idnits 2.17.00 (12 Aug 2021) /tmp/idnits57104/draft-ietf-abfab-arch-10.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (December 31, 2013) is 3062 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- -- Looks like a reference, but probably isn't: '1' on line 143 -- Looks like a reference, but probably isn't: '2' on line 752 == Outdated reference: draft-ietf-abfab-gss-eap has been published as RFC 7055 == Outdated reference: draft-ietf-abfab-aaa-saml has been published as RFC 7833 == Outdated reference: draft-ietf-radext-nai has been published as RFC 7542 == Outdated reference: draft-ietf-radext-radius-fragmentation has been published as RFC 7499 -- No information found for draft-ietf-emu-eap-tunnel-method - is the name correct? == Outdated reference: draft-ietf-radext-dtls has been published as RFC 7360 == Outdated reference: draft-ietf-radext-dynamic-discovery has been published as RFC 7585 Summary: 0 errors (**), 0 flaws (~~), 7 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 ABFAB J. Howlett 3 Internet-Draft JANET(UK) 4 Intended status: Informational S. Hartman 5 Expires: July 4, 2014 Painless Security 6 H. Tschofenig 7 Nokia Siemens Networks 8 E. Lear 9 Cisco Systems GmbH 10 J. Schaad 11 Soaring Hawk Consulting 12 December 31, 2013 14 Application Bridging for Federated Access Beyond Web (ABFAB) 15 Architecture 16 draft-ietf-abfab-arch-10.txt 18 Abstract 20 Over the last decade a substantial amount of work has occurred in the 21 space of federated access management. Most of this effort has 22 focused on two use cases: network access and web-based access. 23 However, the solutions to these use cases that have been proposed and 24 deployed tend to have few common building blocks in common. 26 This memo describes an architecture that makes use of extensions to 27 the commonly used security mechanisms for both federated and non- 28 federated access management, including the Remote Authentication Dial 29 In User Service (RADIUS) the Generic Security Service (GSS), the 30 Extensible Authentication Protocol (EAP) and the Security Assertion 31 Markup Language (SAML). The architecture addresses the problem of 32 federated access management to primarily non-web-based services, in a 33 manner that will scale to large numbers of identity providers, 34 relying parties, and federations. 36 Status of This Memo 38 This Internet-Draft is submitted in full conformance with the 39 provisions of BCP 78 and BCP 79. 41 Internet-Drafts are working documents of the Internet Engineering 42 Task Force (IETF). Note that other groups may also distribute 43 working documents as Internet-Drafts. The list of current Internet- 44 Drafts is at http://datatracker.ietf.org/drafts/current/. 46 Internet-Drafts are draft documents valid for a maximum of six months 47 and may be updated, replaced, or obsoleted by other documents at any 48 time. It is inappropriate to use Internet-Drafts as reference 49 material or to cite them other than as "work in progress." 51 This Internet-Draft will expire on July 4, 2014. 53 Copyright Notice 55 Copyright (c) 2013 IETF Trust and the persons identified as the 56 document authors. All rights reserved. 58 This document is subject to BCP 78 and the IETF Trust's Legal 59 Provisions Relating to IETF Documents 60 (http://trustee.ietf.org/license-info) in effect on the date of 61 publication of this document. Please review these documents 62 carefully, as they describe your rights and restrictions with respect 63 to this document. Code Components extracted from this document must 64 include Simplified BSD License text as described in Section 4.e of 65 the Trust Legal Provisions and are provided without warranty as 66 described in the Simplified BSD License. 68 Table of Contents 70 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 71 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 5 72 1.1.1. Channel Binding . . . . . . . . . . . . . . . . . . . 6 73 1.2. An Overview of Federation . . . . . . . . . . . . . . . . 7 74 1.3. Challenges for Contemporary Federation . . . . . . . . . 10 75 1.4. An Overview of ABFAB-based Federation . . . . . . . . . . 10 76 1.5. Design Goals . . . . . . . . . . . . . . . . . . . . . . 13 77 2. Architecture . . . . . . . . . . . . . . . . . . . . . . . . 14 78 2.1. Relying Party to Identity Provider . . . . . . . . . . . 15 79 2.1.1. AAA, RADIUS and Diameter . . . . . . . . . . . . . . 16 80 2.1.2. Discovery and Rules Determination . . . . . . . . . . 18 81 2.1.3. Routing and Technical Trust . . . . . . . . . . . . . 19 82 2.1.4. AAA Security . . . . . . . . . . . . . . . . . . . . 20 83 2.1.5. SAML Assertions . . . . . . . . . . . . . . . . . . . 21 84 2.2. Client To Identity Provider . . . . . . . . . . . . . . . 23 85 2.2.1. Extensible Authentication Protocol (EAP) . . . . . . 23 86 2.2.2. EAP Channel Binding . . . . . . . . . . . . . . . . . 25 87 2.3. Client to Relying Party . . . . . . . . . . . . . . . . . 25 88 2.3.1. GSS-API . . . . . . . . . . . . . . . . . . . . . . . 25 89 2.3.2. Protocol Transport . . . . . . . . . . . . . . . . . 27 90 2.3.3. Reauthentication . . . . . . . . . . . . . . . . . . 27 91 3. Application Security Services . . . . . . . . . . . . . . . . 28 92 3.1. Authentication . . . . . . . . . . . . . . . . . . . . . 28 93 3.2. GSS-API Channel Binding . . . . . . . . . . . . . . . . . 29 94 3.3. Host-Based Service Names . . . . . . . . . . . . . . . . 30 95 3.4. Additional GSS-API Services . . . . . . . . . . . . . . . 32 97 4. Privacy Considerations . . . . . . . . . . . . . . . . . . . 32 98 4.1. Entities and their roles . . . . . . . . . . . . . . . . 33 99 4.2. Privacy Aspects of ABFAB Communication Flows . . . . . . 34 100 4.2.1. Client to RP . . . . . . . . . . . . . . . . . . . . 34 101 4.2.2. Client to IdP (via Federation Substrate) . . . . . . 35 102 4.2.3. IdP to RP (via Federation Substrate) . . . . . . . . 36 103 4.3. Relationship between User and Entities . . . . . . . . . 37 104 4.4. Accounting Information . . . . . . . . . . . . . . . . . 37 105 4.5. Collection and retention of data and identifiers . . . . 37 106 4.6. User Participation . . . . . . . . . . . . . . . . . . . 38 107 5. Security Considerations . . . . . . . . . . . . . . . . . . . 38 108 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 39 109 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 39 110 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 40 111 8.1. Normative References . . . . . . . . . . . . . . . . . . 40 112 8.2. Informative References . . . . . . . . . . . . . . . . . 41 113 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 43 115 1. Introduction 117 The Internet uses numerous security mechanisms to manage access to 118 various resources. These mechanisms have been generalized and scaled 119 over the last decade through mechanisms such as Simple Authentication 120 and Security Layer (SASL) with the Generic Security Server 121 Application Program Interface (GSS-API) (known as the GS2 family) 122 [RFC5801], Security Assertion Markup Language (SAML) 123 [OASIS.saml-core-2.0-os], and the Authentication, Authorization, and 124 Accounting (AAA) architecture as embodied in RADIUS [RFC2865] and 125 Diameter [RFC6733]. 127 A Relying Party (RP) is the entity that manages access to some 128 resource. The entity that is requesting access to that resource is 129 often described as the Client. Many security mechanisms are 130 manifested as an exchange of information between these entities. The 131 RP is therefore able to decide whether the Client is authorized, or 132 not. 134 Some security mechanisms allow the RP to delegate aspects of the 135 access management decision to an entity called the Identity Provider 136 (IdP). This delegation requires technical signaling, trust and a 137 common understanding of semantics between the RP and IdP. These 138 aspects are generally managed within a relationship known as a 139 'federation'. This style of access management is accordingly 140 described as 'federated access management'. 142 Federated access management has evolved over the last decade through 143 specifications like SAML [OASIS.saml-core-2.0-os], OpenID [1], OAuth 145 [RFC6749] and WS-Trust [WS-TRUST]. The benefits of federated access 146 management include: 148 Single or Simplified sign-on: 150 An Internet service can delegate access management, and the 151 associated responsibilities such as identity management and 152 credentialing, to an organization that already has a long-term 153 relationship with the Client. This is often attractive as Relying 154 Parties frequently do not want these responsibilities. The Client 155 also requires fewer credentials, which is also desirable. 157 Data Minimization and User Participation: 159 Often a Relying Party does not need to know the identity of a 160 Client to reach an access management decision. It is frequently 161 only necessary for the Relying Party to know specific attributes 162 about the client, for example, that the client is affiliated with 163 a particular organization or has a certain role or entitlement. 164 Sometimes the RP only needs to know a pseudonym of the client. 166 Prior to the release of attributes to the RP from the IdP, the IdP 167 will check configuration and policy to determine if the attributes 168 are to be released. There is currently no direct client 169 participation in this decision. 171 Provisioning: 173 Sometimes a Relying Party needs, or would like, to know more about 174 a client than an affiliation or a pseudonym. For example, a 175 Relying Party may want the Client's email address or name. Some 176 federated access management technologies provide the ability for 177 the IdP to supply this information, either on request by the RP or 178 unsolicited. 180 This memo describes the Application Bridging for Federated Access 181 Beyond the Web (ABFAB) architecture. This architecture makes use of 182 extensions to the commonly used security mechanisms for both 183 federated and non-federated access management, including RADIUS, the 184 Generic Security Service (GSS), the Extensible Authentication 185 Protocol (EAP) and SAML. The architecture should be extended to use 186 Diameter in the future. The architecture addresses the problem of 187 federated access management primarily for non-web-based services. It 188 does so in a manner that will scale to large numbers of identity 189 providers, relying parties, and federations. 191 1.1. Terminology 193 This document uses identity management and privacy terminology from 194 [RFC6973]. In particular, this document uses the terms identity 195 provider, relying party, identifier, pseudonymity, unlinkability, and 196 anonymity. 198 In this architecture the IdP consists of the following components: an 199 EAP server, a RADIUS server, and optionally a SAML Assertion service. 201 This document uses the term Network Access Identifier (NAI), as 202 defined in [I-D.ietf-radext-nai]. An NAI consists of a realm 203 identifier, which is associated with an IdP and a username which is 204 associated with a specific client of the IdP. 206 One of the problems people will find with reading this document is 207 that the terminology sometimes appears to be inconsistent. This is 208 due the fact that the terms used by the different standards we are 209 referencing are not consistent. In general the document uses either 210 the ABFAB term or the term associated with the standard under 211 discussion as appropriate. For reference we include this table which 212 maps the different terms into a single table. 214 +----------+-----------+--------------------+-----------------------+ 215 | Protocol | Client | Relying Party | Identity Provider | 216 +----------+-----------+--------------------+-----------------------+ 217 | ABFAB | Client | Relying Party (RP) | Identity Provider | 218 | | | | (IdP) | 219 | | | | | 220 | | Initiator | Acceptor | | 221 | | | | | 222 | | | Server | | 223 | | | | | 224 | SAML | Subject | Service Provider | Issuer | 225 | | | | | 226 | GSS-API | Initiator | Acceptor | | 227 | | | | | 228 | EAP | EAP peer | EAP Authenticator | EAP server | 229 | | | | | 230 | AAA | | AAA Client | AAA server | 231 | | | | | 232 | RADIUS | user | NAS | RADIUS server | 233 | | | | | 234 | | | RADIUS client | | 235 +----------+-----------+--------------------+-----------------------+ 237 Table 1. Terminology 239 Note that in some cases a cell has been left empty; in these cases 240 there is no name that represents the entity. 