 2081 In session-less applications, the lifetime of the Session-Id is a 2082 single Diameter transaction, i.e. the session is implicitly 2083 terminated after a single Diameter transaction and a new Session-Id 2084 is generated for each Diameter request. 2086 For the purposes of this discussion, session-less applications are 2087 further divided into two types of applications: 2089 Stateless Applications: 2091 Requests within a stateless application have no relationship to 2092 each other. The 3GPP defined S13 application is an example of a 2093 stateless application [S13], where only a Diameter command is 2094 defined between a client and a server and no state is maintained 2095 between two consecutive transactions. 2097 Pseudo-Session Applications: 2099 Applications that do not rely on the Session-Id AVP for 2100 correlation of application messages related to the same session 2101 but use other session-related information in the Diameter requests 2102 for this purpose. The 3GPP defined Cx application [Cx] is an 2103 example of a pseudo-session application. 2105 The handling of overload reports must take the type of application 2106 into consideration, as discussed in Appendix D.2. 2108 D.2. Application Type Overload Implications 2110 This section discusses considerations for mitigating overload 2111 reported by a Diameter entity. This discussion focuses on the type 2112 of application. Appendix D.3 discusses considerations for handling 2113 various request types when the target server is known to be in an 2114 overloaded state. 2116 These discussions assume that the strategy for mitigating the 2117 reported overload is to reduce the overall workload sent to the 2118 overloaded entity. The concept of applying overload treatment to 2119 requests targeted for an overloaded Diameter entity is inherent to 2120 this discussion. The method used to reduce offered load is not 2121 specified here but could include routing requests to another Diameter 2122 entity known to be able to handle them, or it could mean rejecting 2123 certain requests. For a Diameter agent, rejecting requests will 2124 usually mean generating appropriate Diameter error responses. For a 2125 Diameter client, rejecting requests will depend upon the application. 2126 For example, it could mean giving an indication to the entity 2127 requesting the Diameter service that the network is busy and to try 2128 again later. 2130 Stateless Applications: 2132 By definition there is no relationship between individual requests 2133 in a stateless application. As a result, when a request is sent 2134 or relayed to an overloaded Diameter entity - either a Diameter 2135 Server or a Diameter Agent - the sending or relaying entity can 2136 choose to apply the overload treatment to any request targeted for 2137 the overloaded entity. 2139 Pseudo-Session Applications: 2141 For pseudo-session applications, there is an implied ordering of 2142 requests. As a result, decisions about which requests towards an 2143 overloaded entity to reject could take the command code of the 2144 request into consideration. This generally means that 2145 transactions later in the sequence of transactions should be given 2146 more favorable treatment than messages earlier in the sequence. 2147 This is because more work has already been done by the Diameter 2148 network for those transactions that occur later in the sequence. 2149 Rejecting them could result in increasing the load on the network 2150 as the transactions earlier in the sequence might also need to be 2151 repeated. 2153 Session-Based Applications: 2155 Overload handling for session-based applications must take into 2156 consideration the work load associated with setting up and 2157 maintaining a session. As such, the entity sending requests 2158 towards an overloaded Diameter entity for a session-based 2159 application might tend to reject new session requests prior to 2160 rejecting intra-session requests. In addition, session ending 2161 requests might be given a lower probability of being rejected as 2162 rejecting session ending requests could result in session status 2163 being out of sync between the Diameter clients and servers. 2164 Application designers that would decide to reject mid-session 2165 requests will need to consider whether the rejection invalidates 2166 the session and any resulting session cleanup procedures. 2168 D.3. Request Transaction Classification 2170 Independent Request: 2172 An independent request is not correlated to any other requests 2173 and, as such, the lifetime of the session-id is constrained to an 2174 individual transaction. 2176 Session-Initiating Request: 2178 A session-initiating request is the initial message that 2179 establishes a Diameter session. The ACR message defined in 2180 [RFC6733] is an example of a session-initiating request. 