idnits 2.17.00 (12 Aug 2021) /tmp/idnits15680/draft-ietf-dime-ovli-03.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 : ---------------------------------------------------------------------------- ** There are 6 instances of too long lines in the document, the longest one being 14 characters in excess of 72. ** The abstract seems to contain references ([RFC2119]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 641 has weird spacing: '...rotocol stan...' == The document seems to use 'NOT RECOMMENDED' as an RFC 2119 keyword, but does not include the phrase in its RFC 2119 key words list. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: The lack of end-to-end confidentiality protection means that any Diameter agent in the path of an overload report can view the contents of that report. In addition to the requirement to select which peers are trusted to send overload reports, operators MUST be able to select which peers are authorized to receive reports. A node MUST not send an overload report to a peer not authorized to receive it. Furthermore, an agent MUST remove any overload reports that might have been inserted by other nodes before forwarding a Diameter message to a peer that is not authorized to receive overload reports. -- The document date (July 3, 2014) is 2879 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) == Unused Reference: 'RFC5905' is defined on line 1824, but no explicit reference was found in the text == Unused Reference: 'RFC5729' is defined on line 1847, but no explicit reference was found in the text ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) == Outdated reference: draft-ietf-dime-e2e-sec-req has been published as RFC 7966 -- Obsolete informational reference (is this intentional?): RFC 4006 (Obsoleted by RFC 8506) Summary: 3 errors (**), 0 flaws (~~), 7 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: January 4, 2015 B. Campbell 6 Oracle 7 L. Morand 8 Orange Labs 9 July 3, 2014 11 Diameter Overload Indication Conveyance 12 draft-ietf-dime-ovli-03.txt 14 Abstract 16 This specification documents a Diameter Overload Control (DOC) base 17 solution and the dissemination of the overload report information. 19 Requirements 21 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 22 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 23 document are to be interpreted as described in RFC 2119 [RFC2119]. 25 Status of This Memo 27 This Internet-Draft is submitted in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF). Note that other groups may also distribute 32 working documents as Internet-Drafts. The list of current Internet- 33 Drafts is at http://datatracker.ietf.org/drafts/current/. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 This Internet-Draft will expire on January 4, 2015. 42 Copyright Notice 44 Copyright (c) 2014 IETF Trust and the persons identified as the 45 document authors. All rights reserved. 47 This document is subject to BCP 78 and the IETF Trust's Legal 48 Provisions Relating to IETF Documents 49 (http://trustee.ietf.org/license-info) in effect on the date of 50 publication of this document. Please review these documents 51 carefully, as they describe your rights and restrictions with respect 52 to this document. Code Components extracted from this document must 53 include Simplified BSD License text as described in Section 4.e of 54 the Trust Legal Provisions and are provided without warranty as 55 described in the Simplified BSD License. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 60 2. Terminology and Abbreviations . . . . . . . . . . . . . . . . 4 61 3. Solution Overview . . . . . . . . . . . . . . . . . . . . . . 4 62 3.1. Overload Control Endpoints (Non normative) . . . . . . . 6 63 3.2. Piggybacking Principle (Non normative) . . . . . . . . . 10 64 3.3. DOIC Capability Announcement (Non normative) . . . . . . 11 65 3.4. DOIC Overload Condition Reporting (Non normative) . . . . 12 66 3.5. DOIC Extensibility (Non normative) . . . . . . . . . . . 13 67 3.6. Simplified Example Architecture (Non normative) . . . . . 14 68 3.7. Considerations for Applications Integrating the DOIC 69 Solution (Non normative) . . . . . . . . . . . . . . . . 15 70 3.7.1. Application Classification (Non normative) . . . . . 15 71 3.7.2. Application Type Overload Implications (Non 72 normative) . . . . . . . . . . . . . . . . . . . . . 16 73 3.7.3. Request Transaction Classification (Non normative) . 18 74 3.7.4. Request Type Overload Implications (Non normative) . 18 75 4. Solution Procedures (Normative) . . . . . . . . . . . . . . . 20 76 4.1. Capability Announcement (Normative) . . . . . . . . . . . 20 77 4.1.1. Reacting Node Behavior (Normative) . . . . . . . . . 20 78 4.1.2. Reporting Node Behavior (Normative) . . . . . . . . 21 79 4.1.3. Agent Behavior (Normative) . . . . . . . . . . . . . 22 80 4.2. Overload Report Processing (Normative) . . . . . . . . . 22 81 4.2.1. Overload Control State (Normative) . . . . . . . . . 22 82 4.2.2. Reacting Node Behavior (Normative) . . . . . . . . . 24 83 4.2.3. Reporting Node Behavior (Normative) . . . . . . . . 26 84 4.2.4. Agent Behavior (Normative) . . . . . . . . . . . . . 26 85 4.3. Protocol Extensibility (Normative) . . . . . . . . . . . 27 86 5. Loss Algorithm (Normative) . . . . . . . . . . . . . . . . . 28 87 5.1. Overview (Non normative) . . . . . . . . . . . . . . . . 28 88 5.2. Use of OC-Reduction-Percentage AVP . . . . . . . . . . . 29 89 5.3. Reporting Node Behavior (Normative) . . . . . . . . . . . 29 90 5.4. Reacting Node Behavior (Normative) . . . . . . . . . . . 29 91 6. Attribute Value Pairs (Normative) . . . . . . . . . . . . . . 30 92 6.1. OC-Supported-Features AVP . . . . . . . . . . . . . . . . 31 93 6.2. OC-Feature-Vector AVP . . . . . . . . . . . . . . . . . . 31 94 6.3. OC-OLR AVP . . . . . . . . . . . . . . . . . . . . . . . 32 95 6.4. OC-Sequence-Number AVP . . . . . . . . . . . . . . . . . 33 96 6.5. OC-Validity-Duration AVP . . . . . . . . . . . . . . . . 33 97 6.6. OC-Report-Type AVP . . . . . . . . . . . . . . . . . . . 34 98 6.7. OC-Reduction-Percentage AVP . . . . . . . . . . . . . . . 35 99 6.8. Attribute Value Pair flag rules . . . . . . . . . . . . . 35 100 7. Error Response Codes . . . . . . . . . . . . . . . . . . . . 36 101 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 36 102 8.1. AVP codes . . . . . . . . . . . . . . . . . . . . . . . . 36 103 8.2. New registries . . . . . . . . . . . . . . . . . . . . . 37 104 9. Security Considerations . . . . . . . . . . . . . . . . . . . 37 105 9.1. Potential Threat Modes . . . . . . . . . . . . . . . . . 37 106 9.2. Denial of Service Attacks . . . . . . . . . . . . . . . . 38 107 9.3. Non-Compliant Nodes . . . . . . . . . . . . . . . . . . . 39 108 9.4. End-to End-Security Issues . . . . . . . . . . . . . . . 39 109 10. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 40 110 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 40 111 11.1. Normative References . . . . . . . . . . . . . . . . . . 40 112 11.2. Informative References . . . . . . . . . . . . . . . . . 41 113 Appendix A. Issues left for future specifications . . . . . . . 41 114 A.1. Additional traffic abatement algorithms . . . . . . . . . 41 115 A.2. Agent Overload . . . . . . . . . . . . . . . . . . . . . 41 116 A.3. DIAMETER_TOO_BUSY clarifications . . . . . . . . . . . . 42 117 Appendix B. Examples . . . . . . . . . . . . . . . . . . . . . . 42 118 B.1. Mix of Destination-Realm routed requests and Destination- 119 Host routed requests . . . . . . . . . . . . . . . . . . 42 120 Appendix C. Restructuring of -02 version of the draft . . . . . 45 121 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 48 123 1. Introduction 125 This specification defines a base solution for Diameter Overload 126 Control (DOC), refered to as Diameter Overload Indication Conveyance 127 (DOIC). The requirements for the solution are described and 128 discussed in the corresponding design requirements document 129 [RFC7068]. Note that the overload control solution defined in this 130 specification does not address all the requirements listed in 131 [RFC7068]. A number of overload control related features are left 132 for the future specifications. 134 The solution defined in this specification addresses Diameter 135 overload control between two endpoints (see Section 3.1). 136 Furthermore, the solution is designed to apply to existing and future 137 Diameter applications, requires no changes to the Diameter base 138 protocol [RFC6733] and is deployable in environments where some 139 Diameter nodes do not implement the Diameter overload control 140 solution defined in this specification. 142 2. Terminology and Abbreviations 144 Abatement Algorithm 146 An algorithm requested by reporting nodes and used by reacting 147 nodes to reduce the amount of traffic sent during an occurrence of 148 overload control. 150 Throttling 152 Throttling is the reduction of the number of requests sent to an 153 entity. Throttling can include a client dropping requests, or an 154 agent rejecting requests with appropriate error responses. 155 Clients and agents can also choose to redirect throttled requests 156 to some other entity or entities capable of handling them. 158 Editor's note: Propose to add a definition of Abatement to include 159 both throttling and diversion (redirecting of messages) actions. 160 Then to modify this definition to include just the rejecting of 161 requests and adding a definition of diversion. 163 Reporting Node 165 A Diameter node that generates an overload report. (This may or 166 may not be the overloaded node.) 168 Reacting Node 170 A Diameter node that consumes and acts upon a report. Note that 171 "act upon" does not necessarily mean the reacting node applies an 172 abatement algorithm; it might decide to delegate that downstream, 173 in which case it also becomes a "reporting node". 175 Overload Control State (OCS) 177 State describing an occurrence of overload control maintained by 178 reporting and reacting nodes. 180 Overload Report (OLR) 182 A set of AVPs sent by a reporting node indicating the start or 183 continuation of an occurrence of overload control. 185 3. Solution Overview 187 The Diameter Overload Information Conveyance (DOIC) mechanism allows 188 Diameter nodes to request other nodes to perform overload abatement 189 actions, that is, actions to reduce the load offered to the 190 overloaded node or realm. 192 A Diameter node that supports DOIC is known as a "DOIC endpoint". 193 Any Diameter node can act as a DOIC endpoint, including clients, 194 servers, and agents. DOIC endpoints are further divided into 195 "Reporting Nodes" and "Reacting Nodes." A reporting node requests 196 overload abatement by sending an Overload Report (OLR) to one or more 197 reacting nodes. 199 A reacting node consumes OLRs, and performs whatever actions are 200 needed to fulfill the abatement requests included in the OLRs. A 201 Reporting node may report overload on its own behalf, or on behalf of 202 other (typically upstream) nodes. Likewise, a reacting node may 203 perform overload abatement on its own behalf, or on behalf of other 204 (typically downstream) nodes. 206 A node's role as a DOIC endpoint is independent of its Diameter role. 207 For example, Diameter relay and proxy agents may act as DOIC 208 endpoints, even though they are not endpoints in the Diameter sense. 209 Since Diameter enables bi-directional applications, where Diameter 210 servers can send requests towards Diameter clients, a given Diameter 211 node can simultaneously act as a reporting node and a reacting node. 213 Likewise, a relay or proxy agent may act as a reacting node from the 214 perspective of upstream nodes, and a reporting node from the 215 perspective of downstream nodes. 217 DOIC endpoints do not generate new messages to carry DOIC related 218 information. Rather, they "piggyback" DOIC information over existing 219 Diameter messages by inserting new AVPs into existing Diameter 220 requests and responses. Nodes indicate support for DOIC, and any 221 needed DOIC parameters by inserting an OC_Supported_Features AVP 222 (Section 6.2) into existing requests and responses. Reporting nodes 223 send OLRs by inserting OC-OLR AVPs (Section 6.3). 225 A given OLR applies to the Diameter realm and application of the 226 Diameter message that carries it. If a reporting node supports more 227 than one realm and/or application, it reports independently for each 228 combination of realm and application. Similarly, OC-Feature-Vector 229 AVPs apply to the realm and application of the enclosing message. 