242 1.1.1. Channel Binding 244 This document uses the term channel binding with two different 245 meanings. 247 EAP channel binding is used to provide GSS-API naming semantics. EAP 248 channel binding sends a set of attributes from the peer to the EAP 249 server either as part of the EAP conversation or as part of a secure 250 association protocol. In addition, attributes are sent in the 251 backend protocol from the EAP authenticator to the EAP server. The 252 EAP server confirms the consistency of these attributes and provides 253 the confirmation back to the peer. In this document, channel binding 254 without qualification refers to EAP channel binding. 256 GSS-API channel binding provides protection against man-in-the-middle 257 attacks when GSS-API is used for authentication inside of some 258 tunnel; it is similar to a facility called cryptographic binding in 259 EAP. The binding works by each side deriving a cryptographic value 260 from the tunnel itself and then using that cryptographic value to 261 prove to the other side that it knows the value. 263 See [RFC5056] for a discussion of the differences between these two 264 facilities. However, the difference can be summarized as GSS-API 265 channel binding says that there is nobody between the client and the 266 EAP authenticator while EAP channel binding allows the client to have 267 knowledge about attributes of the EAP authenticator (such as its 268 name). 270 Typically when considering both EAP and GSS-API channel binding, 271 people think of channel binding in combination with mutual 272 authentication. This is sufficiently common that without additional 273 qualification channel binding should be assumed to imply mutual 274 authentication. In GSS-API, without mutualtion only the acceptor has 275 authenticated the initator. Similarly in EAP, only the EAP server 276 has authenticated the peer. That's sometimes useful. Consider for 277 example a user who wishes to access a protected resource for a shared 278 whiteboard in a conference room. The whiteboard is the acceptor; it 279 knows that the initiator is authorized to give it a presentation and 280 the user can validate the whitebord got the correct presentation by 281 visual means. (The presention should not be confiduatal in this 282 case.) If channel binding is used without mutual authentication, it 283 is effectively a request to disclose the resource in the context of a 284 particular channel. Such an authentication would be similar in 285 concept to a holder-of-key SAML assertion. However, also note that 286 while it is not happening in the protocol, mutual authentication is 287 happening in the overall system: the user is able to visually 288 authenticate the content. This is consistent with all uses of 289 channel binding without protocol level mutual authentication found so 290 far. 292 1.2. An Overview of Federation 294 In the previous section we introduced the following entities: 296 o the Client, 298 o the Identity Provider, and 300 o the Relying Party. 302 The final entity that needs to be introduced is the Individual. An 303 Individual is a human being that is using the Client. In any given 304 situation, an Individual may or may not exist. Clients can act 305 either as front ends for Individuals or they may be independent 306 entities that are setup and allowed to run autonomously. An example 307 of such an entity can be found in the trust routing protocol where 308 the routers use ABFAB to authenticate to each other. 310 These entities and their relationships are illustrated graphically in 311 Figure 1. 313 ,----------\ ,---------\ 314 | Identity | Federation | Relying | 315 | Provider + <-------------------> + Party | 316 `----------' '---------' 317 < 318 \ 319 \ Authentication 320 \ 321 \ 322 \ 323 \ 324 \ +---------+ 325 \ | | O 326 v| Client | \|/ Individual 327 | | | 328 +---------+ / \ 330 Figure 1: Entities and their Relationships 332 The relationships between the entities in Figure 1 are: 334 Federation 336 The Identity Provider and the Relying Parties are part of a 337 Federation. The relationship may be direct (they have an explicit 338 trust relationship) or transitive (the trust relationship is 339 mediated by one or more entities). The federation relationship is 340 governed by a federation agreement. Within a single federation, 341 there may be multiple Identity Providers as well as multiple 342 Relying Parties. 344 Authentication 346 There is a direct relationship between the Client and the Identity 347 Provider by which the entities trust and can securely authenticate 348 each other. 350 A federation agreement typically encompasses operational 351 specifications and legal rules: 353 Operational Specifications: 355 These include the technical specifications (e.g. protocols used to 356 communicate between the three parties), process standards, 357 policies, identity proofing, credential and authentication 358 algorithm requirements, performance requirements, assessment and 359 audit criteria, etc. The goal of operational specifications is to 360 provide enough definition that the system works and 361 interoperability is possible. 363 Legal Rules: 365 The legal rules take the legal framework into consideration and 366 provide contractual obligations for each entity. The rules define 367 the responsibilities of each party and provide further 368 clarification of the operational specifications. These legal 369 rules regulate the operational specifications, make operational 370 specifications legally binding to the participants, define and 371 govern the rights and responsibilities of the participants. The 372 legal rules may, for example, describe liability for losses, 373 termination rights, enforcement mechanisms, measures of damage, 374 dispute resolution, warranties, etc. 376 The Operational Specifications can demand the usage of a 377 sophisticated technical infrastructure, including requirements on the 378 message routing intermediaries, to offer the required technical 379 functionality. In other environments, the Operational Specifications 380 require fewer technical components in order to meet the required 381 technical functionality. 383 The Legal Rules include many non-technical aspects of federation, 384 such as business practices and legal arrangements, which are outside 385 the scope of the IETF. The Legal Rules can still have an impact on 386 the architectural setup or on how to ensure the dynamic establishment 387 of trust. 389 While a federation agreement is often discussed within the context of 390 formal relationships, such as between an enterprise and an employee 391 or a government and a citizen, a federation agreement does not have 392 to require any particular level of formality. For an IdP and a 393 Client, it is sufficient for a relationship to be established by 394 something as simple as using a web form and confirmation email. For 395 an IdP and an RP, it is sufficient for the IdP to publish contact 396 information along with a public key and for the RP to use that data. 397 Within the framework of ABFAB, it will generally be required that a 398 mechanism exists for the IdP to be able to trust the identity of the 399 RP, if this is not present then the IdP cannot provide the assurances 400 to the client that the identity of the RP has been established. 402 The nature of federation dictates that there is some form of 403 relationship between the identity provider and the relying party. 404 This is particularly important when the relying party wants to use 405 information obtained from the identity provider for access management 406 decisions and when the identity provider does not want to release 407 information to every relying party (or only under certain 408 conditions). 410 While it is possible to have a bilateral agreement between every IdP 411 and every RP; on an Internet scale this setup requires the 412 introduction of the multi-lateral federation concept, as the 413 management of such pair-wise relationships would otherwise prove 414 burdensome. 416 The IdP will typically have a long-term relationship with the Client. 417 This relationship typically involves the IdP positively identifying 418 and credentialing the Client (for example, at time of employment 419 within an organization). When dealing with individuals, this process 420 is called identity proofing [NIST-SP.800-63]. The relationship will 421 often be instantiated within an agreement between the IdP and the 422 Client (for example, within an employment contract or terms of use 423 that stipulates the appropriate use of credentials and so forth). 425 The nature and quality of the relationship between the Client and the 426 IdP is an important contributor to the level of trust that an RP may 427 attribute to an assertion describing a Client made by an IdP. This 428 is sometimes described as the Level of Assurance [NIST-SP.800-63]. 430 Federation does not require an a priori relationship or a long-term 431 relationship between the RP and the Client; it is this property of 432 federation that yields many of its benefits. However, federation 433 does not preclude the possibility of a pre-existing relationship 434 between the RP and the Client, nor that they may use the introduction 435 to create a new long-term relationship independent of the federation. 437 Finally, it is important to reiterate that in some scenarios there 438 might indeed be an Individual behind the Client and in other cases 439 the Client may be autonomous. 441 1.3. Challenges for Contemporary Federation 443 As the number of federated services has proliferated, the role of the 444 individual can become ambiguous in certain circumstances. For 445 example, a school might provide online access for a student's grades 446 to their parents for review, and to the student's teacher for 447 modification. A teacher who is also a parent must clearly 448 distinguish her role upon access. 450 Similarly, as the number of federations proliferates, it becomes 451 increasingly difficult to discover which identity provider(s) a user 452 is associated with. This is true for both the web and non-web case, 453 but is particularly acute for the latter as many non-web 454 authentication systems are not semantically rich enough on their own 455 to allow for such ambiguities. For instance, in the case of an email 456 provider, the use of SMTP and IMAP protocols do not have the ability 457 for the server to get additional information, beyond the clients NAI, 458 in order to provide additional input to decide between multiple 459 federations it may be associated with. However, the building blocks 460 do exist to add this functionality. 462 1.4. An Overview of ABFAB-based Federation 464 The previous section described the general model of federation, and 465 the application of access management within the federation. This 466 section provides a brief overview of ABFAB in the context of this 467 model. 469 In this example, a client is attempting to connect to a server in 470 order to either get access to some data or perform some type of 471 transaction. In order for the client to mutually authenticate with 472 the server, the following steps are taken in an ABFAB architecture: 474 1. Client Configuration: The Client Application is configured with 475 an NAI assigned by the IdP. It is also configured with any 476 keys, certificates, passwords or other secret and public 477 information needed to run the EAP protocols between it and the 478 IdP. 480 2. Authentication mechanism selection: The GSS-EAP GSS-API 481 mechanism is selected for authentication/authorization. 483 3. Client provides an NAI to RP: The client application sets up a 484 transport to the RP and begins the GSS-EAP authentication. In 485 response, the RP sends an EAP request message (nested in the 486 GSS-EAP protocol) asking for the Client's name. The Client 487 sends an EAP response with an NAI name form that, at a minimum, 488 contains the realm portion of its full NAI. 490 4. Discovery of federated IdP: The RP uses pre-configured 491 information or a federation proxy to determine what IdP to use 492 based on policy and the realm portion of the provided Client 493 NAI. This is discussed in detail below (Section 2.1.2). 