idnits 2.17.1 draft-ietf-mif-mpvd-arch-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** There are 5 instances of too long lines in the document, the longest one being 6 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 206 has weird spacing: '...xchange or...' -- The document date (September 12, 2014) is 3513 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'IEEE802.11u' is defined on line 1030, but no explicit reference was found in the text == Outdated reference: A later version (-05) exists of draft-korhonen-dmm-prefix-properties-03 -- Obsolete informational reference (is this intentional?): RFC 3315 (Obsoleted by RFC 8415) -- Obsolete informational reference (is this intentional?): RFC 5996 (Obsoleted by RFC 7296) -- Obsolete informational reference (is this intentional?): RFC 6555 (Obsoleted by RFC 8305) Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 MIF Working Group D. Anipko, Ed. 3 Internet-Draft Microsoft Corporation 4 Intended status: Informational September 12, 2014 5 Expires: March 14, 2015 7 Multiple Provisioning Domain Architecture 8 draft-ietf-mif-mpvd-arch-04 10 Abstract 12 This document is a product of the work of the MIF Architecture Design 13 team. It outlines a solution framework for some of the issues 14 experienced by nodes that can be attached to multiple networks 15 simultaneously. The framework defines the concept of a Provisioning 16 Domain (PvD) which is a a consistent set of network configuration 17 information. PvD aware nodes learn PvD specific information from the 18 networks they are attached to and / or other sources. PvDs are used 19 to enable separation and configuration consistency in presence of 20 multiple concurrent connections. 22 Status of this Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on March 14, 2015. 39 Copyright Notice 41 Copyright (c) 2014 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents (http://trustee.ietf.org/ 46 license-info) in effect on the date of publication of this document. 47 Please review these documents carefully, as they describe your rights 48 and restrictions with respect to this document. Code Components 49 extracted from this document must include Simplified BSD License text 50 as described in Section 4.e of the Trust Legal Provisions and are 51 provided without warranty as described in the Simplified BSD License. 53 Table of Contents 55 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 56 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 57 2. Definitions and Types of PvDs . . . . . . . . . . . . . . . . 4 58 2.1. Explicit PvDs . . . . . . . . . . . . . . . . . . . . . . 4 59 2.2. Implicit PvDs and Incremental Adoption of Explicit PvDs . 5 60 2.3. Relationship Between PvDs and Interfaces . . . . . . . . . 6 61 2.4. PvD Identity / Naming . . . . . . . . . . . . . . . . . . 6 62 2.5. The Relationship to Dual-Stack Networks . . . . . . . . . 7 63 3. Conveying PvD information using DHCPv6 and Router Advertisement 7 64 3.1. Separate Messages or One Message? . . . . . . . . . . . . 8 65 3.2. Securing PvD Information . . . . . . . . . . . . . . . . . 8 66 3.3. Backward Compatibility . . . . . . . . . . . . . . . . . . 8 67 3.4. Selective Propagation . . . . . . . . . . . . . . . . . . 8 68 3.5. Retracting / Updating PvD Information . . . . . . . . . . 9 69 3.6. Conveying Configuration Information using IKEv2 . . . . . 9 70 4. Example Network Configurations . . . . . . . . . . . . . . . . 10 71 4.1. A Mobile Node . . . . . . . . . . . . . . . . . . . . . . 10 72 4.2. A Node with a VPN Connection . . . . . . . . . . . . . . . 11 73 4.3. A Home Network and a Network Operator with Multiple PvDs . 12 74 5. Reference Model for the PvD-aware Node . . . . . . . . . . . . 12 75 5.1. Constructions and Maintenance of Separate PvDs . . . . . . 13 76 5.2. Consistent use of PvDs for Network Connections . . . . . . 13 77 5.2.1. Name Resolution . . . . . . . . . . . . . . . . . . . 13 78 5.2.2. Next-hop and Source Address Selection . . . . . . . . 14 79 5.2.3. Listening Applications . . . . . . . . . . . . . . . . 15 80 5.2.3.1. Processing of Incoming Traffic . . . . . . . . . . 15 81 5.2.3.1.1. Connection-oriented APIs . . . . . . . . . . . 15 82 5.2.3.1.2. Connectionless APIs . . . . . . . . . . . . . 16 83 5.2.4. Enforcement of Security Policies . . . . . . . . . . . 16 84 5.3. Connectivity Tests . . . . . . . . . . . . . . . . . . . . 16 85 5.4. Relationship to Interface Management and Connection Manage 17 86 6. PvD support in APIs . . . . . . . . . . . . . . . . . . . . . 17 87 6.1. Basic . . . . . . . . . . . . . . . . . . . . . . . . . . 18 88 6.2. Intermediate . . . . . . . . . . . . . . . . . . . . . . . 18 89 6.3. Advanced . . . . . . . . . . . . . . . . . . . . . . . . . 18 90 7. PvD Trust for PvD-Aware Node . . . . . . . . . . . . . . . . . 18 91 7.1. Untrusted PvDs . . . . . . . . . . . . . . . . . . . . . . 18 92 7.2. Trusted PvDs . . . . . . . . . . . . . . . . . . . . . . . 19 93 7.2.1. Authenticated PvDs . . . . . . . . . . . . . . . . . . 19 94 7.2.2. PvDs Trusted by Attachment . . . . . . . . . . . . . . 20 95 8. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 20 96 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 97 10. Security Considerations . . . . . . . . . . . . . . . . . . . 20 98 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 21 99 11.1. Normative References . . . . . . . . . . . . . . . . . . 21 100 11.2. Informative References . . . . . . . . . . . . . . . . . 21 101 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 23 103 1. Introduction 105 Nodes attached to multiple networks may encounter problems from 106 conflicting configuration between the networks, or attempts to 107 simultaneously use more than one network. While various techniques 108 are currently used to tackle these problems ([RFC6419]), in many 109 cases issues may still appear. The MIF problem statement document 110 [RFC6418] describes the general landscape and discusses many of the 111 specific issues and scenario details. 