230 This implies that a node may support DOIC for one application and/or 231 realm, but not another, and may indicate different DOIC parameters 232 for each application and realm for which it supports DOIC. 234 Reacting nodes perform overload abatement according to an agreed-upon 235 abatement algorithm. An abatement algorithm defines the meaning of 236 the parameters of an OLR, and the procedures required for overload 237 abatement. This document specifies a single must-support algorithm, 238 namely the "loss" algorithm Section 5). Future specifications may 239 introduce new algorithms. 241 Overload conditions may vary in scope. For example, a single 242 Diameter node may be overloaded, in which case reacting nodes may 243 reasonably attempt to send throttled requests to other destinations 244 or via other agents. On the other hand, an entire Diameter realm may 245 be overloaded, in which case such attempts would do harm. DOIC OLRs 246 have a concept of "report type" (Section 6.6), where the type defines 247 such behaviors. Report types are extensible. This document defines 248 report types for overload of a specific server, and for overload of 249 an entire realm. 251 While a reporting node sends OLRs to "adjacent" reacting nodes, nodes 252 that are "adjacent" for DOIC purposes may not be adjacent from a 253 Diameter, or transport, perspective. For example, one or more 254 Diameter agents that do not support DOIC may exist between a given 255 pair of reporting and reacting nodes, as long as those agents pass 256 unknown AVPs through unmolested. The report types described in this 257 document can safely pass through non-supporting agents. This may not 258 be true for report types defined in future specifications. Documents 259 that introduce new report types MUST describe any limitations on 260 their use across non-supporting agents. 262 3.1. Overload Control Endpoints (Non normative) 264 The overload control solution can be considered as an overlay on top 265 of an arbitrary Diameter network. The overload control information 266 is exchanged over on a "DOIC association" established between two 267 communication endpoints. The endpoints, namely the "reacting node" 268 and the "reporting node" do not need to be adjacent Diameter peer 269 nodes, nor they need to be the end-to-end Diameter nodes in a typical 270 "client-server" deployment with multiple intermediate Diameter agent 271 nodes in between. The overload control endpoints are the two 272 Diameter nodes that decide to exchange overload control information 273 between each other. How the endpoints are determined is specific to 274 a deployment, a Diameter node role in that deployment and local 275 configuration. 277 The following diagrams illustrate the concept of Diameter Overload 278 Endpoints and how they differ from the standard [RFC6733] defined 279 client, server and agent Diameter nodes. The following is the key to 280 the elements in the diagrams: 282 C Diameter client as defined in [RFC6733]. 284 S Diameter server as defined in [RFC6733]. 286 A Diameter agent, in either a relay or proxy mode, as defined in 287 [RFC6733]. 289 DEP Diameter Overload Endpoint as defined in this document. In the 290 following figures a DEP may terminate two different DOIC 291 associations being a reporter and reactor at the same time. 293 Diameter Session A Diameter session as defined in [RFC6733]. 295 DOIC Association A DOIC association exists between two Diameter 296 Overload Endpoints. One of the endpoints is the overload reporter 297 and the other is the overload reactor. 299 Figure 1 illustrates the most basic configuration where a client is 300 connected directly to a server. In this case, the Diameter session 301 and the DOIC association are both between the client and server. 303 +-----+ +-----+ 304 | C | | S | 305 +-----+ +-----+ 306 | DEP | | DEP | 307 +--+--+ +--+--+ 308 | | 309 | | 310 |{Diameter Session}| 311 | | 312 |{DOIC Association}| 313 | | 315 Figure 1: Basic DOIC deployment 317 In Figure 2 there is an agent that is not participating directly in 318 the exchange of overload reports. As a result, the Diameter session 319 and the DOIC association are still established between the client and 320 the server. 322 +-----+ +-----+ +-----+ 323 | C | | A | | S | 324 +-----+ +--+--+ +-----+ 325 | DEP | | | DEP | 326 +--+--+ | +--+--+ 327 | | | 328 | | | 329 |----------{Diameter Session}---------| 330 | | | 331 |----------{DOIC Association}---------| 332 | | | 334 Figure 2: DOIC deployment with non participating agent 336 Figure 3 illustrates the case where the client does not support 337 Diameter overload. In this case, the DOIC association is between the 338 agent and the server. The agent handles the role of the reactor for 339 overload reports generated by the server. 341 +-----+ +-----+ +-----+ 342 | C | | A | | S | 343 +--+--+ +-----+ +-----+ 344 | | DEP | | DEP | 345 | +--+--+ +--+--+ 346 | | | 347 | | | 348 |----------{Diameter Session}---------| 349 | | | 350 | |{DOIC Association}| 351 | | | 353 Figure 3: DOIC deployment with non-DOIC client and DOIC enabled agent 355 In Figure 4 there is a DOIC association between the client and the 356 agent and a second DOIC association between the agent and the server. 357 One use case requiring this configuration is when the agent is 358 serving as a SFE for a set of servers. 360 +-----+ +-----+ +-----+ 361 | C | | A | | S | 362 +-----+ +-----+ +-----+ 363 | DEP | | DEP | | DEP | 364 +--+--+ +--+--+ +--+--+ 365 | | | 366 | | | 367 |----------{Diameter Session}---------| 368 | | | 369 |{DOIC Association}|{DOIC Association}| 370 | | and/or 371 |----------{DOIC Association}---------| 372 | | | 374 Figure 4: A deployment where all nodes support DOIC 376 Figure 5 illustrates a deployment where some clients support Diameter 377 overload control and some do not. In this case the agent must 378 support Diameter overload control for the non supporting client. It 379 might also need to have a DOIC association with the server, as shown 380 here, to handle overload for a server farm and/or for managing Realm 381 overload. 383 +-----+ +-----+ +-----+ +-----+ 384 | C1 | | C2 | | A | | S | 385 +-----+ +--+--+ +-----+ +-----+ 386 | DEP | | | DEP | | DEP | 387 +--+--+ | +--+--+ +--+--+ 388 | | | | 389 | | | | 390 |-------------------{Diameter Session}-------------------| 391 | | | | 392 | |--------{Diameter Session}-----------| 393 | | | | 394 |---------{DOIC Association}----------|{DOIC Association}| 395 | | | and/or 396 |-------------------{DOIC Association}-------------------| 397 | | | | 399 Figure 5: A deployment with DOIC and non-DOIC supporting clients 401 Editor's note: Propose to remove C1, which is already shown in a 402 previous figure. Have this focus just on the non supporting client 403 scenario. 405 Figure 6 illustrates a deployment where some agents support Diameter 406 overload control and others do not. 408 +-----+ +-----+ +-----+ +-----+ 409 | C | | A | | A | | S | 410 +-----+ +--+--+ +-----+ +-----+ 411 | DEP | | | DEP | | DEP | 412 +--+--+ | +--+--+ +--+--+ 413 | | | | 414 | | | | 415 |-------------------{Diameter Session}-------------------| 416 | | | | 417 | | | | 418 |---------{DOIC Association}----------|{DOIC Association}| 419 | | | and/or 420 |-------------------{DOIC Association}-------------------| 421 | | | | 423 Figure 6: A deployment with DOIC and non-DOIC supporting agents 425 Editor's note: Propose to add a non supporting server scenario. 427 3.2. Piggybacking Principle (Non normative) 429 The overload control AVPs defined in this specification have been 430 designed to be piggybacked on top of existing application message 431 exchanges. This is made possible by adding overload control top 432 level AVPs, the OC-OLR AVP and the OC-Supported-Features AVP as 433 optional AVPs into existing commands when the corresponding Command 434 Code Format (CCF) specification allows adding new optional AVPs (see 435 Section 1.3.4 of [RFC6733]). 437 Reacting nodes indicate support for DOIC by including the OC- 438 Supported-Features AVP all request messages originated or relayed by 439 the Diameter node. 441 Reporting nodes indicate support for DOIC by including the OC- 442 Supported-Features AVP in all answer messages originated or relayed 443 by the Diameter node. Reporting nodes also include overload reports 444 using the OC-OLR AVP in answer messages. 446 Note: There is no new Diameter application defined to carry 447 overload related AVPs. The DOIC AVPs are carried in existing 448 Diameter application messages. 450 Note that the overload control solution does not have fixed server 451 and client roles. The endpoint role is determined based on the 452 message type: whether the message is a request (i.e. sent by a 453 "reacting node") or an answer (i.e. send by a "reporting node"). 454 Therefore, in a typical "client-server" deployment, the "client" MAY 455 report its overload condition to the "server" for any server 456 initiated message exchange. An example of such is the server 457 requesting a re-authentication from a client. 459 3.3. DOIC Capability Announcement (Non normative) 461 The DOIC solutions supports the ability for Diameter nodes to 462 determine if other nodes in the path of a request support the 463 solution. This capability is refered to as DOIC Capability 464 Announcement (DCA) and is separate from Diameter Capability Exchange. 466 The DCA mechanism is built around the piggybacking principle used for 467 transporting Diameter overload AVPs. This includes both DCA AVPs and 468 AVPs associated with Diameter overload reports. This allows for the 469 DCA AVPs to be carried across Diameter nodes that do not support the 470 DOIC solution. 472 The DCA mechanism uses the OC-Supported-Features AVPs to indicate the 473 Diameter overload features supported. 475 The first node in the path of a Diameter request that supports the 476 DOIC solution inserts the OC-Supported-Feature AVP in the request 477 message. This includes an indication that it supports the loss 478 overload abatement algorithm defined in this specification (see 479 Section 5). This insures that there is at least one commonly 480 supported overload abatement algorithm between the reporting node and 481 the reacting nodes in the path of the request. 483 DOIC must support deployments where Diameter Clients and/or 484 Diameter servers do not support the DOIC solution. In this 485 scenario, it is assumed that Diameter Agents that support the DOIC 486 solution will handle overload abatement for the non supporting 487 clients. In this case the DOIC agent will insert the OC- 488 Supporting-Features AVP in requests that do not already contain 489 one, telling the reporting node that there is a DOIC node that 490 will handle overload abatement. 492 The reporting node inserts the OC-Supported-Feature AVP in all answer 493 messages to requests that contained the OC-Supported-Feature AVP. 494 The contents of the reporting node's OC-Supported-Feature AVP 495 indicate the set of Diameter overload features supported by the 496 reporting node with one exception. 498 The reporting node only includes an indication of support for one 499 overload abatement algorithm. This is the algorithm that the 500 reporting node intends to use should it enter an overload condition. 501 Reacting nodes can use the indicated overload abatement algorithm to 502 prepare for possible overload reports. 504 Note that the loss algorithm defined in this document is a 505 stateless abatement algorithm. As a result it does not require 506 any actions by reacting nodes prior to the receipt of an overload 507 report. Stateful abatement algorithms that base the abatement 508 logic on a history of request messages sent might require reacting 509 nodes to maintain state to insure that overload reports can be 510 properly handled. 512 The individual features supported by the DOIC nodes are indicated in 513 the OC-Feature-Vector AVP. Any semantics associated with the 514 features will be defined in extension specifications that introduce 515 the features. 517 The DCA mechanism must also support the scenario where the set of 518 features supported by the sender of a request and by agents in the 519 path of a request differ. In this case, the agent updates the OC- 520 Supported-Feature AVP to reflect the mixture of the two sets of 521 supported features. 523 The logic to determine the content of the modified OC-Supported- 524 Feature AVP is out-of-scope for this specification and is left to 525 implementation decisions. Care must be taken in doing so not to 526 introduce interoperability issues for downstream or upstream DOIC 527 nodes. 529 3.4. DOIC Overload Condition Reporting (Non normative) 531 As with DOIC Capability Announcement, Overload Condition Reporting 532 uses new AVPs (Section 6.3) to indicate an overload condition. 