495 5. Request from Relying Party to IdP: Once the RP knows who the IdP 496 is, it (or its agent) will send a RADIUS request to the IdP. 497 The RADIUS access request encapsulates the EAP response. At 498 this stage, the RP will likely have no idea who the client is. 499 The RP sends its identity to the IdP in AAA attributes, and it 500 may send a SAML Attribute Request in a AAA attribute. The AAA 501 network checks that the identity claimed by the RP is valid. 503 6. IdP begins EAP with the client: The IdP sends an EAP message to 504 the client with an EAP method to be used. The IdP should not 505 re-request the clients name in this message, but clients need to 506 be able to handle it. In this case the IdP must accept a realm 507 only in order to protect the client's name from the RP. The 508 available and appropriate methods are discussed below in this 509 memo (Section 2.2.1). 511 7. The EAP protocol is run: A bunch of EAP messages are passed 512 between the client (EAP peer) and the IdP (EAP server), until 513 the result of the authentication protocol is determined. The 514 number and content of those messages depends on the EAP method 515 selected. If the IdP is unable to authenticate the client, the 516 IdP sends an EAP failure message to the RP. As part of the EAP 517 protocol, the client sends a channel bindings EAP message to the 518 IdP (Section 2.2.2). In the channel binding message the client 519 identifies, among other things, the RP to which it is attempting 520 to authenticate. The IdP checks the channel binding data from 521 the client with that provided by the RP via the AAA protocol. 522 If the bindings do not match the IdP sends an EAP failure 523 message to the RP. 525 8. Successful EAP Authentication: At this point, the IdP (EAP 526 server) and client (EAP peer) have mutually authenticated each 527 other. As a result, the client and the IdP hold two 528 cryptographic keys: a Master Session Key (MSK), and an Extended 529 MSK (EMSK). At this point the client has a level of assurance 530 about the identity of the RP based on the name checking the IdP 531 has done using the RP naming information from the AAA framework 532 and from the client (by the channel binding data). 534 9. Local IdP Policy Check: At this stage, the IdP checks local 535 policy to determine whether the RP and client are authorized for 536 a given transaction/service, and if so, what if any, attributes 537 will be released to the RP. If the IdP gets a policy failure, 538 it sends an EAP failure message to the RP. (The RP will have 539 done its policy checks during the discovery process.) 541 10. IdP provides the RP with the MSK: The IdP sends a positive 542 result EAP to the RP, along with an optional set of AAA 543 attributes associated with the client (usually as one or more 544 SAML assertions). In addition, the EAP MSK is returned to the 545 RP. 547 11. RP Processes Results: When the RP receives the result from the 548 IdP, it should have enough information to either grant or refuse 549 a resource access request. It may have information that 550 associates the client with specific authorization identities. 551 If additional attributes are needed from the IdP the RP may make 552 a new SAML Request to the IdP. It will apply these results in 553 an application-specific way. 555 12. RP returns results to client: Once the RP has a response it must 556 inform the client application of the result. If all has gone 557 well, all are authenticated, and the application proceeds with 558 appropriate authorization levels. The client can now complete 559 the authentication of the RP by the use of the EAP MSK value. 561 An example communication flow is given below: 563 Relying Client Identity 564 Party App Provider 566 | (1) | Client Configuration 567 | | | 568 |<-----(2)----->| | Mechanism Selection 569 | | | 570 |<-----(3)-----<| | NAI transmitted to RP 571 | | | 572 |<=====(4)====================>| Discovery 573 | | | 574 |>=====(5)====================>| Access request from RP to IdP 575 | | | 576 | |< - - (6) - -<| EAP method to Client 577 | | | 578 | |< - - (7) - ->| EAP Exchange to authenticate 579 | | | Client 580 | | | 581 | | (8 & 9) Local Policy Check 582 | | | 583 |<====(10)====================<| IdP Assertion to RP 584 | | | 585 (11) | | RP processes results 586 | | | 587 |>----(12)----->| | Results to client app. 589 ----- = Between Client App and RP 590 ===== = Between RP and IdP 591 - - - = Between Client App and IdP (via RP) 593 1.5. Design Goals 595 Our key design goals are as follows: 597 o Each party of a transaction will be authenticated, although 598 perhaps not identified, and the client will be authorized for 599 access to a specific resource. 601 o Means of authentication is decoupled so as to allow for multiple 602 authentication methods. 604 o The architecture requires no sharing of long term private keys 605 between clients and servers. 607 o The system will scale to large numbers of identity providers, 608 relying parties, and users. 610 o The system will be designed primarily for non-Web-based 611 authentication. 613 o The system will build upon existing standards, components, and 614 operational practices. 616 Designing new three party authentication and authorization protocols 617 is hard and fraught with risk of cryptographic flaws. Achieving 618 widespread deployment is even more difficult. A lot of attention on 619 federated access has been devoted to the Web. This document instead 620 focuses on a non-Web-based environment and focuses on those protocols 621 where HTTP is not used. Despite the increased excitement for 622 layering every protocol on top of HTTP there are still a number of 623 protocols available that do not use HTTP-based transports. Many of 624 these protocols are lacking a native authentication and authorization 625 framework of the style shown in Figure 1. 627 2. Architecture 629 We have already introduced the federated access architecture, with 630 the illustration of the different actors that need to interact, but 631 did not expand on the specifics of providing support for non-Web 632 based applications. This section details this aspect and motivates 633 design decisions. The main theme of the work described in this 634 document is focused on re-using existing building blocks that have 635 been deployed already and to re-arrange them in a novel way. 637 Although this architecture assumes updates to the relying party, the 638 client application, and the Identity Provider, those changes are kept 639 at a minimum. A mechanism that can demonstrate deployment benefits 640 (based on ease of update of existing software, low implementation 641 effort, etc.) is preferred and there may be a need to specify 642 multiple mechanisms to support the range of different deployment 643 scenarios. 645 There are a number of ways for encapsulating EAP into an application 646 protocol. For ease of integration with a wide range of non-Web based 647 application protocols the usage of the GSS-API was chosen. A 648 description of the technical specification can be found in 649 [I-D.ietf-abfab-gss-eap]. 651 The architecture consists of several building blocks, which is shown 652 graphically in Figure 2. In the following sections, we discuss the 653 data flow between each of the entities, the protocols used for that 654 data flow and some of the trade-offs made in choosing the protocols. 656 +--------------+ 657 | Identity | 658 | Provider | 659 | (IdP) | 660 +-^----------^-+ 661 * EAP o RADIUS 662 * o 663 --v----------v-- 664 /// \\\ 665 // \\ 666 | Federation | 667 | Substrate | 668 \\ // 669 \\\ /// 670 --^----------^-- 671 * EAP o RADIUS 672 * o 673 +-------------+ +-v----------v--+ 674 | | | | 675 | Client | EAP/EAP Method | Relying Party | 676 | Application |<****************>| (RP) | 677 | | GSS-API | | 678 | |<---------------->| | 679 | | Application | | 680 | | Protocol | | 681 | |<================>| | 682 +-------------+ +---------------+ 684 Legend: 686 <****>: Client-to-IdP Exchange 687 <---->: Client-to-RP Exchange 688 : RP-to-IdP Exchange 689 <====>: Protocol through which GSS-API/GS2 exchanges are tunneled 691 Figure 2: ABFAB Protocol Instantiation 693 2.1. Relying Party to Identity Provider 695 Communications between the Relying Party and the Identity Provider is 696 done by the federation substrate. This communication channel is 697 responsible for: 699 o Establishing the trust relationship between the RP and the IdP. 701 o Determining the rules governing the relationship. 703 o Conveying authentication packets from the client to the IdP and 704 back. 706 o Providing the means of establishing a trust relationship between 707 the RP and the client. 709 o Providing a means for the RP to obtain attributes about the client 710 from the IdP. 712 The ABFAB working group has chosen the AAA framework for the messages 713 transported between the RP and IdP. The AAA framework supports the 714 requirements stated above as follows: 716 o The AAA backbone supplies the trust relationship between the RP 717 and the IdP. 719 o The agreements governing a specific AAA backbone contains the 720 rules governing the relationships within the AAA federation. 722 o A method exists for carrying EAP packets within RADIUS [RFC3579] 723 and Diameter [RFC4072]. 725 o The use of EAP channel binding [RFC6677] along with the core ABFAB 726 protocol provide the pieces necessary to establish the identities 727 of the RP and the client, while EAP provides the cryptographic 728 methods for the RP and the client to validate they are talking to 729 each other. 731 o A method exists for carrying SAML packets within RADIUS 732 [I-D.ietf-abfab-aaa-saml] which allows the RP to query attributes 733 about the client from the IdP. 735 Future protocols that support the same framework but do different 736 routing may be used in the future. One such effort is to setup a 737 framework that creates a trusted point-to-point channel on the fly. 739 2.1.1. AAA, RADIUS and Diameter 741 Interestingly, for network access authentication the usage of the AAA 742 framework with RADIUS [RFC2865] and Diameter [RFC6733] was quite 743 successful from a deployment point of view. To map to the 744 terminology used in Figure 1 to the AAA framework the IdP corresponds 745 to the AAA server, the RP corresponds to the AAA client, and the 746 technical building blocks of a federation are AAA proxies, relays and 747 redirect agents (particularly if they are operated by third parties, 748 such as AAA brokers and clearing houses). The front-end, i.e. the 749 end host to AAA client communication, is in case of network access 750 authentication offered by link layer protocols that forward 751 authentication protocol exchanges back-and-forth. An example of a 752 large scale RADIUS-based federation is EDUROAM [2]. 754 By using the AAA framework, ABFAB gets a lot of mileage as many of 755 the federation agreements already exist and merely need to be 756 expanded to cover the ABFAB additions. The AAA framework has already 757 addressed some of the problems outlined above. For example, 759 o It already has a method for routing requests based on a domain. 761 o It already has an extensible architecture allowing for new 762 attributes to be defined and transported. 764 o Pre-existing relationships can be re-used. 766 The astute reader will notice that RADIUS and Diameter have 767 substantially similar characteristics. Why not pick one? RADIUS and 768 Diameter are deployed in different environments. RADIUS can often be 769 found in enterprise and university networks, and is also in use by 770 fixed network operators. Diameter, on the other hand, is deployed by 771 mobile operators. Another key difference is that today RADIUS is 772 largely transported upon UDP. We leave as a deployment decision, 773 which protocol will be appropriate. The protocol defines all the 774 necessary new AAA attributes as RADIUS attributes. A future document 775 would define the same AAA attributes for a Diameter environment. We 776 also note that there exist proxies which convert from RADIUS to 777 Diameter and back. This makes it possible for both to be deployed in 778 a single federation substrate. 780 Through the integrity protection mechanisms in the AAA framework, the 781 identity provider can establish technical trust that messages are 782 being sent by the appropriate relying party. Any given interaction 783 will be associated with one federation at the policy level. The 784 legal or business relationship defines what statements the identity 785 provider is trusted to make and how these statements are interpreted 786 by the relying party. The AAA framework also permits the relying 787 party or elements between the relying party and identity provider to 788 make statements about the relying party. 