113 Problems, enumerated in [RFC6418], can be grouped into 3 categories: 115 1. Lack of consistent and distinctive management of configuration 116 elements associated with different networks. 118 2. Inappropriate mixed use of configuration elements associated with 119 different networks during a particular network activity or 120 connection. 122 3. Usage of a particular network that is not consistent with the 123 intent of the scenario or involved parties leading to 124 connectivity failure and / or other undesired consequences. 126 An example of (1) is a single, node-scoped list of DNS server IP 127 addresses learned from different networks leading to failures or 128 delays in resolution of names from particular namespaces; an example 129 of (2) is an attempt to resolve the name of an HTTP proxy server 130 learned from network A using a DNS server learned from network B; an 131 example of (3) is the use of an employer-provided VPN connection for 132 peer-to-peer connectivity unrelated to employment activities. 134 This architecture provides solutions to these categories of problems, 135 respectively, by: 137 1. Introducing the formal notion of PvDs, including identity for 138 PvDs, and describing mechanisms for nodes to learn the intended 139 associations between acquired network configuration information 140 elements. 142 2. Introducing a reference model for PvD-aware nodes that prevents 143 the inadvertent mixed use of configuration information which may 144 belong to different PvDs. 146 3. Providing recommendations on PvD selection based on PvD identity 147 and connectivity tests for common scenarios. 149 1.1. Requirements Language 151 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 152 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 153 document are to be interpreted as described in RFC 2119 [RFC2119]. 155 2. Definitions and Types of PvDs 157 Provisioning Domain: 158 A consistent set of network configuration information. 159 Classically, all of the configuration information available on a 160 single interface is provided by a single source (such as a network 161 administrator) and can therefore be treated as a single 162 provisioning domain. In modern IPv6 networks, multihoming can 163 result in more than one provisioning domain being present on a 164 single link. In some scenarios, it is also possible for elements 165 of the same PvD to be present on multiple links. 167 Typical examples of information in a provisioning domain learned 168 from the network are: 170 * Source address prefixes for use by connections within the 171 provisioning domain 173 * IP address(es) of DNS server(s) 175 * Name of HTTP proxy server (if available) 177 * DNS suffixes associated with the network 179 * Default gateway address 181 PvD-aware node: 182 A node that supports the association of network configuration 183 information into PvDs and the use of these PvDs to serve requests 184 for network connections in ways consistent with the 185 recommendations of this architecture. 187 2.1. Explicit PvDs 189 A node may receive explicit information from the network and / or 190 other sources conveying the presence of PvDs and the association of 191 particular network information with a particular PvD. PvDs that are 192 constructed based on such information are referred to as "explicit" 193 in this document. 195 Protocol changes or extensions will likely be required to support 196 explicit PvDs through IETF-defined mechanisms. As an example, one 197 could think of one or more DHCP options carrying PvD identity and / 198 or its elements. 200 A different approach could be the introduction of a DHCP option which 201 only carries the identity of a PvD. Here, the associations between 202 network information elements with the identity is implemented by the 203 respective protocols, for example with a Router Discovery [RFC4861] 204 option associating an address range with a PvD. 206 Another example of a delivery mechanism for PvDs are key exchange or 207 tunneling protocols, such as IKEv2 [RFC5996] that allow the 208 transport of host configuration information. 210 Specific, existing or new features of networking protocols that 211 enable the delivery of PvD identity and association with various 212 network information elements will be defined in companion design 213 documents. 215 Link-specific and / or vendor-proprietary mechanisms for the 216 discovery of PvD information (differing from IETF-defined mechanisms) 217 can be used by nodes either separate from, or in conjunction with, 218 IETF-defined mechanisms; providing they allow the discovery of the 219 necessary elements of the PvD(s). 221 In all cases, nodes must by default ensure that the lifetime of all 222 dynamically discovered PvD configuration is appropriately limited by 223 relevant events. For example, if an interface media state change is 224 indicated, previously discovered information relevant to that 225 interface may no longer be valid and so need to be confirmed or re- 226 discovered. 228 It is expected that the way a node makes use of PvD information is 229 generally independent of the specific mechanism / protocol that the 230 information was received by. 232 In some network topologies, network infrastructure elements may need 233 to advertise multiple PvDs. Generally, the details of how this is 234 performed will be defined in companion design documents. However, 235 where different design choices are possible, the choice that requires 236 a smaller number of packets shall be preferred for efficiency. 