534 The OC-OLR AVP is referred to as an overload report. The OC-OLR AVP 535 includes the type of report, an overload report ID, the length of 536 time that the report is valid and abatement algorithm specific AVPs. 538 Two types of overload reports are defined in this document, host 539 reports and realm reports. 541 Host reports apply to traffic that is sent to a specific Diameter 542 host. The applies to requests that contain the Destination-Host AVP 543 that contains a DiameterIdentity that matches that of the overload 544 report. These requests are referred to as host-routed requests. A 545 host report also applies to realm-routed requests, requests that do 546 not have a Destination-Host AVP, when the selected route for the 547 request is a connection to the impacted host. 549 Realm reports apply to realm-routed requests for a specific realm as 550 indicated in the Destination-Realm AVP. 552 Reporting nodes are responsible for determining the need for a 553 reduction of traffic. The method for making this determination is 554 implementation specific and depend on the type of overload report 555 being generated. A host report, for instance, will generally be 556 generated by tracking utilization of resources required by the host 557 to handle transactions for the the Diameter application. A realm 558 report will generally impact the traffic sent to multiple hosts and, 559 as such, will typically require tracking the capacity of the servers 560 able to handle realm-routed requests for the application. 562 Once a reporting node determines the need for a reduction in traffic, 563 it uses the DOIC defined AVPs to report on the condition. These AVPs 564 are included in answer messages sent or relayed by the reporting 565 node. The reporting node indicates the overload abatement algorithm 566 that is to be used to handle the traffic reduction in the OC- 567 Supported-Features AVP. The OC-OLR AVP is used to communicate 568 information about the requested reduction. 570 Reacting nodes, upon receipt of an overload report, are responsible 571 for applying the abatement algorithm to traffic impacted by the 572 overload report. The method used for that abatement is dependent on 573 the abatement algorithm. The loss abatement algorithm is defined in 574 this document (Section 5). Other abatement algorithms can be defined 575 in extensions to the DOIC solutions. 577 As the conditions that lead to the generation of the overload report 578 change the reporting node can send new overload reports requesting 579 greater reduction if the condition gets worse or less reduction if 580 the condition improves. The reporting node sends an overload report 581 with a duration of zero to indicate that the overlaod condition has 582 ended and use of the abatement algorithm is no longer needed. 584 The reacting node also determines when the overload report expires 585 based on the OC-Validaty-Duration AVP in the overload report and 586 stops applying the abatement algorithm when the report expires. 588 3.5. DOIC Extensibility (Non normative) 590 The DOIC solutions is designed to be extensible. This extensibility 591 is based on existing Diameter based extensibility mechanisms. 593 There are multiple categories of extensions that are expected. This 594 includes the definition of new overload abatement algorithms, the 595 definition of new report types and new definitions of the scope of 596 messages impacted by an overload report. 598 The DOIC solution uses the OC-Supported-Features AVP for DOIC nodes 599 to communicate supported features. The specific features supported 600 by the DOIC node are indicated in the OC-Feature-Vector AVP. DOIC 601 extensions must define new values for the OC-Feature-Vector AVP. 602 DOIC extensions also have the ability to add new AVPs to the OC- 603 Supported-Features AVP, if additional information about the new 604 feature is required to be communicate. 606 Overload abatement algorithms use the OC-OLR AVP to communicate 607 overload occurances. This AVP can also be extended to add new AVPs 608 allowing a reporting nodes to communicate additional information 609 about handling an overload condition. 611 If necessary, new extensions can also define new top level AVPs. It 612 is, however, recommended that DOIC extensions use the OC-Supported- 613 Features and OC-OLR to carry all DOIC related AVPs. 615 3.6. Simplified Example Architecture (Non normative) 617 Figure 7 illustrates the simplified architecture for Diameter 618 overload information conveyance. See Section 3.1 for more discussion 619 and details how different Diameter nodes fit into the architecture 620 from the DOIC point of view. 622 Realm X Same or other Realms 623 <--------------------------------------> <----------------------> 625 +--^-----+ : (optional) : 626 |Diameter| : : 627 |Server A|--+ .--. : +---^----+ : .--. 628 +--------+ | _( `. : |Diameter| : _( `. +---^----+ 629 +--( )--:-| Agent |-:--( )--|Diameter| 630 +--------+ | ( ` . ) ) : +-----^--+ : ( ` . ) ) | Client | 631 |Diameter|--+ `--(___.-' : : `--(___.-' +-----^--+ 632 |Server B| : : 633 +---^----+ : : 635 End-to-end Overload Indication 636 1) <-----------------------------------------------> 637 Diameter Application Y 639 Overload Indication A Overload Indication A' 640 2) <----------------------> <----------------------> 641 standard base protocol standard base protocol 643 Figure 7: Simplified architecture choices for overload indication 644 delivery 646 In Figure 7, the Diameter overload indication can be conveyed (1) 647 end-to-end between servers and clients or (2) between servers and 648 Diameter agent inside the realm and then between the Diameter agent 649 and the clients when the Diameter agent acting as back-to-back-agent 650 for DOIC purposes. 652 3.7. Considerations for Applications Integrating the DOIC Solution (Non 653 normative) 655 THis section outlines considerations to be taken into account when 656 integrating the DOIC solution into Diameter applications. 658 3.7.1. Application Classification (Non normative) 660 The following is a classification of Diameter applications and 661 requests. This discussion is meant to document factors that play 662 into decisions made by the Diameter identity responsible for handling 663 overload reports. 665 Section 8.1 of [RFC6733] defines two state machines that imply two 666 types of applications, session-less and session-based applications. 668 The primary difference between these types of applications is the 669 lifetime of Session-Ids. 671 For session-based applications, the Session-Id is used to tie 672 multiple requests into a single session. 674 In session-less applications, the lifetime of the Session-Id is a 675 single Diameter transaction, i.e. the session is implicitly 676 terminated after a single Diameter transaction and a new Session-Id 677 is generated for each Diameter request. 679 For the purposes of this discussion, session-less applications are 680 further divided into two types of applications: 682 Stateless applications: 684 Requests within a stateless application have no relationship to 685 each other. The 3GPP defined S13 application is an example of a 686 stateless application [S13], --> where only a Diameter command is 687 defined between a client and a server and no state is maintained 688 between two consecutive transactions. 690 Pseudo-session applications: 692 Applications that do not rely on the Session-Id AVP for 693 correlation of application messages related to the same session 694 but use other session-related information in the Diameter requests 695 for this purpose. The 3GPP defined Cx application [Cx] is an 696 example of a pseudo-session application. 698 The Credit-Control application defined in [RFC4006] is an example of 699 a Diameter session-based application. 701 The handling of overload reports must take the type of application 702 into consideration, as discussed in Section 3.7.2. 704 3.7.2. Application Type Overload Implications (Non normative) 706 This section discusses considerations for mitigating overload 707 reported by a Diameter entity. This discussion focuses on the type 708 of application. Section 3.7.3 discusses considerations for handling 709 various request types when the target server is known to be in an 710 overloaded state. 712 These discussions assume that the strategy for mitigating the 713 reported overload is to reduce the overall workload sent to the 714 overloaded entity. The concept of applying overload treatment to 715 requests targeted for an overloaded Diameter entity is inherent to 716 this discussion. The method used to reduce offered load is not 717 specified here but could include routing requests to another Diameter 718 entity known to be able to handle them, or it could mean rejecting 719 certain requests. For a Diameter agent, rejecting requests will 720 usually mean generating appropriate Diameter error responses. For a 721 Diameter client, rejecting requests will depend upon the application. 722 For example, it could mean giving an indication to the entity 723 requesting the Diameter service that the network is busy and to try 724 again later. 726 Stateless applications: 728 By definition there is no relationship between individual requests 729 in a stateless application. As a result, when a request is sent 730 or relayed to an overloaded Diameter entity - either a Diameter 731 Server or a Diameter Agent - the sending or relaying entity can 732 choose to apply the overload treatment to any request targeted for 733 the overloaded entity. 735 Pseudo-session applications: 737 For pseudo-session applications, there is an implied ordering of 738 requests. As a result, decisions about which requests towards an 739 overloaded entity to reject could take the command code of the 740 request into consideration. This generally means that 741 transactions later in the sequence of transactions should be given 742 more favorable treatment than messages earlier in the sequence. 743 This is because more work has already been done by the Diameter 744 network for those transactions that occur later in the sequence. 745 Rejecting them could result in increasing the load on the network 746 as the transactions earlier in the sequence might also need to be 747 repeated. 749 Session-based applications: 751 Overload handling for session-based applications must take into 752 consideration the work load associated with setting up and 753 maintaining a session. As such, the entity sending requests 754 towards an overloaded Diameter entity for a session-based 755 application might tend to reject new session requests prior to 756 rejecting intra-session requests. In addition, session ending 757 requests might be given a lower probability of being rejected as 758 rejecting session ending requests could result in session status 759 being out of sync between the Diameter clients and servers. 760 Application designers that would decide to reject mid-session 761 requests will need to consider whether the rejection invalidates 762 the session and any resulting session clean-up procedures. 764 3.7.3. Request Transaction Classification (Non normative) 766 Independent Request: 768 An independent request is not correlated to any other requests 769 and, as such, the lifetime of the session-id is constrained to an 770 individual transaction. 772 Session-Initiating Request: 774 A session-initiating request is the initial message that 775 establishes a Diameter session. The ACR message defined in 776 [RFC6733] is an example of a session-initiating request. 778 Correlated Session-Initiating Request: 780 There are cases when multiple session-initiated requests must be 781 correlated and managed by the same Diameter server. It is notably 782 the case in the 3GPP PCC architecture [PCC], where multiple 783 apparently independent Diameter application sessions are actually 784 correlated and must be handled by the same Diameter server. 786 Intra-Session Request: 788 An intra session request is a request that uses the same Session- 789 Id than the one used in a previous request. An intra session 790 request generally needs to be delivered to the server that handled 791 the session creating request for the session. The STR message 792 defined in [RFC6733] is an example of an intra-session requests. 794 Pseudo-Session Requests: 796 Pseudo-session requests are independent requests and do not use 797 the same Session-Id but are correlated by other session-related 798 information contained in the request. There exists Diameter 799 applications that define an expected ordering of transactions. 800 This sequencing of independent transactions results in a pseudo 801 session. The AIR, MAR and SAR requests in the 3GPP defined Cx 802 [Cx] application are examples of pseudo-session requests. 