790 The AAA framework provides transport for attributes. Statements made 791 about the client by the identity provider, statements made about the 792 relying party and other information are transported as attributes. 794 One demand that the AAA substrate makes of the upper layers is that 795 they must properly identify the end points of the communication. It 796 must be possible for the AAA client at the RP to determine where to 797 send each RADIUS or Diameter message. Without this requirement, it 798 would be the RP's responsibility to determine the identity of the 799 client on its own, without the assistance of an IdP. This 800 architecture makes use of the Network Access Identifier (NAI), where 801 the IdP is indicated by the realm component [I-D.ietf-radext-nai]. 802 The NAI is represented and consumed by the GSS-API layer as 803 GSS_C_NT_USER_NAME as specified in [RFC2743]. The GSS-API EAP 804 mechanism includes the NAI in the EAP Response/Identity message. 806 As of the time this document was published, no profiles for the use 807 of Diameter have been created. 809 2.1.2. Discovery and Rules Determination 811 While we are using the AAA protocols to communicate with the IdP, the 812 RP may have multiple federation substrates to select from. The RP 813 has a number of criteria that it will use in selecting which of the 814 different federations to use: 816 o The federation selected must be able to communicate with the IdP. 818 o The federation selected must match the business rules and 819 technical policies required for the RP security requirements. 821 The RP needs to discover which federation will be used to contact the 822 IdP. The first selection criteria used during discovery is going to 823 be the name of the IdP to be contacted. The second selection 824 criteria used during discovery is going to be the set of business 825 rules and technical policies governing the relationship; this is 826 called rules determination. The RP also needs to establish technical 827 trust in the communications with the IdP. 829 Rules determination covers a broad range of decisions about the 830 exchange. One of these is whether the given RP is permitted to talk 831 to the IdP using a given federation at all, so rules determination 832 encompasses the basic authorization decision. Other factors are 833 included, such as what policies govern release of information about 834 the client to the RP and what policies govern the RP's use of this 835 information. While rules determination is ultimately a business 836 function, it has significant impact on the technical exchanges. The 837 protocols need to communicate the result of authorization. When 838 multiple sets of rules are possible, the protocol must disambiguate 839 which set of rules are in play. Some rules have technical 840 enforcement mechanisms; for example in some federations 841 intermediaries validate information that is being communicated within 842 the federation. 844 At the time of writing no protocol mechanism has been specified to 845 allow a AAA client to determine whether a AAA proxy will indeed be 846 able to route AAA requests to a specific IdP. The AAA routing is 847 impacted by business rules and technical policies that may be quite 848 complex and at the present time, the route selection is based on 849 manual configuration. 851 2.1.3. Routing and Technical Trust 853 Several approaches to having messages routed through the federation 854 substrate are possible. These routing methods can most easily be 855 classified based on the mechanism for technical trust that is used. 856 The choice of technical trust mechanism constrains how rules 857 determination is implemented. Regardless of what deployment strategy 858 is chosen, it is important that the technical trust mechanism be able 859 to validate the identities of both parties to the exchange. The 860 trust mechanism must ensure that the entity acting as IdP for a given 861 NAI is permitted to be the IdP for that realm, and that any service 862 name claimed by the RP is permitted to be claimed by that entity. 863 Here are the categories of technical trust determination: 865 AAA Proxy: 866 The simplest model is that an RP is an AAA client and can send the 867 request directly to an AAA proxy. The hop-by-hop integrity 868 protection of the AAA fabric provides technical trust. An RP can 869 submit a request directly to the correct federation. 870 Alternatively, a federation disambiguation fabric can be used. 871 Such a fabric takes information about what federations the RP is 872 part of and what federations the IdP is part of and routes a 873 message to the appropriate federation. The routing of messages 874 across the fabric plus attributes added to requests and responses 875 provides rules determination. For example, when a disambiguation 876 fabric routes a message to a given federation, that federation's 877 rules are chosen. Name validation is enforced as messages travel 878 across the fabric. The entities near the RP confirm its identity 879 and validate names it claims. The fabric routes the message 880 towards the appropriate IdP, validating the name of the IdP in the 881 process. The routing can be statically configured. Alternatively 882 a routing protocol could be developed to exchange reachability 883 information about a given IdP and to apply policy across the AAA 884 fabric. Such a routing protocol could flood naming constraints to 885 the appropriate points in the fabric. 887 Trust Broker: 888 Instead of routing messages through AAA proxies, some trust broker 889 could establish keys between entities near the RP and entities 890 near the IdP. The advantage of this approach is efficiency of 891 message handling. Fewer entities are needed to be involved for 892 each message. Security may be improved by sending individual 893 messages over fewer hops. Rules determination involves decisions 894 made by trust brokers about what keys to grant. Also, associated 895 with each credential is context about rules and about other 896 aspects of technical trust including names that may be claimed. A 897 routing protocol similar to the one for AAA proxies is likely to 898 be useful to trust brokers in flooding rules and naming 899 constraints. 901 Global Credential: 902 A global credential such as a public key and certificate in a 903 public key infrastructure can be used to establish technical 904 trust. A directory or distributed database such as the Domain 905 Name System is used by the RP to discover the endpoint to contact 906 for a given NAI. Either the database or certificates can provide 907 a place to store information about rules determination and naming 908 constraints. Provided that no intermediates are required (or 909 appear to be required) and that the RP and IdP are sufficient to 910 enforce and determine rules, rules determination is reasonably 911 simple. However applying certain rules is likely to be quite 912 complex. For example if multiple sets of rules are possible 913 between an IdP and RP, confirming the correct set is used may be 914 difficult. This is particularly true if intermediates are 915 involved in making the decision. Also, to the extent that 916 directory information needs to be trusted, rules determination may 917 be more complex. 919 Real-world deployments are likely to be mixtures of these basic 920 approaches. For example, it will be quite common for an RP to route 921 traffic to a AAA proxy within an organization. That proxy could then 922 use any of the three methods to get closer to the IdP. It is also 923 likely that rather than being directly reachable, the IdP may have a 924 proxy on the edge of its organization. Federations will likely 925 provide a traditional AAA proxy interface even if they also provide 926 another mechanism for increased efficiency or security. 928 2.1.4. AAA Security 930 For the AAA framework there are two different places where security 931 needs to be examined. The first is the security that is in place for 932 the links in the AAA backbone being used. The second is the nodes 933 that the backbone consists of. 935 The default link security for RADIUS is showing its age as it uses 936 MD5 and a shared secret to both obfuscate passwords and to provide 937 integrity on the RADIUS messages. While some EAP methods have 938 designed in the ability to protect the client authentication 939 credentials, the MSK returned from the IDP to the RP is protected 940 only by the RADIUS security. In many environments this is considered 941 to be insufficient, especially as not all attributes are obfuscated 942 and can thus leak information to a passive eavesdropper. The use of 943 RADIUS with TLS [RFC6614] and/or DTLS [I-D.ietf-radext-dtls] 944 addresses these attacks. The same level of security is included in 945 the base Diameter specifications. 947 2.1.5. SAML Assertions 949 For the traditional use of AAA frameworks, network access, the only 950 requirement that was necessary to grant access was an affirmative 951 response from the IdP. In the ABFAB world, the RP may need to get 952 additional information about the client before granting access. 953 ABFAB therefore has a requirement that it can transport an arbitrary 954 set of attributes about the client from the IdP to the RP. 956 Security Assertions Markup Language (SAML) [OASIS.saml-core-2.0-os] 957 was designed in order to carry an extensible set of attributes about 958 a subject. Since SAML is extensible in the attribute space, ABFAB 959 has no immediate needs to update the core SAML specifications for our 960 work. It will be necessary to update IdPs that need to return SAML 961 assertions to RPs and for both the IdP and the RP to implement a new 962 SAML profile designed to carry SAML assertions in AAA. The new 963 profile can be found in RFCXXXX [I-D.ietf-abfab-aaa-saml]. As SAML 964 statements will frequently be large, RADIUS servers and clients that 965 deal with SAML statements will need to implement RFC XXXX 966 [I-D.ietf-radext-radius-fragmentation] 968 There are several issues that need to be highlighted: 970 o The security of SAML assertions. 972 o Namespaces and mapping of SAML attributes. 974 o Subject naming of entities. 976 o Making multiple queries about the subject(s). 978 o Level of Assurance for authentication. 980 SAML assertions have an optional signature that can be used to 981 protect and provide origination of the assertion. These signatures 982 are normally based on asymmetric key operations and require that the 983 verifier be able to check not only the cryptographic operation, but 984 also the binding of the originators name and the public key. In a 985 federated environment it will not always be possible for the RP to 986 validate the binding, for this reason the technical trust established 987 in the federation is used as an alternate method of validating the 988 origination and integrity of the SAML Assertion. 990 Attributes placed in SAML assertions can have different namespaces 991 assigned to the same name. In many, but not all, cases the 992 federation agreements will determine what attributes can be used in a 993 SAML statement. This means that the RP needs to map from the 994 federation names, types and semantics into the ones that the policies 995 of the RP are written in. In other cases the federation substrate 996 may modify the SAML assertions in transit to do the necessary 997 namespace, naming and semantic mappings as the assertion crosses the 998 different boundaries in the federation. If the proxies are modifying 999 the SAML Assertion, then they will obviously remove any signatures as 1000 they would no longer validate. In this case the technical trust is 1001 the required mechanism for validating the integrity of the assertion. 1002 Finally, the attributes may still be in the namespace of the 1003 originating IdP. When this occurs the RP will need to get the 1004 required mapping operations from the federation agreements and do the 1005 appropriate mappings itself. 