2182 Correlated Session-Initiating Request: 2184 There are cases when multiple session-initiated requests must be 2185 correlated and managed by the same Diameter server. It is notably 2186 the case in the 3GPP PCC architecture [PCC], where multiple 2187 apparently independent Diameter application sessions are actually 2188 correlated and must be handled by the same Diameter server. 2190 Intra-Session Request: 2192 An intra-session request is a request that uses the same Session- 2193 Id than the one used in a previous request. An intra-session 2194 request generally needs to be delivered to the server that handled 2195 the session creating request for the session. The STR message 2196 defined in [RFC6733] is an example of an intra-session request. 2198 Pseudo-Session Requests: 2200 Pseudo-session requests are independent requests and do not use 2201 the same Session-Id but are correlated by other session-related 2202 information contained in the request. There exists Diameter 2203 applications that define an expected ordering of transactions. 2204 This sequencing of independent transactions results in a pseudo 2205 session. The AIR, MAR and SAR requests in the 3GPP defined Cx 2206 [Cx] application are examples of pseudo-session requests. 2208 D.4. Request Type Overload Implications 2210 The request classes identified in Appendix D.3 have implications on 2211 decisions about which requests should be throttled first. The 2212 following list of request treatment regarding throttling is provided 2213 as guidelines for application designers when implementing the 2214 Diameter overload control mechanism described in this document. The 2215 exact behavior regarding throttling is a matter of local policy, 2216 unless specifically defined for the application. 2218 Independent Requests: 2220 Independent requests can generally be given equal treatment when 2221 making throttling decisions, unless otherwise indicated by 2222 application requirements or local policy. 2224 Session-Initiating Requests: 2226 Session-initiating requests often represent more work than 2227 independent or intra-session requests. Moreover, session- 2228 initiating requests are typically followed by other session- 2229 related requests. Since the main objective of the overload 2230 control is to reduce the total number of requests sent to the 2231 overloaded entity, throttling decisions might favor allowing 2232 intra-session requests over session-initiating requests. In the 2233 absence of local policies or application specific requirements to 2234 the contrary, Individual session-initiating requests can be given 2235 equal treatment when making throttling decisions. 2237 Correlated Session-Initiating Requests: 2239 A Request that results in a new binding, where the binding is used 2240 for routing of subsequent session-initiating requests to the same 2241 server, represents more work load than other requests. As such, 2242 these requests might be throttled more frequently than other 2243 request types. 2245 Pseudo-Session Requests: 2247 Throttling decisions for pseudo-session requests can take into 2248 consideration where individual requests fit into the overall 2249 sequence of requests within the pseudo session. Requests that are 2250 earlier in the sequence might be throttled more aggressively than 2251 requests that occur later in the sequence. 2253 Intra-Session Requests: 2255 There are two types of intra-sessions requests, requests that 2256 terminate a session and the remainder of intra-session requests. 2257 Implementers and operators may choose to throttle session- 2258 terminating requests less aggressively in order to gracefully 2259 terminate sessions, allow cleanup of the related resources (e.g. 2260 session state) and avoid the need for additional intra-session 2261 requests. Favoring session-termination requests may reduce the 2262 session management impact on the overloaded entity. The default 2263 handling of other intra-session requests might be to treat them 2264 equally when making throttling decisions. There might also be 2265 application level considerations whether some request types are 2266 favored over others. 2268 Authors' Addresses 2270 Jouni Korhonen (editor) 2271 Broadcom 2272 Porkkalankatu 24 2273 Helsinki FIN-00180 2274 Finland 2276 Email: jouni.nospam@gmail.com 2278 Steve Donovan (editor) 2279 Oracle 2280 7460 Warren Parkway 2281 Frisco, Texas 75034 2282 United States 2284 Email: srdonovan@usdonovans.com 2286 Ben Campbell 2287 Oracle 2288 7460 Warren Parkway 2289 Frisco, Texas 75034 2290 United States 2292 Email: ben@nostrum.com 2293 Lionel Morand 2294 Orange Labs 2295 38/40 rue du General Leclerc 2296 Issy-Les-Moulineaux Cedex 9 92794 2297 France 2299 Phone: +33145296257 2300 Email: lionel.morand@orange.com