804 3.7.4. Request Type Overload Implications (Non normative) 806 The request classes identified in Section 3.7.3 have implications on 807 decisions about which requests should be throttled first. The 808 following list of request treatment regarding throttling is provided 809 as guidelines for application designers when implementing the 810 Diameter overload control mechanism described in this document. The 811 exact behavior regarding throttling is a matter of local policy, 812 unless specifically defined for the application. 814 Independent requests: 816 Independent requests can be given equal treatment when making 817 throttling decisions. 819 Session-initiating requests: 821 Session-initiating requests represent more work than independent 822 or intra-session requests. Moreover, session-initiating requests 823 are typically followed by other session-related requests. As 824 such, as the main objective of the overload control is to reduce 825 the total number of requests sent to the overloaded entity, 826 throttling decisions might favor allowing intra-session requests 827 over session-initiating requests. Individual session-initiating 828 requests can be given equal treatment when making throttling 829 decisions. 831 Correlated session-initiating requests: 833 A Request that results in a new binding, where the binding is used 834 for routing of subsequent session-initiating requests to the same 835 server, represents more work load than other requests. As such, 836 these requests might be throttled more frequently than other 837 request types. 839 Pseudo-session requests: 841 Throttling decisions for pseudo-session requests can take into 842 consideration where individual requests fit into the overall 843 sequence of requests within the pseudo session. Requests that are 844 earlier in the sequence might be throttled more aggressively than 845 requests that occur later in the sequence. 847 Intra-session requests 849 There are two classes of intra-sessions requests. The first class 850 consists of requests that terminate a session. The second one 851 contains the set of requests that are used by the Diameter client 852 and server to maintain the ongoing session state. Session 853 terminating requests should be throttled less aggressively in 854 order to gracefully terminate sessions, allow clean-up of the 855 related resources (e.g. session state) and get rid of the need for 856 other intra-session requests, reducing the session management 857 impact on the overloaded entity. The default handling of other 858 intra-session requests might be to treat them equally when making 859 throttling decisions. There might also be application level 860 considerations whether some request types are favored over others. 862 4. Solution Procedures (Normative) 864 This section outlines the normative behavior associated with the DOIC 865 solution. 867 4.1. Capability Announcement (Normative) 869 This section defines DOIC Capability Announcement (DCA) behavior. 871 The DCA procedures are used to indicate support for DOIC and support 872 for DOIC features. The DOIC features include overload abatement 873 algorithms supported. It might also include new report types or 874 other extensions documented in the future. 876 Diameter nodes indicate support for DOIC by including the OC- 877 Supported-Features AVP in messages sent or handled by the node. 879 Diameter agents that support DOIC MUST ensure that all messages have 880 the OC-Supporting-Features AVP. If a message handled by the DOIC 881 agent does not include the OC-Supported-Features AVP then the DOIC 882 agent inserts the AVP. If the message already has the AVP then the 883 agent either leaves it unchanged in the relayed message or modifies 884 it to reflect a mixed set of DOIC features. 886 4.1.1. Reacting Node Behavior (Normative) 888 A reacting node MUST include the OC-Supported-Features AVP in all 889 request messages. 891 A reacting node MUST include the OC-Feature-Vector AVP with an 892 indication of the loss algorithm. 894 A reacting node SHOULD indicate support for all other DOIC features 895 it supports. 897 An OC-Supported-Features AVP in answer messages indicates there is a 898 reporting node for the transaction. The reacting node MAY take 899 action based on the features indicated in the OC-Feature-Vector AVP. 901 Note that the loss abatement algorithm is the only feature 902 described in this document and it does not require action to be 903 taken by the reacting node except when the answer message also has 904 an overload report. This behavior is described in Section 4.2 and 905 Section 5. 907 4.1.2. Reporting Node Behavior (Normative) 909 Upon receipt of a request message, a reporting node determines if 910 there is a reacting node for the transaction based on the presence of 911 the OC-Supported-Features AVP. 913 Based on the content of the OC-Supported-Features AVP in the request 914 message, the reporting node knows what overload control functionality 915 supported by reacting node(s). The reporting node then acts 916 accordingly for the subsequent answer messages it initiates. 918 If the reqeust message contains an OC-Supported-Features AVP then the 919 reporting node MUST include the OC-Supported-Features AVP in the 920 answer message for that transaction. 922 The reporting node MUST indicate support for one and only one 923 abatement algorithm in the OC-Feature-Vector AVP. The abatement 924 algorithm included MUST be from the set of abatement algorithms 925 contained in the request messages OC-Supported-Features AVP. The 926 abatement algorithm included indicates the abatement algorithm the 927 reporting node wants the reacting node to use when the reporting node 928 enters an overload condition. 930 The reporting node MUST NOT change the selected algorithm during a 931 period of time that it is in an overload condition and, as a result, 932 is sending OC-OLR AVPs in answer messages. 934 The reporting node SHOULD indicate support for other DOIC features it 935 supports and that apply to the transaction. 937 Note that not all DOIC features will apply to all Diameter 938 applications or deployment scenarios. The features included in 939 the OC-Feature-Vector AVP is based on local reporting node policy. 941 The reporting node MUST NOT include the OC-Supported-Features AVP, 942 OC-OLR AVP or any other overload control AVPs defined in extension 943 drafts in response messages for transactions where the request 944 message does not include the OC-Supported-Features AVP. Lack of the 945 OC-Supported-Features AVP in the request message indicates that there 946 is no reacting node for the transaction. 948 An agent MAY modify the OC-Supported-Features AVP carried in answer 949 messages. 951 4.1.3. Agent Behavior (Normative) 953 Editor's note -- Need to add this section. 955 4.2. Overload Report Processing (Normative) 957 4.2.1. Overload Control State (Normative) 959 Both reacting and reporting nodes maintain an overload control state 960 (OCS) for each endpoint (a host or a realm) they communicate with and 961 both endpoints have announced support for DOIC. See Sections 6.1 and 962 4.1 for discussion about how the support for DOIC is determined. 964 4.2.1.1. Overload Control State for Reacting Nodes 966 A reacting node maintains the following OCS per supported Diameter 967 application: 969 o A host-type Overload Control State for each Destination-Host 970 towards which it sends host-type requests and 972 o A realm-type Overload Control State for each Destination-Realm 973 towards which it sends realm-type requests. 975 A host-type Overload Control State may be identified by the pair of 976 Application-Id and Destination-Host. A realm-type Overload Control 977 State may be identified by the pair of Application-Id and 978 Destination-Realm. The host-type/realm-type Overload Control State 979 for a given pair of Application and Destination-Host / Destination- 980 Realm could include the following information: 982 o Sequence number (as received in OC-OLR) 984 o Time of expiry (deviated from validity duration as received in OC- 985 OLR and time of reception) 987 o Selected Abatement Algorithm (as received in OC-Supported- 988 Features) 990 o Algorithm specific input data (as received within OC-OLR, e.g. 991 Reduction Percentage for Loss) 993 4.2.1.2. Overload Control States for Reporting Nodes 995 A reporting node maintains per supported Diameter application and per 996 supported (and eventually selected) Abatement Algorithm an Overload 997 Control State. 999 An Overload Control State may be identified by the pair of 1000 Application-Id and supported Abatement Algorithm. 1002 The Overload Control State for a given pair of Application and 1003 Abatement Algorithm could include the information: 1005 o Sequence number 1007 o Validity Duration and Expiry Time 1009 o Algorithm specific input data (e.g. Reduction Percentage for 1010 Loss) 1012 Overload Control States for reporting nodes containing a validity 1013 duration of 0 sec. should not expire before any previously sent 1014 (stale) OLR has timed out at any reacting node. 1016 Editor's note: This statement is unclear and contradictory with other 1017 statements. A validity timer of zero seconds indicates that the 1018 overload condition has ended and abatement is no longer requested. 1020 4.2.1.3. Maintaining Overload Control State 1022 Reacting nodes create a host-type OCS identified by OCS-Id = (app- 1023 id,host-id) when receiving an answer message of application app-id 1024 containing an Orig-Host of host-id and a host-type OC-OLR AVP unless 1025 such host-type OCS already exists. 1027 Reacting nodes create a realm-type OCS identified by OCS-Id = (app- 1028 id,realm-id) when receiving an answer message of application app-id 1029 containing an Orig-Realm of realm-id and a realm-type OC-OLR AVP 1030 unless such realm type OCS already exists. 1032 Reacting nodes delete an OCS when it expires (i.e. when current time 1033 minus reception time is greater than validity duration). 1035 Editor's note: Reacting nodes also delete on OCS with an updated OLR 1036 is received with a validity duration of zero. 1038 Reacting nodes update the host-type OCS identified by OCS-Id = (app- 1039 id,host-id) when receiving an answer message of application app-id 1040 containing an Orig-Host of host-id and a host-type OC-OLR AVP with a 1041 sequence number higher than the stored sequence number. 1043 Reacting nodes update the realm-type OCS identified by OCS-Id = (app- 1044 id,realm-id) when receiving an answer message of application app-id 1045 containing an Orig-Realm of realm-id and a realm-type OC-OLR AVP with 1046 a sequence number higher than the stored sequence number. 1048 Reacting nodes do not delete an OCS when receiving an answer message 1049 that does not contain an OC-OLR AVP (i.e. absence of OLR means "no 1050 change"). 1052 Reporting nodes create an OCS identified by OCS-Id = (app-id,Alg) 1053 when receiving a request of application app-id containing an OC- 1054 Supported-Features AVP indicating support of the Abatement Algorithm 1055 Alg (which the reporting node selects) while being overloaded, unless 1056 such OCS already exists. 1058 Reporting nodes delete an OCS when it expires. 1060 Editor's note: Reporting nodes should send updated overload reports 1061 with a validity duration of zero for a period of time after an OCS 1062 expires or is removed due to the overload condition ending. 1064 Reporting nodes update the OCS identified by OCS-Id = (app-id,Alg) 1065 when they detect the need to modify the requested amount of 1066 application app-id traffic reduction. 1068 4.2.2. Reacting Node Behavior (Normative) 1070 Once a reacting node receives an OC-OLR AVP from a reporting node, it 1071 applies traffic abatement based on the selected algorithm with the 1072 reporting node and the current overload condition. The reacting node 1073 learns the reporting node supported abatement algorithms directly 1074 from the received answer message containing the OC-Supported-Features 1075 AVP. 1077 The received OC-Supported-Features AVP does not change the existing 1078 overload condition and/or traffic abatement algorithm settings if the 1079 OC-Sequence-Number AVP contains a value that is equal to the 1080 previously received/recorded value. If the OC-Supported-Features AVP 1081 is received for the first time for the reporting node or the OC- 1082 Sequence-Number AVP value is less than the previously received/ 1083 recorded value (and is outside the valid overflow window), then the 1084 sequence number is stale (e.g. an intentional or unintentional 1085 replay) and SHOULD be silently discarded. 