1007 The RADIUS SAML RFC [I-D.ietf-abfab-aaa-saml] has defined a new SAML 1008 name format that corresponds to the NAI name form defined by RFC XXXX 1009 [I-D.ietf-radext-nai]. This allows for easy name matching in many 1010 cases as the name form in the SAML statement and the name form used 1011 in RADIUS or Diameter will be the same. In addition to the NAI name 1012 form, the document also defines a pair of implicit name forms 1013 corresponding to the Client and the Client's machine. These implicit 1014 name forms are based on the Identity-Type enumeration defined in TEAP 1015 [I-D.ietf-emu-eap-tunnel-method]. If the name form returned in a 1016 SAML statement is not based on the NAI, then it is a requirement on 1017 the EAP server that it validate that the subject of the SAML 1018 assertion, if any, is equivalent to the subject identified by the NAI 1019 used in the RADIUS or Diameter session. 1021 RADIUS has the ability to deal with multiple SAML queries for those 1022 EAP Servers which follow RFC 5080 [RFC5080]. In this case a State 1023 attribute will always be returned with the Access-Accept. The EAP 1024 client can then send a new Access-Request with the State attribute 1025 and the new SAML Request Multiple SAML queries can then be done by 1026 making a new Access-Request using the State attribute returned in the 1027 last Access-Accept to link together the different RADIUS sessions. 1029 Some RPs need to ensure that specific criteria are met during the 1030 authentication process. This need is met by using Levels of 1031 Assurance. The way a Level of Assurance is communicated to the RP 1032 from the EAP server is by the use of a SAML Authentication Request 1033 using the Authentication Profile from RFC XXX 1034 [I-D.ietf-abfab-aaa-saml] When crossing boundaries between different 1035 federations, either the policy specified will need to be shared 1036 between the two federations, the policy will need to be mapped by the 1037 proxy server on the boundary or the proxy server on the boundary will 1038 need to supply information the EAP server so that it can do the 1039 required mapping. If this mapping is not done, then the EAP server 1040 will not be able to enforce the desired Level of Assurance as it will 1041 not understand the policy requirements. 1043 2.2. Client To Identity Provider 1045 Looking at the communications between the client and the IdP, the 1046 following items need to be dealt with: 1048 o The client and the IdP need to mutually authenticate each other. 1050 o The client and the IdP need to mutually agree on the identity of 1051 the RP. 1053 ABFAB selected EAP for the purposes of mutual authentication and 1054 assisted in creating some new EAP channel binding documents for 1055 dealing with determining the identity of the RP. A framework for the 1056 channel binding mechanism has been defined in RFC 6677 [RFC6677] that 1057 allows the IdP to check the identity of the RP provided by the AAA 1058 framework with that provided by the client. 1060 2.2.1. Extensible Authentication Protocol (EAP) 1062 Traditional web federation does not describe how a client interacts 1063 with an identity provider for authentication. As a result, this 1064 communication is not standardized. There are several disadvantages 1065 to this approach. Since the communication is not standardized, it is 1066 difficult for machines to correctly enter their credentials with 1067 different authentications, where Individuals can correctly identify 1068 the entire mechanism on the fly. The use of browsers for 1069 authentication restricts the deployment of more secure forms of 1070 authentication beyond plaintext username and password known by the 1071 server. In a number of cases the authentication interface may be 1072 presented before the client has adequately validated they are talking 1073 to the intended server. By giving control of the authentication 1074 interface to a potential attacker, the security of the system may be 1075 reduced and phishing opportunities introduced. 1077 As a result, it is desirable to choose some standardized approach for 1078 communication between the client's end-host and the identity 1079 provider. There are a number of requirements this approach must 1080 meet. 1082 Experience has taught us one key security and scalability 1083 requirement: it is important that the relying party not get 1084 possession of the long-term secret of the client. Aside from a 1085 valuable secret being exposed, a synchronization problem can develop 1086 when the client changes keys with the IdP. 1088 Since there is no single authentication mechanism that will be used 1089 everywhere there is another associated requirement: The 1090 authentication framework must allow for the flexible integration of 1091 authentication mechanisms. For instance, some IdPs require hardware 1092 tokens while others use passwords. A service provider wants to 1093 provide support for both authentication methods, and other methods 1094 from IdPs not yet seen. 1096 These requirements can be met by utilizing standardized and 1097 successfully deployed technology, namely by the Extensible 1098 Authentication Protocol (EAP) framework [RFC3748]. Figure 2 1099 illustrates the integration graphically. 1101 EAP is an end-to-end framework; it provides for two-way communication 1102 between a peer (i.e. client or individual) through the EAP 1103 authenticator (i.e., relying party) to the back-end (i.e., identity 1104 provider). Conveniently, this is precisely the communication path 1105 that is needed for federated identity. Although EAP support is 1106 already integrated in AAA systems (see [RFC3579] and [RFC4072]) 1107 several challenges remain: 1109 o The first is how to carry EAP payloads from the end host to the 1110 relying party. 1112 o Another is to verify statements the relying party has made to the 1113 client, confirm these statements are consistent with statements 1114 made to the identity provider and confirm all of the above are 1115 consistent with the federation and any federation-specific policy 1116 or configuration. 1118 o Another challenge is choosing which identity provider to use for 1119 which service. 1121 The EAP method used for ABFAB needs to meet the following 1122 requirements: 1124 o It needs to provide mutual authentication of the client and IdP. 1126 o It needs to support channel binding. 1128 As of this writing, the only EAP method that meets these criteria is 1129 TEAP [I-D.ietf-emu-eap-tunnel-method] either alone (if client 1130 certificates are used) or with an inner EAP method that does mutual 1131 authentication. 1133 2.2.2. EAP Channel Binding 1135 EAP channel binding is easily confused with a facility in GSS-API 1136 also called channel binding. GSS-API channel binding provides 1137 protection against man-in-the-middle attacks when GSS-API is used as 1138 authentication inside some tunnel; it is similar to a facility called 1139 cryptographic binding in EAP. See [RFC5056] for a discussion of the 1140 differences between these two facilities. 1142 The client knows, in theory, the name of the RP that it attempted to 1143 connect to, however in the event that an attacker has intercepted the 1144 protocol, the client and the IdP need to be able to detect this 1145 situation. A general overview of the problem along with a 1146 recommended way to deal with the channel binding issues can be found 1147 in RFC 6677 [RFC6677]. 1149 Since that document was published, a number of possible attacks were 1150 found and methods to address these attacks have been outlined in 1151 [RFC7029]. 1153 2.3. Client to Relying Party 1155 The final set of interactions between the parties to consider are 1156 those between the client and the RP. In some ways this is the most 1157 complex set since at least part of it is outside the scope of the 1158 ABFAB work. The interactions between these parties include: 1160 o Running the protocol that implements the service that is provided 1161 by the RP and desired by the client. 1163 o Authenticating the client to the RP and the RP to the client. 1165 o Providing the necessary security services to the service protocol 1166 that it needs beyond authentication. 1168 o Deal with client re-authentication where desired. 1170 2.3.1. GSS-API 1172 One of the remaining layers is responsible for integration of 1173 federated authentication into the application. There are a number of 1174 approaches that applications have adopted for security. So, there 1175 may need to be multiple strategies for integration of federated 1176 authentication into applications. However, we have started with a 1177 strategy that provides integration to a large number of application 1178 protocols. 1180 Many applications such as SSH [RFC4462], NFS [RFC2203], DNS [RFC3645] 1181 and several non-IETF applications support the Generic Security 1182 Services Application Programming Interface [RFC2743]. Many 1183 applications such as IMAP, SMTP, XMPP and LDAP support the Simple 1184 Authentication and Security Layer (SASL) [RFC4422] framework. These 1185 two approaches work together nicely: by creating a GSS-API mechanism, 1186 SASL integration is also addressed. In effect, using a GSS-API 1187 mechanism with SASL simply requires placing some headers on the front 1188 of the mechanism and constraining certain GSS-API options. 1190 GSS-API is specified in terms of an abstract set of operations which 1191 can be mapped into a programming language to form an API. When 1192 people are first introduced to GSS-API, they focus on it as an API. 1193 However, from the prospective of authentication for non-web 1194 applications, GSS-API should be thought of as a protocol as well as 1195 an API. When looked at as a protocol, it consists of abstract 1196 operations such as the initial context exchange, which includes two 1197 sub-operations (gss_init_sec_context and gss_accept_sec_context). An 1198 application defines which abstract operations it is going to use and 1199 where messages produced by these operations fit into the application 1200 architecture. A GSS-API mechanism will define what actual protocol 1201 messages result from that abstract message for a given abstract 1202 operation. So, since this work is focusing on a particular GSS-API 1203 mechanism, we generally focus on protocol elements rather than the 1204 API view of GSS-API. 1206 The API view of GSS-API does have significant value as well, since 1207 the abstract operations are well defined, the set of information that 1208 a mechanism gets from the application is well defined. Also, the set 1209 of assumptions the application is permitted to make is generally well 1210 defined. As a result, an application protocol that supports GSS-API 1211 or SASL is very likely to be usable with a new approach to 1212 authentication including this one with no required modifications. In 1213 some cases, support for a new authentication mechanism has been added 1214 using plugin interfaces to applications without the application being 1215 modified at all. Even when modifications are required, they can 1216 often be limited to supporting a new naming and authorization model. 1217 For example, this work focuses on privacy; an application that 1218 assumes it will always obtain an identifier for the client will need 1219 to be modified to support anonymity, unlinkability or pseudonymity. 1221 So, we use GSS-API and SASL because a number of the application 1222 protocols we wish to federate support these strategies for security 1223 integration. What does this mean from a protocol standpoint and how 1224 does this relate to other layers? This means we need to design a 1225 concrete GSS-API mechanism. We have chosen to use a GSS-API 1226 mechanism that encapsulates EAP authentication. So, GSS-API (and 1227 SASL) encapsulates EAP between the end-host and the service. The AAA 1228 framework encapsulates EAP between the relying party and the identity 1229 provider. The GSS-API mechanism includes rules about how initiators 1230 and services are named as well as per-message security and other 1231 facilities required by the applications we wish to support. 1233 2.3.2. Protocol Transport 1235 The transport of data between the client and the relying party is not 1236 provided by GSS-API. GSS-API creates and consumes messages, but it 1237 does not provide the transport itself, instead the protocol using 1238 GSS-API needs to provide the transport. In many cases HTTP or HTTPS 1239 is used for this transport, but other transports are perfectly 1240 acceptable. The core GSS-API document [RFC2743] provides some 1241 details on what requirements exist. 1243 In addition we highlight the following: 1245 o The transport does not need to provide either privacy or 1246 integrity. After GSS-EAP has finished negotiation, GSS-API can be 1247 used to provide both services. If the negotiation process itself 1248 needs protection from eavesdroppers then the transport would need 1249 to provide the necessary services. 1251 o The transport needs to provide reliable transport of the messages. 1253 o The transport needs to ensure that tokens are delivered in order 1254 during the negotiation process. 1256 o GSS-API messages need to be delivered atomically. If the 1257 transport breaks up a message it must also reassemble the message 1258 before delivery. 1260 2.3.3. Reauthentication 1262 There are circumstances where the server will want to have the client 1263 reauthenticate itself. These include very long sessions, where the 1264 original authentication is time limited or cases where in order to 1265 complete an operation a different authentication is required. GSS- 1266 EAP does not have any mechanism for the server to initiate a 1267 reauthentication as all authentication operations start from the 1268 client. If a protocol using GSS-EAP needs to support 1269 reauthentication that is initiated by the server, then a request from 1270 the server to the client for the reauthentiction to start needs to be 1271 placed in the protocol. 1273 Clients can re-use the existing secure connection established by GSS- 1274 API to run the new authentication in by calling GSS_Init_sec_context. 1275 At this point a full reauthentication will be done. 1277 3. Application Security Services 1279 One of the key goals is to integrate federated authentication into 1280 existing application protocols and where possible, existing 1281 implementations of these protocols. Another goal is to perform this 1282 integration while meeting the best security practices of the 1283 technologies used to perform the integration. This section describes 1284 security services and properties required by the EAP GSS-API 1285 mechanism in order to meet these goals. This information could be 1286 viewed as specific to that mechanism. However, other future 1287 application integration strategies are very likely to need similar 1288 services. So, it is likely that these services will be expanded 1289 across application integration strategies if new application 1290 integration strategies are adopted. 1292 3.1. Authentication 1294 GSS-API provides an optional security service called mutual 1295 authentication. This service means that in addition to the initiator 1296 providing (potentially anonymous or pseudonymous) identity to the 1297 acceptor, the acceptor confirms its identity to the initiator. 1298 Especially for the ABFAB context, this service is confusingly named. 1299 We still say that mutual authentication is provided when the identity 1300 of an acceptor is strongly authenticated to an anonymous initiator. 1302 RFC 2743, unfortunately, does not explicitly talk about what mutual 1303 authentication means. Within this document we therefore define it 1304 as: 1306 o If a target name is configured for the initiator, then the 1307 initiator trusts that the supplied target name describes the 1308 acceptor. This implies both that appropriate cryptographic 1309 exchanges took place for the initiator to make such a trust 1310 decision, and that after evaluating the results of these 1311 exchanges, the initiator's policy trusts that the target name is 1312 accurate. 1314 o If no target name is configured for the initiator, then the 1315 initiator trusts that the acceptor name, supplied by the acceptor, 1316 correctly names the entity it is communicating with. 1318 o Both the initiator and acceptor have the same key material for 1319 per-message keys and both parties have confirmed they actually 1320 have the key material. In EAP terms, there is a protected 1321 indication of success. 1323 Mutual authentication is an important defense against certain aspects 1324 of phishing. Intuitively, clients would like to assume that if some 1325 party asks for their credentials as part of authentication, 1326 successfully gaining access to the resource means that they are 1327 talking to the expected party. Without mutual authentication, the 1328 server could "grant access" regardless of what credentials are 1329 supplied. Mutual authentication better matches this user intuition. 1331 It is important, therefore, that the GSS-EAP mechanism implement 1332 mutual authentication. That is, an initiator needs to be able to 1333 request mutual authentication. When mutual authentication is 1334 requested, only EAP methods capable of providing the necessary 1335 service can be used, and appropriate steps need to be taken to 1336 provide mutual authentication. While a broader set of EAP methods 1337 could be supported by not requiring mutual authentication, it was 1338 decided that the client needs to always have the ability to request 1339 it. In some cases the IdP and the RP will not support mutual 1340 authentication, however the client will always be able to detect this 1341 and make an appropriate security decision. 1343 The AAA infrastructure may hide the initiator's identity from the 1344 GSS-API acceptor, providing anonymity between the initiator and the 1345 acceptor. At this time, whether the identity is disclosed is 1346 determined by EAP server policy rather than by an indication from the 1347 initiator. Also, initiators are unlikely to be able to determine 1348 whether anonymous communication will be provided. For this reason, 1349 initiators are unlikely to set the anonymous return flag from 1350 GSS_Init_Sec_context. 1352 3.2. GSS-API Channel Binding 1354 [RFC5056] defines a concept of channel binding which is used prevent 1355 man-in-the-middle attacks. The channel binding works by taking a 1356 cryptographic value from the transport security and checks that both 1357 sides of the GSS-API conversation know this value. Transport Layer 1358 Security (TLS) is the most common transport security layer used for 1359 this purpose. 1361 It needs to be stressed that RFC 5056 channel binding (also called 1362 GSS-API channel binding when GSS-API is involved) is not the same 1363 thing as EAP channel binding. GSS-API channel binding is used for 1364 detecting Man-In-The-Middle attacks. EAP channel binding is used for 1365 mutual authentication and acceptor naming checks. Details are 1366 discussed in the mechanisms specification [I-D.ietf-abfab-gss-eap]. 1367 A fuller description of the differences between the facilities can be 1368 found in RFC 5056 [RFC5056]. 1370 The use of TLS can provide both encryption and integrity on the 1371 channel. It is common to provide SASL and GSS-API with these other 1372 security services. 1374 One of the benefits that the use of TLS provides, is that client has 1375 the ability to validate the name of the server. However this 1376 validation is predicated on a couple of things. The TLS sessions 1377 needs to be using certificates and not be an anonymous session. The 1378 client and the TLS server need to share a common trust point for the 1379 certificate used in validating the server. TLS provides its own 1380 server authentication. However there are a variety of situations 1381 where this authentication is not checked for policy or usability 1382 reasons. When the TLS authentication is checked, if the trust 1383 infrastructure behind the TLS authentication is different from the 1384 trust infrastructure behind the GSS-API mutual authentication then 1385 confirming the end-points using both trust infrastructures is likely 1386 to enhance security. If the endpoints of the GSS-API authentication 1387 are different than the endpoints of the lower layer, this is a strong 1388 indication of a problem such as a man-in-the-middle attack. Channel 1389 binding provides a facility to determine whether these endpoints are 1390 the same. 1392 The GSS-EAP mechanism needs to support channel binding. When an 1393 application provides channel binding data, the mechanism needs to 1394 confirm this is the same on both sides consistent with the GSS-API 1395 specification. 1397 3.3. Host-Based Service Names 1399 IETF security mechanisms typically take a host name and perhaps a 1400 service, entered by a user, and make some trust decision about 1401 whether the remote party in the interaction is the intended party. 1402 This decision can be made by the use of certificates, pre-configured 1403 key information or a previous leap of trust. GSS-API has defined a 1404 relatively flexible name convention, however most of the IETF 1405 applications that use GSS-API (including SSH, NFS, IMAP, LDAP and 1406 XMPP) have chosen to use a more restricted naming convention based on 1407 the host name. The GSS-EAP mechanism needs to support host-based 1408 service names in order to work with existing IETF protocols. 1410 The use of host-based service names leads to a challenging trust 1411 delegation problem. Who is allowed to decide whether a particular 1412 host name maps to a specific entity? Possible solutions to this 1413 problem have been looked at. 1415 o The public-key infrastructure (PKI) used by the web has chosen to 1416 have a number of trust anchors (root certificate authorities) each 1417 of which can map any host name to a public key. 1419 o A number of GSS-API mechanisms, such as Kerberos [RFC1964], have 1420 split the problem into two parts. A new concept called a realm is 1421 introduced, the realm is responsible for host mapping within that 1422 realm. The mechanism then decides what realm is responsible for a 1423 given name. This is the approach adopted by ABFAB. 1425 GSS-EAP defines a host naming convention that takes into account the 1426 host name, the realm, the service and the service parameters. An 1427 example of GSS-API service name is "xmpp/foo@example.com". This 1428 identifies the XMPP service on the host foo in the realm example.com. 1429 Any of the components, except for the service name may be omitted 1430 from a name. When omitted, then a local default would be used for 1431 that component of the name. 1433 While there is no requirement that realm names map to Fully Qualified 1434 Domain Names (FQDN) within DNS, in practice this is normally true. 1435 Doing so allows for the realm portion of service names and the 1436 portion of NAIs to be the same. It also allows for the use of DNS in 1437 locating the host of a service while establishing the transport 1438 channel between the client and the relying party. 1440 It is the responsibility of the application to determine the server 1441 that it is going to communicate with; GSS-API has the ability to help 1442 confirm that the server is the desired server but not to determine 1443 the name of the server to use. It is also the responsibility of the 1444 application to determine how much of the information identifying the 1445 service needs to be validated by the ABFAB system. The information 1446 that needs to be validated is used to build up the service name 1447 passed into the GSS-EAP mechanism. What information is to be 1448 validated will depend on both what information was provided by the 1449 client, and what information is considered significant. If the 1450 client only cares about getting a specific service, then the host and 1451 realm that provides the service does not need to be validated. 1453 Applications may retrieve information about providers of services 1454 from DNS. Service Records (SRV) and Naming Authority Pointer (NAPTR) 1455 records are used to help find a host that provides a service; however 1456 the necessity of having DNSSEC on the queries depends on how the 1457 information is going to be used. If the host name returned is not 1458 going to be validated by EAP channel binding, because only the 1459 service is being validated, then DNSSEC is not required. However, if 1460 the host name is going to be validated by EAP channel binding then 1461 DNSSEC needs to be use to ensure that the correct host name is 1462 validated. In general, if the information that is returned from the 1463 DNS query is to be validated, then it needs to be obtained in a 1464 secure manner. 