1087 As described in Section 6.3, the OC-OLR AVP contains the necessary 1088 information for the overload condition on the reporting node. 1090 From the OC-Report-Type AVP contained in the OC-OLR AVP, the reacting 1091 node learns whether the overload condition report concerns a specific 1092 host (as identified by the Origin-Host AVP of the answer message 1093 containing the OC-OLR AVP) or the entire realm (as identified by the 1094 Origin-Realm AVP of the answer message containing the OC-OLR AVP). 1095 The reacting node learns the Diameter application to which the 1096 overload report applies from the Application-ID of the answer message 1097 containing the OC-OLR AVP. The reacting node MUST use this 1098 information as an input for its traffic abatement algorithm. The 1099 idea is that the reacting node applies different handling of the 1100 traffic abatement, whether sent request messages are targeted to a 1101 specific host (identified by the Diameter-Host AVP in the request) or 1102 to any host in a realm (when only the Destination-Realm AVP is 1103 present in the request). Note that future specifications MAY define 1104 new OC-Report-Type AVP values that imply different handling of the 1105 OC-OLR AVP. For example, in a form of new additional AVPs inside the 1106 Grouped OC-OLR AVP that would define report target in a finer 1107 granularity than just a host. 1109 Editor's note: The above behavior for Realm reports is 1110 inconsistent with the definition of realm reports in section 1111 Section 6.6. 1113 If the OC-OLR AVP is received for the first time, the reacting node 1114 MUST create overload control state associated with the related realm 1115 or a specific host in the realm identified in the message carrying 1116 the OC-OLR AVP, as described in Section 4.2.1. 1118 If the value of the OC-Sequence-Number AVP contained in the received 1119 OC-OLR AVP is equal to or less than the value stored in an existing 1120 overload control state, the received OC-OLR AVP SHOULD be silently 1121 discarded. If the value of the OC-Sequence-Number AVP contained in 1122 the received OC-OLR AVP is greater than the value stored in an 1123 existing overload control state or there is no previously recorded 1124 sequence number, the reacting node MUST update the overload control 1125 state associated with the realm or the specific node in the realm. 1127 When an overload control state is created or updated, the reacting 1128 node MUST apply the traffic abatement requested in the OC-OLR AVP 1129 using the algorithm announced in the OC-Supported-Features AVP 1130 contained in the received answer message along with the OC-OLR AVP. 1132 The validity duration of the overload information contained in the 1133 OC-OLR AVP is either explicitly indicated in the OC-Validity-Duration 1134 AVP or is implicitly equals to the default value (5 seconds) if the 1135 OC-Validity-Duration AVP is absent. The reacting node MUST maintain 1136 the validity duration in the overload control state. Once the 1137 validity duration times out, the reacting node MUST assume the 1138 overload condition reported in a previous OC-OLR AVP has ended. 1140 A value of zero ("0") received in the OC-Validity-Duration in an 1141 updated overload report indicates that the overload condition has 1142 ended and that the overload state is no longer valid. 1144 In the case that the validity duration expires or is explicitly 1145 signaled as being no longer valid the state associated with the 1146 overload report MUST be removed and any abatement associated with the 1147 overload report MUST be ended in a controlled fashion. After 1148 removing the overload state the sequence number MUST NOT be used for 1149 future comparisons of sequence numbers. 1151 4.2.3. Reporting Node Behavior (Normative) 1153 A reporting node is a Diameter node inserting an OC-OLR AVP in a 1154 Diameter message in order to inform a reacting node about an overload 1155 condition and request Diameter traffic abatement. 1157 The operation on the reporting node is straight forward. The 1158 reporting node learns the capabilities of the reacting node when it 1159 receives the OC-Supported-Features AVP as part of any Diameter 1160 request message. If the reporting node shares at least one common 1161 feature with the reacting node, then the DOIC can be enabled between 1162 these two endpoints. See Section 4.1 for further discussion on the 1163 capability and feature announcement between two endpoints. 1165 When a traffic reduction is required due to an overload condition and 1166 the overload control solution is supported by the sender of the 1167 Diameter request, the reporting node MUST include an OC-Supported- 1168 Features AVP and an OC-OLR AVP in the corresponding Diameter answer. 1169 The OC-OLR AVP contains the required traffic reduction and the OC- 1170 Supported-Features AVP indicates the traffic abatement algorithm to 1171 apply. This algorithm MUST be one of the algorithms advertised by 1172 the request sender. 1174 A reporting node MAY rely on the OC-Validity-Duration AVP values for 1175 the implicit overload control state cleanup on the reacting node. 1176 However, it is RECOMMENDED that the reporting node always explicitly 1177 indicates the end of a overload condition. 1179 The reporting node SHOULD indicate the end of an overload occurrence 1180 by sending a new OLR with OC-Validity-Duration set to a value of zero 1181 ("0"). The reporting node SHOULD insure that all reacting nodes 1182 receive the updated overload report. 1184 4.2.4. Agent Behavior (Normative) 1186 Editor's note -- Need to add this section. 1188 4.3. Protocol Extensibility (Normative) 1190 The overload control solution can be extended, e.g. with new traffic 1191 abatement algorithms, new report types or other new functionality. 1193 When defining a new extension a new feature bit MUST be defined for 1194 the OC-Feature-Vector. This feature bit is used to communicate 1195 support for the new feature. 1197 The extention may also define new AVPs for use in DOIC Capability 1198 Anouncement and for use in DOIC Overload reporting. These new AVP 1199 should be defined to be extensions to the OC-Supported-Features and 1200 OC-OLR AVPs defined in this document. 1202 It should be noted that [RFC6733] defined Grouped AVP extension 1203 mechanisms apply. This allows, for example, defining a new feature 1204 that is mandatory to be understood even when piggybacked on an 1205 existing applications. More specifically, the sub-AVPs inside the 1206 OC-Supported-Features and OC-OLR AVP MAY have the M-bit set. 1207 However, when overload control AVPs are piggybacked on top of an 1208 existing applications, setting M-bit in sub-AVPs is NOT RECOMMENDED. 1210 The handling of feature bits in the OC-Feature-Vector AVP that are 1211 not associated with overload abatement algorithms MUST be specified 1212 by the extensions that define the features. 1214 When defining new report type values, the corresponding specification 1215 MUST define the semantics of the new report types and how they affect 1216 the OC-OLR AVP handling. The specification MUST also reserve a 1217 corresponding new feature, see the OC-Supported-Features and OC- 1218 Feature-Vector AVPs. 1220 The OC-OLR AVP can be expanded with optional sub-AVPs only if a 1221 legacy implementation can safely ignore them without breaking 1222 backward compatibility for the given OC-Report-Type AVP value implied 1223 report handling semantics. If the new sub-AVPs imply new semantics 1224 for handling the indicated report type, then a new OC-Report-Type AVP 1225 value MUST be defined. 1227 New features (feature bits in the OC-Feature-Vector AVP) and report 1228 types (in the OC-Report-Type AVP) MUST be registered with IANA. As 1229 with any Diameter specification, new AVPs MUST also be registered 1230 with IANA. See Section 8 for the required procedures. 1232 5. Loss Algorithm (Normative) 1234 This section documents the Diameter overload loss abatement 1235 algorithm. 1237 5.1. Overview (Non normative) 1239 The DOIC specification supports the ability for multiple overload 1240 abatement algorithms to be specified. The abatement algorithm used 1241 for any instance of overload is determined by the Diameter Overload 1242 Capability Announcement process documented in Section 4.1. 1244 The loss algorithm described in this section is the default algorithm 1245 that must be supported by all Diameter nodes that support DOIC. 1247 The loss algorithm is designed to be a straightforward and stateless 1248 overload abatement algorithm. It is used by reporting nodes to 1249 request a percentage reduction in the amount of traffic sent. The 1250 traffic impacted by the requested reduction depends on the type of 1251 overload report. 1253 Reporting nodes use a strategy of applying abatement logic to the 1254 requested percentage of request messages sent (or handled in the case 1255 of agents) by the reacting node that are impacted by the overload 1256 report. 1258 From a conceptual level, the logic at the reacting node could be 1259 outlined as follows. In this discussion assume that the reacting 1260 node is also the sending node. 1262 1. An overload report is received and the associated overload state 1263 is saved by the reacting node. 1265 2. A new Diameter request is generated by the application running on 1266 the reacting node. 1268 3. The reacting node determines that an active overload report 1269 applies to the request. 1271 4. The reacting node determines if abatement should be applied to 1272 the request. One approach that could be taken would be to select 1273 a random number between 1 and 100. If the random number is less 1274 than the indicated reduction percentage then the request is given 1275 abatement treatment, otherwise the request is given normal 1276 routing treatment. 1278 5.2. Use of OC-Reduction-Percentage AVP 1280 A reporting node using the loss algorithm must use the OC-Reduction- 1281 Percentage AVP (Section 6.7 to indicated the desired percentage of 1282 traffic reduction.) 1284 Editor's note: The above duplicates what is in the OC-Reduction- 1285 Percentage AVP section can probably be removed. 1287 5.3. Reporting Node Behavior (Normative) 1289 The method a reporting nodes uses to determine the amount of traffic 1290 reduction required to address an overload condition is an 1291 implementation decision. 1293 When a reporting node that has selected the loss abatement algorithm 1294 determines the need to request a traffic reduction it must include an 1295 OC-OLR AVP in all response messages. 1297 The reporting node must indicate a percentage reduction in the OC- 1298 Reduction-Percentage AVP. 1300 The reporting node may change the reduction percentage in subsequent 1301 overload reports. When doing so the reporting node must conform to 1302 overload report handing specified in Section 4.2.3. 1304 When the reporting node determines it no longer needs a reduction in 1305 traffic the reporting node should send an overload report indicating 1306 the overload report is no longer valid, as specified in 1307 Section 4.2.3. 1309 5.4. Reacting Node Behavior (Normative) 1311 The method a reacting node uses to determine which request messages 1312 are given abatement treatment is an implementation decision. 1314 When receiving an OC-OLR in an answer message where the algorithm 1315 indicated in the OC-Supported-Features AVP is the loss algorithm, the 1316 reacting node must attempt to apply abatement treatment to the 1317 requested percentage of request messages sent. 1319 Note: the loss algorithm is a stateless algorithm. As a result, 1320 the reacting node does not guarantee that there will be an 1321 absolute reduction in traffic sent. Rather, it guarantees that 1322 the requested percentage of new requests will be given abatement 1323 treatment. 1325 If reacting node comes out of the 100 percent traffic reduction as a 1326 result of the overload report timing out, the following concerns are 1327 RECOMMENDED to be applied. The reacting node sending the traffic 1328 should be conservative and, for example, first send "probe" messages 1329 to learn the overload condition of the overloaded node before 1330 converging to any traffic amount/rate decided by the sender. Similar 1331 concerns apply in all cases when the overload report times out unless 1332 the previous overload report stated 0 percent reduction. 1334 Editor's note: Need to add additional guidance to slowly increase 1335 the rate of traffic sent to avoid a sudden spike in traffic, as 1336 the spike in traffic could result in oscillation of the need for 1337 overload control. 1339 If the reacting node does not receive a an OLR in messages sent to 1340 the formally overloaded node then the reacting node should slowly 1341 increase the rate of traffic sent to the overloaded node. 1343 It is suggested that the reacting node decrease the amount of traffic 1344 given abatement treatment by 20% each second until the reduction is 1345 completely removed and no traffic is given abatement treatment. 1347 The goal of this behavior is to reduce the probability of overload 1348 condition thrashing where an immediate transition from 100% 1349 reduction to 0% reduction results in the reporting node moving 1350 quickly back into an overload condition. 1352 6. Attribute Value Pairs (Normative) 1354 This section describes the encoding and semantics of the Diameter 1355 Overload Indication Attribute Value Pairs (AVPs) defined in this 1356 document. 1358 When added to existing commands, both OC-Feature-Vector and OC-OLR 1359 AVPs SHOULD have the M-bit flag cleared to avoid backward 1360 compatibility issues. 1362 A new application specification can incorporate the overload control 1363 mechanism specified in this document by making it mandatory to 1364 implement for the application and referencing this specification 1365 normatively. In such a case, the OC-Feature-Vector and OC-OLR AVPs 1366 reused in newly defined Diameter applications SHOULD have the M-bit 1367 flag set. However, it is the responsibility of the Diameter 1368 application designers to define how overload control mechanisms works 1369 on that application. 1371 6.1. OC-Supported-Features AVP 1373 The OC-Supported-Features AVP (AVP code TBD1) is type of Grouped and 1374 serves for two purposes. First, it announces a node's support for 1375 the DOIC in general. Second, it contains the description of the 1376 supported DOIC features of the sending node. The OC-Supported- 1377 Features AVP MUST be included in every Diameter message a DOIC 1378 supporting node sends. 1380 OC-Supported-Features ::= < AVP Header: TBD1 > 1381 [ OC-Feature-Vector ] 1382 * [ AVP ] 1384 The OC-Feature-Vector sub-AVP is used to announce the DOIC features 1385 supported by the endpoint, in the form of a flag bits field in which 1386 each bit announces one feature or capability supported by the node 1387 (see Section 6.2). The absence of the OC-Feature-Vector AVP 1388 indicates that only the default traffic abatement algorithm described 1389 in this specification is supported. 1391 A reacting node includes this AVP to indicate its capabilities to a 1392 reporting node. For example, the endpoint (reacting node) may 1393 indicate which (future defined) traffic abatement algorithms it 1394 supports in addition to the default. 1396 During the message exchange the overload control endpoints express 1397 their common set of supported capabilities. The reacting node 1398 includes the OC-Supported-Features AVP that announces what it 1399 supports. The reporting node that sends the answer also includes the 1400 OC-Supported-Features AVP that describes the capabilities it 1401 supports. The set of capabilities advertised by the reporting node 1402 depends on local policies. At least one of the announced 1403 capabilities MUST match. If there is no single matching capability 1404 the reacting node MUST act as if it does not implement DOIC and cease 1405 inserting any DOIC related AVPs into any Diameter messages with this 1406 specific reacting node. 1408 Editor's note: The last sentence conflicts with the last sentence 1409 two paragraphs up. In reality, there will always be at least one 1410 matching capability as all nodes supporting DOIC must support the 1411 loss algorithm. Suggest removing the last sentence. 1413 6.2. OC-Feature-Vector AVP 1415 The OC-Feature-Vector AVP (AVP code TBD6) is type of Unsigned64 and 1416 contains a 64 bit flags field of announced capabilities of an 1417 overload control endpoint. The value of zero (0) is reserved. 1419 The following capabilities are defined in this document: 1421 OLR_DEFAULT_ALGO (0x0000000000000001) 1423 When this flag is set by the overload control endpoint it means 1424 that the default traffic abatement (loss) algorithm is supported. 1426 6.3. OC-OLR AVP 1428 The OC-OLR AVP (AVP code TBD2) is type of Grouped and contains the 1429 necessary information to convey an overload report. The OC-OLR AVP 1430 does not explicitly contain all information needed by the reacting 1431 node to decide whether a subsequent request must undergo a throttling 1432 process with the received reduction percentage. The value of the OC- 1433 Report-Type AVP within the OC-OLR AVP indicates which implicit 1434 information is relevant for this decision (see Section 6.6). The 1435 application the OC-OLR AVP applies to is the same as the Application- 1436 Id found in the Diameter message header. The identity the OC-OLR AVP 1437 concerns is determined from the Origin-Host AVP (and Origin-Realm AVP 1438 as well) found from the encapsulating Diameter command. The OC-OLR 1439 AVP is intended to be sent only by a reporting node. 1441 OC-OLR ::= < AVP Header: TBD2 > 1442 < OC-Sequence-Number > 1443 < OC-Report-Type > 1444 [ OC-Reduction-Percentage ] 1445 [ OC-Validity-Duration ] 1446 * [ AVP ] 1448 The OC-Validity-Duration AVP indicates the validity time of the 1449 overload report associated with a specific sequence number, measured 1450 after reception of the OC-OLR AVP. The validity time MUST NOT be 1451 updated after reception of subsequent OC-OLR AVPs with the same 1452 sequence number. The default value for the OC-Validity-Duration AVP 1453 value is 5 (i.e., 5 seconds). When the OC-Validity-Duration AVP is 1454 not present in the OC-OLR AVP, the default value applies. 1456 Note that if a Diameter command were to contain multiple OC-OLR AVPs 1457 they all MUST have different OC-Report-Type AVP value. OC-OLR AVPs 1458 with unknown values SHOULD be silently discarded and the event SHOULD 1459 be logged. 1461 Editor's note: Need to specify what happens when two reports of 1462 the same type are received. 1464 6.4. OC-Sequence-Number AVP 1466 The OC-Sequence-Number AVP (AVP code TBD3) is type of Unsigned64. 1467 Its usage in the context of overload control is described in 1468 Section 4.2. 1470 From the functionality point of view, the OC-Sequence-Number AVP MUST 1471 be used as a non-volatile increasing counter between two overload 1472 control endpoints. The sequence number is only required to be unique 1473 between two overload control endpoints. Sequence numbers are treated 1474 in a uni-directional manner, i.e. two sequence numbers on each 1475 direction between two endpoints are not related or correlated. 1477 When generating sequence numbers, the new sequence number MUST be 1478 greater than any sequence number in an active overload report 1479 previously sent by the reporting node. This property MUST hold over 1480 a reboot of the reporting node. 1482 6.5. OC-Validity-Duration AVP 1484 The OC-Validity-Duration AVP (AVP code TBD4) is type of Unsigned32 1485 and indicates in seconds the validity time of the overload report. 1486 The number of seconds is measured after reception of the first OC-OLR 1487 AVP with a given value of OC-Sequence-Number AVP. The default value 1488 for the OC-Validity-Duration AVP is 5 (i.e., 5 seconds). When the 1489 OC-Validity-Duration AVP is not present in the OC-OLR AVP, the 1490 default value applies. Validity duration with values above 86400 1491 (i.e.; 24 hours) MUST NOT be used. Invalid duration values are 1492 treated as if the OC-Validity-Duration AVP were not present and 1493 result in the default value being used. 1495 A timeout of the overload report has specific concerns that need to 1496 be taken into account by the endpoint acting on the earlier received 1497 overload report(s). Section 6.7 discusses the impacts of timeout in 1498 the scope of the traffic abatement algorithms. 1500 When a reporting node has recovered from overload, it SHOULD 1501 invalidate any existing overload reports in a timely matter. This 1502 can be achieved by sending an updated overload report (meaning the 1503 OLR contains a new sequence number) with the OC-Validity-Duration AVP 1504 value set to zero ("0"). If the overload report is about to expire 1505 naturally, the reporting node MAY choose to simply let it do so. 1507 A reacting node MUST invalidate and remove an overload report that 1508 expires without an explicit overload report containing an OC- 1509 Validity-Duration value set to zero ("0"). 1511 6.6. OC-Report-Type AVP 1513 The OC-Report-Type AVP (AVP code TBD5) is type of Enumerated. The 1514 value of the AVP describes what the overload report concerns. The 1515 following values are initially defined: 1517 0 A host report. The overload treatment should apply to requests 1518 for which all of the following conditions are true: 1520 Either the Destination-Host AVP is present in the request and its 1521 value matches the value of the Origin-Host AVP of the received 1522 message that contained the OC-OLR AVP; or the Destination-Host is 1523 not present in the request but the value of peer identity 1524 associated with the connection used to send the request matches 1525 the value of the Origin-Host AVP of the received message that 1526 contained the OC-OLR AVP. 1528 The value of the Destination-Realm AVP in the request matches the 1529 value of the Origin-Realm AVP of the received message that 1530 contained the OC-OLR AVP. 1532 The value of the Application-ID in the Diameter Header of the 1533 request matches the value of the Application-ID of the Diameter 1534 Header of the received message that contained the OC-OLR AVP. 1536 1 A realm report. The overload treatment should apply to requests 1537 for which all of the following conditions are true: 1539 The Destination-Host AVP is absent in the request. 1541 The value of the Destination-Realm AVP in the request matches the 1542 value of the Origin-Realm AVP of the received message that 1543 contained the OC-OLR AVP. 1545 The value of the Application-ID in the Diameter Header of the 1546 request matches the value of the Application-ID of the Diameter 1547 Header of the received message that contained the OC-OLR AVP. 1549 Editor's note: There is still an open issue on the definition of 1550 Realm reports and whether what report types should be supported. 1551 There is consensus that host reports should be supported. There 1552 is discussion on Realm reports and Realm-Routed-Request reports. 1553 The above definition applies to Realm-Routed-Request reports where 1554 Realm reports are defined to apply to all requests that match the 1555 realm, independent of the presence, absence or value of the 1556 Destination-Host AVP. 1558 The default value of the OC-Report-Type AVP is 0 (i.e. the host 1559 report). 1561 The OC-Report-Type AVP is envisioned to be useful for situations 1562 where a reacting node needs to apply different overload treatments 1563 for different "types" of overload. For example, the reacting node(s) 1564 might need to throttle differently requests sent to a specific server 1565 (identified by the Destination-Host AVP in the request) and requests 1566 that can be handled by any server in a realm. The example in 1567 Appendix B.1 illustrates this usage. 1569 6.7. OC-Reduction-Percentage AVP 1571 The OC-Reduction-Percentage AVP (AVP code TBD8) is type of Unsigned32 1572 and describes the percentage of the traffic that the sender is 1573 requested to reduce, compared to what it otherwise would send. The 1574 OC-Reduction-Percentage AVP applies to the default (loss) algorithm 1575 specified in this specification. However, the AVP can be reused for 1576 future abatement algorithms, if its semantics fit into the new 1577 algorithm. 1579 The value of the Reduction-Percentage AVP is between zero (0) and one 1580 hundred (100). Values greater than 100 are ignored. The value of 1581 100 means that all traffic is to be throttled, i.e. the reporting 1582 node is under a severe load and ceases to process any new messages. 1583 The value of 0 means that the reporting node is in a stable state and 1584 has no need for the other endpoint to apply any traffic abatement. 1585 The default value of the OC-Reduction-Percentage AVP is 0. When the 1586 OC-Reduction-Percentage AVP is not present in the overload report, 1587 the default value applies. 