1466 Another issue that needs to be addressed for host-based service names 1467 is that they do not work ideally when different instances of a 1468 service are running on different ports. If the services are 1469 equivalent, then it does not matter. However if there are 1470 substantial differences in the quality of the service that 1471 information needs to be part of the validation process. If one has 1472 just a host name and not a port in the information being validated, 1473 then this is not going to be a successful strategy. 1475 3.4. Additional GSS-API Services 1477 GSS-API provides per-message security services that can provide 1478 confidentiality and/or integrity. Some IETF protocols such as NFS 1479 and SSH take advantage of these services. As a result GSS-EAP needs 1480 to support these services. As with mutual authentication, per- 1481 message services will limit the set of EAP methods that can be used 1482 to those that generate a Master Session Key (MSK). Any EAP method 1483 that produces an MSK is able to support per-message security services 1484 described in [RFC2743]. 1486 GSS-API provides a pseudo-random function. This function generates a 1487 pseudo-random sequence using the shared session key as the seed for 1488 the bytes generated. This provides an algorithm that both the 1489 initiator and acceptor can run in order to arrive at the same key 1490 value. The use of this feature allows for an application to generate 1491 keys or other shared secrets for use in other places in the protocol. 1492 In this regards, it is similar in concept to the TLS extractor (RFC 1493 5705 [RFC5705].). While no current IETF protocols require this, non- 1494 IETF protocols are expected to take advantage of this in the near 1495 future. Additionally, a number of protocols have found the TLS 1496 extractor to be useful in this regards so it is highly probable that 1497 IETF protocols may also start using this feature. 1499 4. Privacy Considerations 1501 ABFAB, as an architecture designed to enable federated authentication 1502 and allow for the secure transmission of identity information between 1503 entities, obviously requires careful consideration around privacy and 1504 the potential for privacy violations. 1506 This section examines the privacy related information presented in 1507 this document, summarizing the entities that are involved in ABFAB 1508 communications and what exposure they have to identity information. 1509 In discussing these privacy considerations in this section, we use 1510 terminology and ideas from [RFC6973]. 1512 Note that the ABFAB architecture uses at its core several existing 1513 technologies and protocols; detailed privacy discussion around these 1514 is not examined. This section instead focuses on privacy 1515 considerations specifically related to overall architecture and usage 1516 of ABFAB. 1518 +--------+ +---------------+ +--------------+ 1519 | Client | <---> | RP | <---> | AAA Client | 1520 +--------+ +---------------+ +--------------+ 1521 ^ 1522 | 1523 v 1524 +---------------+ +--------------+ 1525 | SAML Server | | AAA Proxy(s) | 1526 +---------------+ +--------------+ 1527 ^ ^ 1528 | | 1529 v v 1530 +------------+ +---------------+ +--------------+ 1531 | EAP Server | <---> | IdP | <---> | AAA Server | 1532 +------------+ +---------------+ +--------------+ 1534 Figure 3: Entities and Data Flow 1536 4.1. Entities and their roles 1538 Categorizing the ABFAB entities shown in the Figure 3 according to 1539 the taxonomy of terms from [RFC6973] the entities shown in Figure 3 1540 is somewhat complicated as during the various phases of ABFAB 1541 communications the roles of each entity changes. The three main 1542 phases of relevance are the Client to RP communication phase, the 1543 Client to IdP (via the Federation Substrate) phase, and the IdP to RP 1544 (via the Federation Substrate) phase. 1546 In the Client to RP communication phase, we have: 1548 Initiator: Client. 1550 Observers: Client, RP. 1552 Recipient: RP. 1554 In the Client to IdP (via the Federation Substrate) communication 1555 phase, we have: 1557 Initiator: Client. 1559 Observers: Client, RP, AAA Client, AAA Proxy(s), AAA Server, IdP. 1561 Recipient: IdP 1563 In the IdP to Relying party (via the Federation Substrate) 1564 communication phase, we have: 1566 Initiator: RP. 1568 Observers: IdP, AAA Server, AAA Proxy(s), AAA Client, RP. 1570 Recipient: IdP 1572 Eavesdroppers and Attackers can reside on any communication link 1573 between entities in Figure 3. 1575 The Federation Substrate consists of all of the AAA entities. In 1576 some cases the AAA Proxies entities may not exist as the AAA Client 1577 can talk directly to the AAA Server. Specifications such as the 1578 Trust Router Protocol and RADIUS dynamic discovery 1579 [I-D.ietf-radext-dynamic-discovery] can be used to shorten the path 1580 between the AAA client and the AAA server (and thus stop these AAA 1581 Proxies from being Observers); however even in these circumstances 1582 there may be AAA Proxies in the path. 1584 In Figure 3 the IdP has been divided into multiple logical pieces, in 1585 actual implementations these pieces will frequently be tightly 1586 coupled. The links between these pieces provide the greatest 1587 opportunity for attackers and eavesdroppers to acquire information, 1588 however, as they are all under the control of a single entity they 1589 are also the easiest to have tightly secured. 1591 4.2. Privacy Aspects of ABFAB Communication Flows 1593 In the ABFAB architecture, there are a few different types of data 1594 and identifiers in use. The best way to understand them, and the 1595 potential privacy impacts of them, is to look at each phase of 1596 communication in ABFAB. 1598 4.2.1. Client to RP 1600 The flow of data between the client and the RP is divided into two 1601 parts. The first part consists of all of the data exchanged as part 1602 of the ABFAB authentication process. The second part consists of all 1603 of the data exchanged after the authentication process has been 1604 finished. 1606 During the initial communications phase, the client sends an NAI (see 1607 [I-D.ietf-radext-nai]) to the RP. Many EAP methods (but not all) 1608 allow for the client to disclose an NAI to RP the in a form that 1609 includes only a realm component during this communications phase. 1610 This is the minimum amount of identity information necessary for 1611 ABFAB to work - it indicates an IdP that the principal has a 1612 relationship with. EAP methods that do not allow this will 1613 necessarily also reveal an identifier for the principal in the IdP 1614 realm (e.g. a username). 1616 The data shared during the initial communication phase may be 1617 protected by a channel protocol such as TLS. This will prevent the 1618 leak of information to passive eavesdroppers, however an active 1619 attacker may still be able to setup as a man-in-the-middle. The 1620 client may not be able to validate the certificates (if any) provided 1621 by the service, deferring the check of the identity of the RP until 1622 the completion of the ABFAB authentication protocol (i.e., using EAP 1623 channel binding). 1625 The data exchanged after the authentication process can have privacy 1626 and authentication using the GSS-API services. If the overall 1627 application protocol allows for the process of re-authentication, 1628 then the same privacy implications as discussed in previous 1629 paragraphs apply. 1631 4.2.2. Client to IdP (via Federation Substrate) 1633 This phase sees a secure TLS tunnel initiated between the Client and 1634 the IdP via the RP and federation substrate. The process is 1635 initiated by the RP using the realm information given to it by the 1636 client. Once set up, the tunnel is used to send credentials to IdP 1637 to authenticate. 1639 Various operational information is transported between RP and IdP, 1640 over the AAA infrastructure, for example using RADIUS headers. As no 1641 end-to-end security is provided by AAA, all AAA entities on the path 1642 between the RP and IdP have the ability to eavesdrop on this 1643 information unless additional security measures are taken (such as 1644 the use of TLS for RADIUS [I-D.ietf-radext-dtls]). Some of this 1645 information may form identifiers or explicit identity information: 1647 o The Relying Party knows the IP address of the Client. It is 1648 possible that the Relying Party could choose to expose this IP 1649 address by including it in a RADIUS header such as Calling Station 1650 ID. This is a privacy consideration to take into account of the 1651 application protocol. 1653 o The EAP MSK is transported between the IdP and the RP over the AAA 1654 infrastructure, for example through RADIUS headers. This is a 1655 particularly important privacy consideration, as any AAA Proxy 1656 that has access to the EAP MSK is able to decrypt and eavesdrop on 1657 any traffic encrypted using that EAP MSK (i.e., all communications 1658 between the Client and RP). This problem can be mitigted by the 1659 application protocol setting up a secure tunnel between the Client 1660 and the RP and performing a cryptographic binding between the 1661 tunnel and EAP MSK. 1663 o Related to the above, the AAA server has access to the material 1664 necessary to derive the session key, thus the AAA server can 1665 observe any traffic encrypted between the Client and RP. This 1666 "feature" was" chosen as a simplification and to make performance 1667 faster; if it was decided that this trade-off was not desirable 1668 for privacy and security reasons, then extensions to ABFAB that 1669 make use of techniques such as Diffie-Helman key exchange would 1670 mitigate against this. 1672 The choice of EAP method used has other potential privacy 1673 implications. For example, if the EAP method in use does not support 1674 trust anchors to enable mutual authentication, then there are no 1675 guarantees that the IdP is who it claims to be, and thus the full NAI 1676 including a username and a realm might be sent to any entity 1677 masquerading as a particular IdP. 1679 Note that ABFAB has not specified any AAA accounting requirements. 1680 Implementations that use the accounting portion of AAA should 1681 consider privacy appropriately when designing this aspect. 1683 4.2.3. IdP to RP (via Federation Substrate) 1685 In this phase, the IdP communicates with the RP informing it as to 1686 the success or failure of authentication of the user, and optionally, 1687 the sending of identity information about the principal. 1689 As in the previous flow (Client to IdP), various operation 1690 information is transported between IdP and RP over the AAA 1691 infrastructure, and the same privacy considerations apply. However, 1692 in this flow, explicit identity information about the authenticated 1693 principal can be sent from the IdP to the RP. This information can 1694 be sent through RADIUS headers, or using SAML 1695 [I-D.ietf-abfab-aaa-saml]. This can include protocol specific 1696 identifiers, such as SAML NameIDs, as well as arbitrary attribute 1697 information about the principal. What information will be released 1698 is controlled by policy on the Identity Provider. As before, when 1699 sending this through RADIUS headers, all AAA entities on the path 1700 between the RP and IdP have the ability to eavesdrop unless 1701 additional security measures are taken (such as the use of TLS for 1702 RADIUS [I-D.ietf-radext-dtls]). When sending this using SAML, as 1703 specified in [I-D.ietf-abfab-aaa-saml], confidentiality of the 1704 information should however be guaranteed as [I-D.ietf-abfab-aaa-saml] 1705 requires the use of TLS for RADIUS. 1707 4.3. Relationship between User and Entities 1709 o Between User and IdP - the IdP is an entity the user will have a 1710 direct relationship with, created when the organization that 1711 operates the entity provisioned and exchanged the user's 1712 credentials. Privacy and data protection guarantees may form a 1713 part of this relationship. 1715 o Between User and RP - the RP is an entity the user may or may not 1716 have a direct relationship with, depending on the service in 1717 question. Some services may only be offered to those users where 1718 such a direct relationship exists (for particularly sensitive 1719 services, for example), while some may not require this and would 1720 instead be satisfied with basic federation trust guarantees 1721 between themselves and the IdP). This may well include the option 1722 that the user stays anonymous with respect to the RP (though 1723 obviously never to the IdP). If attempting to preserve privacy 1724 through the mitigation of data minimization, then the only 1725 attribute information about individuals exposed to the RP should 1726 be that which is strictly necessary for the operation of the 1727 service. 