1589 6.8. Attribute Value Pair flag rules 1590 +---------+ 1591 |AVP flag | 1592 |rules | 1593 +----+----+ 1594 AVP Section | |MUST| 1595 Attribute Name Code Defined Value Type |MUST| NOT| 1596 +--------------------------------------------------+----+----+ 1597 |OC-Supported-Features TBD1 x.x Grouped | | V | 1598 +--------------------------------------------------+----+----+ 1599 |OC-OLR TBD2 x.x Grouped | | V | 1600 +--------------------------------------------------+----+----+ 1601 |OC-Sequence-Number TBD3 x.x Unsigned64 | | V | 1602 +--------------------------------------------------+----+----+ 1603 |OC-Validity-Duration TBD4 x.x Unsigned32 | | V | 1604 +--------------------------------------------------+----+----+ 1605 |OC-Report-Type TBD5 x.x Enumerated | | V | 1606 +--------------------------------------------------+----+----+ 1607 |OC-Reduction | | | 1608 | -Percentage TBD8 x.x Unsigned32 | | V | 1609 +--------------------------------------------------+----+----+ 1610 |OC-Feature-Vector TBD6 x.x Unsigned64 | | V | 1611 +--------------------------------------------------+----+----+ 1613 As described in the Diameter base protocol [RFC6733], the M-bit 1614 setting for a given AVP is relevant to an application and each 1615 command within that application that includes the AVP. 1617 The Diameter overload control AVPs SHOULD always be sent with the 1618 M-bit cleared when used within existing Diameter applications to 1619 avoid backward compatibility issues. Otherwise, when reused in newly 1620 defined Diameter applications, the DOC related AVPs SHOULD have the 1621 M-bit set. 1623 7. Error Response Codes 1625 Editor's note: This section depends on resolution of issue #27. 1627 8. IANA Considerations 1629 8.1. AVP codes 1631 New AVPs defined by this specification are listed in Section 6. All 1632 AVP codes allocated from the 'Authentication, Authorization, and 1633 Accounting (AAA) Parameters' AVP Codes registry. 1635 8.2. New registries 1637 Three new registries are needed under the 'Authentication, 1638 Authorization, and Accounting (AAA) Parameters' registry. 1640 Section 6.2 defines a new "Overload Control Feature Vector" registry 1641 including the initial assignments. New values can be added into the 1642 registry using the Specification Required policy [RFC5226]. See 1643 Section 6.2 for the initial assignment in the registry. 1645 Section 6.6 defines a new "Overload Report Type" registry with its 1646 initial assignments. New types can be added using the Specification 1647 Required policy [RFC5226]. 1649 9. Security Considerations 1651 This mechanism gives Diameter nodes the ability to request that 1652 downstream nodes send fewer Diameter requests. Nodes do this by 1653 exchanging overload reports that directly affect this reduction. 1654 This exchange is potentially subject to multiple methods of attack, 1655 and has the potential to be used as a Denial-of-Service (DoS) attack 1656 vector. 1658 Overload reports may contain information about the topology and 1659 current status of a Diameter network. This information is 1660 potentially sensitive. Network operators may wish to control 1661 disclosure of overload reports to unauthorized parties to avoid its 1662 use for competitive intelligence or to target attacks. 1664 Diameter does not include features to provide end-to-end 1665 authentication, integrity protection, or confidentiality. This may 1666 cause complications when sending overload reports between non- 1667 adjacent nodes. 1669 9.1. Potential Threat Modes 1671 The Diameter protocol involves transactions in the form of requests 1672 and answers exchanged between clients and servers. These clients and 1673 servers may be peers, that is,they may share a direct transport (e.g. 1674 TCP or SCTP) connection, or the messages may traverse one or more 1675 intermediaries, known as Diameter Agents. Diameter nodes use TLS, 1676 DTLS, or IPSec to authenticate peers, and to provide confidentiality 1677 and integrity protection of traffic between peers. Nodes can make 1678 authorization decisions based on the peer identities authenticated at 1679 the transport layer. 1681 When agents are involved, this presents an effectively hop-by-hop 1682 trust model. That is, a Diameter client or server can authorize an 1683 agent for certain actions, but it must trust that agent to make 1684 appropriate authorization decisions about its peers, and so on. 1686 Since confidentiality and integrity protection occurs at the 1687 transport layer. Agents can read, and perhaps modify, any part of a 1688 Diameter message, including an overload report. 1690 There are several ways an attacker might attempt to exploit the 1691 overload control mechanism. An unauthorized third party might inject 1692 an overload report into the network. If this third party is upstream 1693 of an agent, and that agent fails to apply proper authorization 1694 policies, downstream nodes may mistakenly trust the report. This 1695 attack is at least partially mitigated by the assumption that nodes 1696 include overload reports in Diameter answers but not in requests. 1697 This requires an attacker to have knowledge of the original request 1698 in order to construct a response. Therefore, implementations SHOULD 1699 validate that an answer containing an overload report is a properly 1700 constructed response to a pending request prior to acting on the 1701 overload report. 1703 A similar attack involves an otherwise authorized Diameter node that 1704 sends an inappropriate overload report. For example, a server for 1705 the realm "example.com" might send an overload report indicating that 1706 a competitor's realm "example.net" is overloaded. If other nodes act 1707 on the report, they may falsely believe that "example.net" is 1708 overloaded, effectively reducing that realm's capacity. Therefore, 1709 it's critical that nodes validate that an overload report received 1710 from a peer actually falls within that peer's responsibility before 1711 acting on the report or forwarding the report to other peers. For 1712 example, an overload report from an peer that applies to a realm not 1713 handled by that peer is suspect. 1715 An attacker might use the information in an overload report to assist 1716 in certain attacks. For example, an attacker could use information 1717 about current overload conditions to time a DoS attack for maximum 1718 effect, or use subsequent overload reports as a feedback mechanism to 1719 learn the results of a previous or ongoing attack. 1721 9.2. Denial of Service Attacks 1723 Diameter overload reports can cause a node to cease sending some or 1724 all Diameter requests for an extended period. This makes them a 1725 tempting vector for DoS tacks. Furthermore, since Diameter is almost 1726 always used in support of other protocols, a DoS attack on Diameter 1727 is likely to impact those protocols as well. Therefore, Diameter 1728 nodes MUST NOT honor or forward overload reports from unauthorized or 1729 otherwise untrusted sources. 1731 9.3. Non-Compliant Nodes 1733 When a Diameter node sends an overload report, it cannot assume that 1734 all nodes will comply. A non-compliant node might continue to send 1735 requests with no reduction in load. Requirement 28 [RFC7068] 1736 indicates that the overload control solution cannot assume that all 1737 Diameter nodes in a network are necessarily trusted, and that 1738 malicious nodes not be allowed to take advantage of the overload 1739 control mechanism to get more than their fair share of service. 1741 In the absence of an overload control mechanism, Diameter nodes need 1742 to implement strategies to protect themselves from floods of 1743 requests, and to make sure that a disproportionate load from one 1744 source does not prevent other sources from receiving service. For 1745 example, a Diameter server might reject a certain percentage of 1746 requests from sources that exceed certain limits. Overload control 1747 can be thought of as an optimization for such strategies, where 1748 downstream nodes never send the excess requests in the first place. 1749 However, the presence of an overload control mechanism does not 1750 remove the need for these other protection strategies. 1752 9.4. End-to End-Security Issues 1754 The lack of end-to-end security features makes it far more difficult 1755 to establish trust in overload reports that originate from non- 1756 adjacent nodes. Any agents in the message path may insert or modify 1757 overload reports. Nodes must trust that their adjacent peers perform 1758 proper checks on overload reports from their peers, and so on, 1759 creating a transitive-trust requirement extending for potentially 1760 long chains of nodes. Network operators must determine if this 1761 transitive trust requirement is acceptable for their deployments. 1762 Nodes supporting Diameter overload control MUST give operators the 1763 ability to select which peers are trusted to deliver overload 1764 reports, and whether they are trusted to forward overload reports 1765 from non-adjacent nodes. 1767 The lack of end-to-end confidentiality protection means that any 1768 Diameter agent in the path of an overload report can view the 1769 contents of that report. In addition to the requirement to select 1770 which peers are trusted to send overload reports, operators MUST be 1771 able to select which peers are authorized to receive reports. A node 1772 MUST not send an overload report to a peer not authorized to receive 1773 it. Furthermore, an agent MUST remove any overload reports that 1774 might have been inserted by other nodes before forwarding a Diameter 1775 message to a peer that is not authorized to receive overload reports. 1777 At the time of this writing, the DIME working group is studying 1778 requirements for adding end-to-end security 1780 [I-D.ietf-dime-e2e-sec-req] features to Diameter. These features, 1781 when they become available, might make it easier to establish trust 1782 in non-adjacent nodes for overload control purposes. Readers should 1783 be reminded, however, that the overload control mechanism encourages 1784 Diameter agents to modify AVPs in, or insert additional AVPs into, 1785 existing messages that are originated by other nodes. If end-to-end 1786 security is enabled, there is a risk that such modification could 1787 violate integrity protection. The details of using any future 1788 Diameter end-to-end security mechanism with overload control will 1789 require careful consideration, and are beyond the scope of this 1790 document. 1792 10. Contributors 1794 The following people contributed substantial ideas, feedback, and 1795 discussion to this document: 1797 o Eric McMurry 1799 o Hannes Tschofenig 1801 o Ulrich Wiehe 1803 o Jean-Jacques Trottin 1805 o Maria Cruz Bartolome 1807 o Martin Dolly 1809 o Nirav Salot 1811 o Susan Shishufeng 1813 11. References 1815 11.1. Normative References 1817 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1818 Requirement Levels", BCP 14, RFC 2119, March 1997. 1820 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 1821 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 1822 May 2008. 1824 [RFC5905] Mills, D., Martin, J., Burbank, J., and W. Kasch, "Network 1825 Time Protocol Version 4: Protocol and Algorithms 1826 Specification", RFC 5905, June 2010. 1828 [RFC6733] Fajardo, V., Arkko, J., Loughney, J., and G. Zorn, 1829 "Diameter Base Protocol", RFC 6733, October 2012. 1831 11.2. Informative References 1833 [Cx] 3GPP, , "ETSI TS 129 229 V11.4.0", August 2013. 1835 [I-D.ietf-dime-e2e-sec-req] 1836 Tschofenig, H., Korhonen, J., Zorn, G., and K. Pillay, 1837 "Diameter AVP Level Security: Scenarios and Requirements", 1838 draft-ietf-dime-e2e-sec-req-00 (work in progress), 1839 September 2013. 1841 [PCC] 3GPP, , "ETSI TS 123 203 V11.12.0", December 2013. 1843 [RFC4006] Hakala, H., Mattila, L., Koskinen, J-P., Stura, M., and J. 1844 Loughney, "Diameter Credit-Control Application", RFC 4006, 1845 August 2005. 1847 [RFC5729] Korhonen, J., Jones, M., Morand, L., and T. Tsou, 1848 "Clarifications on the Routing of Diameter Requests Based 1849 on the Username and the Realm", RFC 5729, December 2009. 1851 [RFC7068] McMurry, E. and B. Campbell, "Diameter Overload Control 1852 Requirements", RFC 7068, November 2013. 1854 [S13] 3GPP, , "ETSI TS 129 272 V11.9.0", December 2012. 1856 Appendix A. Issues left for future specifications 1858 The base solution for the overload control does not cover all 1859 possible use cases. A number of solution aspects were intentionally 1860 left for future specification and protocol work. 1862 A.1. Additional traffic abatement algorithms 1864 This specification describes only means for a simple loss based 1865 algorithm. Future algorithms can be added using the designed 1866 solution extension mechanism. The new algorithms need to be 1867 registered with IANA. See Sections 6.1 and 8 for the required IANA 1868 steps. 1870 A.2. Agent Overload 1872 This specification focuses on Diameter endpoint (server or client) 1873 overload. A separate extension will be required to outline the 1874 handling the case of agent overload. 