1729 o Between User and Federation substrate - the user is highly likely 1730 to have no knowledge of, or relationship with, any entities 1731 involved with the federation substrate (not that the IdP and/or RP 1732 may, however). Knowledge of attribute information about 1733 individuals for these entities is not necessary, and thus such 1734 information should be protected in such a way as to prevent access 1735 to this information from being possible. 1737 4.4. Accounting Information 1739 Alongside the core authentication and authorization that occurs in 1740 AAA communications, accounting information about resource consumption 1741 may be delivered as part of the accounting exchange during the 1742 lifetime of the granted application session. 1744 4.5. Collection and retention of data and identifiers 1746 In cases where Relying Parties are not required to identify a 1747 particular individual when an individual wishes to make use of their 1748 service, the ABFAB architecture enables anonymous or pseudonymous 1749 access. Thus data and identifiers other than pseudonyms and 1750 unlinkable attribute information need not be stored and retained. 1752 However, in cases where Relying Parties require the ability to 1753 identify a particular individual (e.g. so they can link this identity 1754 information to a particular account in their service, or where 1755 identity information is required for audit purposes), the service 1756 will need to collect and store such information, and to retain it for 1757 as long as they require. Deprovisioning of such accounts and 1758 information is out of scope for ABFAB, but obviously for privacy 1759 protection any identifiers collected should be deleted when they are 1760 no longer needed. 1762 4.6. User Participation 1764 In the ABFAB architecture, by its very nature users are active 1765 participants in the sharing of their identifiers as they initiate the 1766 communications exchange every time they wish to access a server. 1767 They are, however, not involved in control of the set of information 1768 related to them that transmitted from the IdP to RP for authorization 1769 purposes; rather, this is under the control of policy on the IdP. 1770 Due to the nature of the AAA communication flows, with the current 1771 ABFAB architecture there is no place for a process of gaining user 1772 consent for the information to be released from IdP to RP. 1774 5. Security Considerations 1776 This document describes the architecture for Application Bridging for 1777 Federated Access Beyond Web (ABFAB) and security is therefore the 1778 main focus. This section highlights the main communication channels 1779 and their security properties: 1781 Client-to-RP Channel: 1783 The channel binding material is provided by any certificates and 1784 the final message (i.e., a cryptographic token for the channel). 1785 Authentication may be provided by the RP to the client but a 1786 deployment without authentication at the TLS layer is possible as 1787 well. In addition, there is a channel between the GSS requestor 1788 and the GSS acceptor, but the keying material is provided by a 1789 "third party" to both entities. The client can derive keying 1790 material locally, but the RP gets the material from the IdP. In 1791 the absence of a transport that provides encryption and/or 1792 integrity, the channel between the client and the RP has no 1793 ability to have any cryptographic protection until the EAP 1794 authentication has been completed and the MSK is transferred from 1795 the IdP to the RP. 1797 RP-to-IdP Channel: 1799 The security of this communication channel is mainly provided by 1800 the functionality offered via RADIUS and Diameter. At the time of 1801 writing there are no end-to-end security mechanisms standardized 1802 and thereby the architecture has to rely on hop-by-hop security 1803 with trusted AAA entities or, as an alternative but possible 1804 deployment variant, direct communication between the AAA client to 1805 the AAA server. Note that the authorization result the IdP 1806 provides to the RP in the form of a SAML assertion may; however, 1807 be protected such that the SAML related components are secured 1808 end-to-end. 1810 The MSK is transported from the IdP to the RP over this channel. 1811 As no end-to-end security is provided by AAA, all AAA entities on 1812 the path between the RP and IdP have the ability to eavesdrop if 1813 no additional security measures are taken. One such measure is to 1814 use a transport between the client and the IdP that provides 1815 confidentiality. 1817 Client-to-IdP Channel: 1819 This communication interaction is accomplished with the help of 1820 EAP and EAP methods. The offered security protection will depend 1821 on the EAP method that is chosen but a minimum requirement is to 1822 offer mutual authentication, and key derivation. The IdP is 1823 responsible during this process to determine that the RP that is 1824 communication to the client over the RP-to-IdP channel is the same 1825 one talking to the IdP. This is accomplished via the EAP channel 1826 binding. 1828 Partial list of issues to be addressed in this section: Privacy, 1829 SAML, Trust Anchors, EAP Algorithm Selection, Diameter/RADIUS/AAA 1830 Issues, Naming of Entities, Protection of passwords, Channel Binding, 1831 End-point-connections (TLS), Proxy problems 1833 When a pseudonym is generated as a unique long term identifier for a 1834 client by an IdP, care must be taken in the algorithm that it cannot 1835 easily be reverse engineered by the service provider. If it can be 1836 reversed then the service provider can consult an oracle to determine 1837 if a given unique long term identifier is associated with a different 1838 known identifier. 1840 6. IANA Considerations 1842 This document does not require actions by IANA. 1844 7. Acknowledgments 1846 We would like to thank Mayutan Arumaithurai, Klaas Wierenga and Rhys 1847 Smith for their feedback. Additionally, we would like to thank Eve 1848 Maler, Nicolas Williams, Bob Morgan, Scott Cantor, Jim Fenton, Paul 1849 Leach, and Luke Howard for their feedback on the federation 1850 terminology question. 1852 Furthermore, we would like to thank Klaas Wierenga for his review of 1853 the pre-00 draft version. 1855 8. References 1857 8.1. Normative References 1859 [RFC2743] Linn, J., "Generic Security Service Application Program 1860 Interface Version 2, Update 1", RFC 2743, January 2000. 1862 [RFC2865] Rigney, C., Willens, S., Rubens, A., and W. Simpson, 1863 "Remote Authentication Dial In User Service (RADIUS)", RFC 1864 2865, June 2000. 1866 [RFC3748] Aboba, B., Blunk, L., Vollbrecht, J., Carlson, J., and H. 1867 Levkowetz, "Extensible Authentication Protocol (EAP)", RFC 1868 3748, June 2004. 1870 [RFC3579] Aboba, B. and P. Calhoun, "RADIUS (Remote Authentication 1871 Dial In User Service) Support For Extensible 1872 Authentication Protocol (EAP)", RFC 3579, September 2003. 1874 [RFC4072] Eronen, P., Hiller, T., and G. Zorn, "Diameter Extensible 1875 Authentication Protocol (EAP) Application", RFC 4072, 1876 August 2005. 1878 [I-D.ietf-abfab-gss-eap] 1879 Hartman, S. and J. Howlett, "A GSS-API Mechanism for the 1880 Extensible Authentication Protocol", draft-ietf-abfab-gss- 1881 eap-09 (work in progress), August 2012. 1883 [I-D.ietf-abfab-aaa-saml] 1884 Howlett, J. and S. Hartman, "A RADIUS Attribute, Binding, 1885 Profiles, Name Identifier Format, and Confirmation Methods 1886 for SAML", draft-ietf-abfab-aaa-saml-08 (work in 1887 progress), November 2013. 1889 [I-D.ietf-radext-nai] 1890 DeKok, A., "The Network Access Identifier", draft-ietf- 1891 radext-nai-05 (work in progress), November 2013. 1893 [RFC6677] Hartman, S., Clancy, T., and K. Hoeper, "Channel-Binding 1894 Support for Extensible Authentication Protocol (EAP) 1895 Methods", RFC 6677, July 2012. 1897 8.2. Informative References 1899 [RFC6733] Fajardo, V., Arkko, J., Loughney, J., and G. Zorn, 1900 "Diameter Base Protocol", RFC 6733, October 2012. 1902 [RFC6749] Hardt, D., "The OAuth 2.0 Authorization Framework", RFC 1903 6749, October 2012. 1905 [RFC6973] Cooper, A., Tschofenig, H., Aboba, B., Peterson, J., 1906 Morris, J., Hansen, M., and R. Smith, "Privacy 1907 Considerations for Internet Protocols", RFC 6973, July 1908 2013. 1910 [I-D.ietf-radext-radius-fragmentation] 1911 Perez-Mendez, A., Lopez, R., Pereniguez-Garcia, F., Lopez- 1912 Millan, G., Lopez, D., and A. DeKok, "Support of 1913 fragmentation of RADIUS packets", draft-ietf-radext- 1914 radius-fragmentation-02 (work in progress), November 2013. 1916 [RFC1964] Linn, J., "The Kerberos Version 5 GSS-API Mechanism", RFC 1917 1964, June 1996. 1919 [RFC2203] Eisler, M., Chiu, A., and L. Ling, "RPCSEC_GSS Protocol 1920 Specification", RFC 2203, September 1997. 1922 [RFC3645] Kwan, S., Garg, P., Gilroy, J., Esibov, L., Westhead, J., 1923 and R. Hall, "Generic Security Service Algorithm for 1924 Secret Key Transaction Authentication for DNS (GSS-TSIG)", 1925 RFC 3645, October 2003. 1927 [RFC4462] Hutzelman, J., Salowey, J., Galbraith, J., and V. Welch, 1928 "Generic Security Service Application Program Interface 1929 (GSS-API) Authentication and Key Exchange for the Secure 1930 Shell (SSH) Protocol", RFC 4462, May 2006. 1932 [RFC4422] Melnikov, A. and K. Zeilenga, "Simple Authentication and 1933 Security Layer (SASL)", RFC 4422, June 2006. 1935 [RFC5056] Williams, N., "On the Use of Channel Bindings to Secure 1936 Channels", RFC 5056, November 2007. 1938 [RFC5080] Nelson, D. and A. DeKok, "Common Remote Authentication 1939 Dial In User Service (RADIUS) Implementation Issues and 1940 Suggested Fixes", RFC 5080, December 2007. 1942 [RFC5705] Rescorla, E., "Keying Material Exporters for Transport 1943 Layer Security (TLS)", RFC 5705, March 2010. 1945 [RFC5801] Josefsson, S. and N. Williams, "Using Generic Security 1946 Service Application Program Interface (GSS-API) Mechanisms 1947 in Simple Authentication and Security Layer (SASL): The 1948 GS2 Mechanism Family", RFC 5801, July 2010. 1950 [RFC6614] Winter, S., McCauley, M., Venaas, S., and K. Wierenga, 1951 "Transport Layer Security (TLS) Encryption for RADIUS", 1952 RFC 6614, May 2012. 1954 [OASIS.saml-core-2.0-os] 1955 Cantor, S., Kemp, J., Philpott, R., and E. Maler, 1956 "Assertions and Protocol for the OASIS Security Assertion 1957 Markup Language (SAML) V2.0", OASIS Standard saml- 1958 core-2.0-os, March 2005. 1960 [RFC7029] Hartman, S., Wasserman, M., and D. Zhang, "Extensible 1961 Authentication Protocol (EAP) Mutual Cryptographic 1962 Binding", RFC 7029, October 2013. 1964 [I-D.ietf-emu-eap-tunnel-method] 1965 Zhou, H., Cam-Winget, N., Salowey, J., and S. Hanna, 1966 "Tunnel EAP Method (TEAP) Version 1", draft-ietf-emu-eap- 1967 tunnel-method-09 (work in progress), September 2013. 1969 [I-D.ietf-radext-dtls] 1970 DeKok, A., "DTLS as a Transport Layer for RADIUS", draft- 1971 ietf-radext-dtls-07 (work in progress), October 2013. 1973 [I-D.ietf-radext-dynamic-discovery] 1974 Winter, S. and M. McCauley, "NAI-based Dynamic Peer 1975 Discovery for RADIUS/TLS and RADIUS/DTLS", draft-ietf- 1976 radext-dynamic-discovery-08 (work in progress), October 1977 2013. 1979 [WS-TRUST] 1980 Lawrence, K., Kaler, C., Nadalin, A., Goodner, M., Gudgin, 1981 M., Barbir, A., and H. Granqvist, "WS-Trust 1.4", OASIS 1982 Standard ws-trust-200902, February 2009, 1983 . 1986 [NIST-SP.800-63] 1987 Burr, W., Dodson, D., and W. Polk, "Electronic 1988 Authentication Guideline", NIST Special Publication 1989 800-63, April 2006. 1991 Authors' Addresses 1993 Josh Howlett 1994 JANET(UK) 1995 Lumen House, Library Avenue, Harwell 1996 Oxford OX11 0SG 1997 UK 1999 Phone: +44 1235 822363 2000 Email: Josh.Howlett@ja.net 2002 Sam Hartman 2003 Painless Security 2005 Email: hartmans-ietf@mit.edu 2007 Hannes Tschofenig 2008 Nokia Siemens Networks 2009 Linnoitustie 6 2010 Espoo 02600 2011 Finland 2013 Phone: +358 (50) 4871445 2014 Email: Hannes.Tschofenig@gmx.net 2015 URI: http://www.tschofenig.priv.at 2017 Eliot Lear 2018 Cisco Systems GmbH 2019 Richtistrasse 7 2020 Wallisellen, ZH CH-8304 2021 Switzerland 2023 Phone: +41 44 878 9200 2024 Email: lear@cisco.com 2026 Jim Schaad 2027 Soaring Hawk Consulting 2029 Email: ietf@augustcellars.com