1876 A.3. DIAMETER_TOO_BUSY clarifications 1878 The current [RFC6733] behavior in a case of DIAMETER_TOO_BUSY is 1879 somewhat under specified. For example, there is no information how 1880 long the specific Diameter node is willing to be unavailable. A 1881 specification updating [RFC6733] should clarify the handling of 1882 DIAMETER_TOO_BUSY from the error answer initiating Diameter node 1883 point of view and from the original request initiating Diameter node 1884 point of view. Further, the inclusion of possible additional 1885 information providing AVPs should be discussed and possible be 1886 recommended to be used. 1888 Appendix B. Examples 1890 B.1. Mix of Destination-Realm routed requests and Destination-Host 1891 routed requests 1893 Diameter allows a client to optionally select the destination server 1894 of a request, even if there are agents between the client and the 1895 server. The client does this using the Destination-Host AVP. In 1896 cases where the client does not care if a specific server receives 1897 the request, it can omit Destination-Host and route the request using 1898 the Destination-Realm and Application Id, effectively letting an 1899 agent select the server. 1901 Clients commonly send mixtures of Destination-Host and Destination- 1902 Realm routed requests. For example, in an application that uses user 1903 sessions, a client typically won't care which server handles a 1904 session-initiating requests. But once the session is initiated, the 1905 client will send all subsequent requests in that session to the same 1906 server. Therefore it would send the initial request with no 1907 Destination-Host AVP. If it receives a successful answer, the client 1908 would copy the Origin-Host value from the answer message into a 1909 Destination-Host AVP in each subsequent request in the session. 1911 An agent has very limited options in applying overload abatement to 1912 requests that contain Destination-Host AVPs. It typically cannot 1913 route the request to a different server than the one identified in 1914 Destination-Host. It's only remaining options are to throttle such 1915 requests locally, or to send an overload report back towards the 1916 client so the client can throttle the requests. The second choice is 1917 usually more efficient, since it prevents any throttled requests from 1918 being sent in the first place, and removes the agent's need to send 1919 errors back to the client for each dropped request. 1921 On the other hand, an agent has much more leeway to apply overload 1922 abatement for requests that do not contain Destination-Host AVPs. If 1923 the agent has multiple servers in its peer table for the given realm 1924 and application, it can route such requests to other, less overloaded 1925 servers. 1927 If the overload severity increases, the agent may reach a point where 1928 there is not sufficient capacity across all servers to handle even 1929 realm-routed requests. In this case, the realm itself can be 1930 considered overloaded. The agent may need the client to throttle 1931 realm-routed requests in addition to Destination-Host routed 1932 requests. The overload severity may be different for each server, 1933 and the severity for the realm at is likely to be different than for 1934 any specific server. Therefore, an agent may need to forward, or 1935 originate, multiple overload reports with differing ReportType and 1936 Reduction-Percentage values. 1938 Figure 8 illustrates such a mixed-routing scenario. In this example, 1939 the servers S1, S2, and S3 handle requests for the realm "realm". 1940 Any of the three can handle requests that are not part of a user 1941 session (i.e. routed by Destination-Realm). But once a session is 1942 established, all requests in that session must go to the same server. 1944 Client Agent S1 S2 S3 1945 | | | | | 1946 |(1) Request (DR:realm) | | 1947 |-------->| | | | 1948 | | | | | 1949 | | | | | 1950 | |Agent selects S1 | | 1951 | | | | | 1952 | | | | | 1953 | | | | | 1954 | |(2) Request (DR:realm) | 1955 | |-------->| | | 1956 | | | | | 1957 | | | | | 1958 | | |S1 overloaded, returns OLR 1959 | | | | | 1960 | | | | | 1961 | | | | | 1962 | |(3) Answer (OR:realm,OH:S1,OLR:RT=DH) 1963 | |<--------| | | 1964 | | | | | 1965 | | | | | 1966 | |sees OLR,routes DR traffic to S2&S3 1967 | | | | | 1968 | | | | | 1969 | | | | | 1970 |(4) Answer (OR:realm,OH:S1, OLR:RT=DH) | 1971 |<--------| | | | 1972 | | | | | 1973 | | | | | 1974 |Client throttles requests with DH:S1 | 1975 | | | | | 1976 | | | | | 1977 | | | | | 1978 |(5) Request (DR:realm) | | 1979 |-------->| | | | 1980 | | | | | 1981 | | | | | 1982 | |Agent selects S2 | | 1983 | | | | | 1984 | | | | | 1985 | | | | | 1986 | |(6) Request (DR:realm) | 1987 | |------------------>| | 1988 | | | | | 1989 | | | | | 1990 | | | |S2 is overloaded... 1991 | | | | | 1992 | | | | | 1993 | | | | | 1994 | |(7) Answer (OH:S2, OLR:RT=DH)| 1995 | |<------------------| | 1996 | | | | | 1997 | | | | | 1998 | |Agent sees OLR, realm now overloaded 1999 | | | | | 2000 | | | | | 2001 | | | | | 2002 |(8) Answer (OR:realm,OH:S2, OLR:RT=DH, OLR: RT=R) 2003 |<--------| | | | 2004 | | | | | 2005 | | | | | 2006 |Client throttles DH:S1, DH:S2, and DR:realm 2007 | | | | | 2008 | | | | | 2009 | | | | | 2010 | | | | | 2011 | | | | | 2013 Figure 8: Mix of Destination-Host and Destination-Realm Routed 2014 Requests 2016 1. The client sends a request with no Destination-Host AVP (that is, 2017 a Destination-Realm routed request.) 2019 2. The agent follows local policy to select a server from its peer 2020 table. In this case, the agent selects S2 and forwards the 2021 request. 2023 3. S1 is overloaded. It sends a answer indicating success, but also 2024 includes an overload report. Since the overload report only 2025 applies to S1, the ReportType is "Destination-Host". 2027 4. The agent sees the overload report, and records that S1 is 2028 overloaded by the value in the Reduction-Percentage AVP. It 2029 begins diverting the indicated percentage of realm-routed traffic 2030 from S1 to S2 and S3. Since it can't divert Destination-Host 2031 routed traffic, it forwards the overload report to the client. 2032 This effectively delegates the throttling of traffic with 2033 Destination-Host:S1 to the client. 2035 5. The client sends another Destination-Realm routed request. 2037 6. The agent selects S2, and forwards the request. 2039 7. It turns out that S2 is also overloaded, perhaps due to all that 2040 traffic it took over for S1. S2 returns an successful answer 2041 containing an overload report. Since this report only applies to 2042 S2, the ReportType is "Destination-Host". 2044 8. The agent sees that S2 is also overloaded by the value in 2045 Reduction-Percentage. This value is probably different than the 2046 value from S1's report. The agent diverts the remaining traffic 2047 to S3 as best as it can, but it calculates that the remaining 2048 capacity across all three servers is no longer sufficient to 2049 handle all of the realm-routed traffic. This means the realm 2050 itself is overloaded. The realm's overload percentage is most 2051 likely different than that for either S1 or S2. The agent 2052 forward's S2's report back to the client in the Diameter answer. 2053 Additionally, the agent generates a new report for the realm of 2054 "realm", and inserts that report into the answer. The client 2055 throttles requests with Destination-Host:S1 at one rate, requests 2056 with Destination-Host:S2 at another rate, and requests with no 2057 Destination-Host AVP at yet a third rate. (Since S3 has not 2058 indicated overload, the client does not throttle requests with 2059 Destination-Host:S3.) 2061 Appendix C. Restructuring of -02 version of the draft 2063 This section captures the initial plan for restructuring the DOIC 2064 specification from the -02 version to the new -03 version. 2066 1. Introduction (non normative) 2067 -- Existing Text from section 1. -- 2068 2. Terminology and Abbreviations (non normative) 2069 -- Existing Text from section 2. -- 2070 3. Solution Overview (Non normative) 2071 -- Existing text from section 3. -- 2072 3.1 Overload Control Endpoints (Non normative) 2073 -- New text leveraging text from existing section 5.1 -- 2074 3.2 Piggybacking Principle (Non normative) 2075 -- Existing text from existing section 5.2, with enhancements -- 2076 3.3 DOIC Capability Discovery (Non normative) 2077 -- New text leveraging text from existing section 5.3 -- 2078 3.4 DOIC Overload Condition Reporting (Non normative) 2079 -- New text -- 2080 3.5 DOIC Extensibility (Non normative) 2081 -- New text leveraging text from existing Section 5.4 -- 2082 3.5 Simplified Example Architecture (Non normative) 2083 -- Existing text from section 3.1.6, with enhancements -- 2084 3.6 Considerations for Applications Integrating the DOIC Solution (Non normative) 2085 -- New text -- 2086 3.6.1. Application Classification (Non normative) 2087 -- Existing text from section 3.1.1 -- 2088 3.6.2. Application Type Overload Implications (Non normative) 2089 -- Existing text from section 3.1.2 -- 2090 3.6.3. Request Transaction Classification (Non normative) 2091 -- Existing text from section 3.1.3 -- 2092 3.6.4. Request Type Overload Implications (Non normative) 2093 -- Existing text from section 3.1.4 -- 2094 4. Solution Procedures (Normative) 2095 4.1 Capability Announcement (Normative) 2096 -- Existing text from section 5.3 -- 2097 4.1.1. Reacting Node Behavior (Normative) 2098 -- Existing text from section 5.3.1 -- 2099 4.1.2. Reporting Node Behavior (Normative) 2100 -- Existing text from section 5.3.2 -- 2101 4.1.3. Agent Behavior (Normative) 2102 -- Existing text from section 5.3.3 -- 2103 4.2. Overload Report Processing (Normative) 2104 4.2.1. Overload Control State (Normative) 2105 -- Existing text from section 5.5.1 -- 2106 4.2.2. Reacting Node Behavior (Normative) 2107 -- Existing text from section 5.5.2 -- 2108 4.2.3. Reporting Node Behavior (Normative) 2109 -- Existing text from section 5.5.3 -- 2110 4.2.4. Agent Behavior (Normative) 2111 -- Existing text from section 5.5.4 -- 2112 4.3. Protocol Extensibility (Normative) 2113 -- Existing text from section 5.4 -- 2114 5. Loss Algorithm (Normative) 2115 -- New text pulling from information spread through the document -- 2116 5.1. Overview (Non normative) 2117 -- New text pulling from information spread through the document -- 2118 5.2. Reporting Node Behavior (Normative) 2119 -- New text pulling from information spread through the document -- 2120 5.3. Reacting Node Behavior (Normative) 2121 -- New text pulling from information spread through the document -- 2122 6. Attribute Value Pairs (Normative) 2123 -- Existing text from section 4. -- 2124 6.1. OC-Supported-Features AVP 2125 -- Existing text from section 4.1 -- 2126 6.2. OC-Feature-Vector AVP 2127 -- Existing text from section 4.2 -- 2128 6.3. OC-OLR AVP 2129 -- Existing text from section 4.3 -- 2130 6.4. OC-Sequence-Number AVP 2131 -- Existing text from section 4.4 -- 2132 6.5. OC-Validity-Duration AVP 2133 -- Existing text from section 4.5 -- 2134 6.6. OC-Report-Type AVP 2135 -- Existing text from section 4.6 -- 2136 6.7. OC-Reduction-Percentage AVP 2137 -- Existing text from section 4.7 -- 2138 6.8. Attribute Value Pair flag rules 2139 -- Existing text from section 4.8 -- 2140 7. Error Response Codes 2141 -- New text based on resolution of issue -- 2142 8. IANA Considerations 2143 -- Existing text from section 7. -- 2144 8.1. AVP codes 2145 -- Existing text from section 7.1 -- 2146 8.2. New registries 2147 -- Existing text from section 7.2 -- 2148 9. Security Considerations 2149 -- Existing text from section 8. -- 2150 9.1. Potential Threat Modes 2151 -- Existing text from section 8.1 -- 2152 9.2. Denial of Service Attacks 2153 -- Existing text from section 8.2 -- 2154 9.3. Non-Compliant Nodes 2155 -- Existing text from section 8.3 -- 2156 9.4. End-to End-Security Issues 2157 -- Existing text from section 8.4 -- 2158 10. Contributors 2159 11. References 2160 11.1. Normative References 2161 11.2. Informative References 2162 Appendix A. Issues left for future specifications 2163 A.1. Additional traffic abatement algorithms 2164 A.2. Agent Overload 2165 A.3. DIAMETER_TOO_BUSY clarifications 2166 A.4. Per reacting node reports 2167 Appendix B. Examples 2168 B.1. Mix of Destination-Realm routed requests and Destination- 2169 Host routed requests 2170 Authors' Addresses 2172 Authors' Addresses 2174 Jouni Korhonen (editor) 2175 Broadcom 2176 Porkkalankatu 24 2177 Helsinki FIN-00180 2178 Finland 2180 Email: jouni.nospam@gmail.com 2182 Steve Donovan (editor) 2183 Oracle 2184 7460 Warren Parkway 2185 Frisco, Texas 75034 2186 United States 2188 Email: srdonovan@usdonovans.com 2190 Ben Campbell 2191 Oracle 2192 7460 Warren Parkway 2193 Frisco, Texas 75034 2194 United States 2196 Email: ben@nostrum.com 2198 Lionel Morand 2199 Orange Labs 2200 38/40 rue du General Leclerc 2201 Issy-Les-Moulineaux Cedex 9 92794 2202 France 2204 Phone: +33145296257 2205 Email: lionel.morand@orange.com