238 2.2. Implicit PvDs and Incremental Adoption of Explicit PvDs 240 For some time it is likely that there will be networks which do not 241 advertise explicit PvD information as the deployment of new features 242 in networking protocols is a relatively slow process. 244 When connected to networks which don't advertise explicit PvD 245 information, a PvD-aware node shall automatically create separate 246 PvDs for received configuration. Such PvDs are referred to in this 247 document as "implicit". 249 Through the use of implicit PvDs, PvD-aware nodes may still provide 250 benefits to their users (when compared to non-PvD aware nodes) by 251 following the best practices described in Section 5, using the 252 network information from different interfaces separately to 253 consistently serve network connection request. 255 In mixed mode, i.e., where of multiple networks are available on an 256 attached link only some of which advertise PvD information, the PvD- 257 aware node shall create explicit PvDs from explicitly learned PvD 258 information and associate other learned configuration (without an 259 explicit PvD) with implicit PvD(s) created for that interface. 261 2.3. Relationship Between PvDs and Interfaces 263 By default, implicit PvDs are limited to the network configuration 264 information received on a single interface and by default one such 265 PvD is formed for each interface. If additional information is 266 available to the host (through mechanisms out of scope of this 267 document), the host may form implicit PvDs with different 268 granularity. For example, PvDs spanning multiple interfaces such a 269 home network with a router that has multiple internal interfaces, or 270 multiple PvDs on a single interface such as a network that has 271 multiple uplink connections. 273 Explicit PvDs, in practice will often also be scoped only for 274 configuration related to a particular interface. However, there are 275 no such requirements or limitations defined in this architecture. 276 Explicit PvDs may include information related to more than one 277 interface if the node learns the presence of the same PvD on those 278 interfaces and the authentication of the PvD ID meets the level 279 required by the node policy (generally, authentication of a PvD ID 280 may be also required in scenarios involving only one connected 281 interface and / or PvD). 283 This architecture intends to support such scenarios, among others. 284 Hence, it shall be noted that no hierarchical relationship exists 285 between interfaces and PvDs: it is possible for multiple PvDs to be 286 simultaneously accessible over one interface, as well as a single PvD 287 to be simultaneously accessible over multiple interfaces. 289 2.4. PvD Identity / Naming 291 For explicit PvDs, the PvD ID is a value that is, or has a high 292 probability of being globally unique, and is received as part of PvD 293 information. It shall be possible to generate a human-readable form 294 of the PvD ID to present to the end-user, either based on the PvD ID 295 itself, or using meta-data associated with the ID. For implicit PvDs, 296 the node assigns a locally generated ID with a high probability of 297 being globally unique to each implicit PvD. 299 A PvD-aware node may use these IDs to select a PvD with a matching ID 300 for special-purpose connection requests in accordance with node 301 policy, as chosen by advanced applications, or to present a human- 302 readable representation of the IDs to the end-user for selection of 303 PvDs. 305 A single network provider may operate multiple networks, including 306 networks at different locations. In such cases, the provider may 307 chose whether to advertise single or multiple PvD identities at all 308 or some of those networks as it suits their business needs. This 309 architecture does not impose any specific requirements in this 310 regard. 312 When multiple nodes are connected to the same link with one or more 313 explicit PvDs available, this architecture assumes that the 314 information about all available PvDs is made available by the 315 networks to all the connected nodes. At the same time, connected 316 nodes may have different heuristics, policies and / or other 317 settings, including their configured sets of trusted PvDs. This may 318 lead to different PvDs actually being used by different nodes for 319 their connections. 321 Possible extensions, whereby networks advertize different sets of 322 PvDs to different connected nodes are out of scope of this document. 324 2.5. The Relationship to Dual-Stack Networks 326 When applied to dual-stack networks, the PvD definition allows for 327 multiple PvDs to be created whereby each PvD contains information 328 relevant to only one address family, or for a single PvD containing 329 information for multiple address families. This architecture 330 requires that accompanying design documents describing PvD-related 331 protocol changes must support PvDs containing information from 332 multiple address families. PvD-aware nodes must be capable of 333 creating and using both single-family and multi-family PvDs. 335 For explicit PvDs, the choice of either of these approaches is a 336 policy decision for the network administrator and / or the node user/ 337 administrator. Since some of the IP configuration information that 338 can be learned from the network can be applicable to multiple address 339 families (for instance DHCP Address Selection Policy Opt [RFC7078]), 340 it is likely that dual-stack networks will deploy single PvDs for 341 both address families. 343 By default for implicit PvDs, PvD-aware nodes shall include multiple 344 IP families into a single implicit PvD created for an interface. At 345 the time of writing, in dual-stack networks it appears to be common 346 practice for the configuration of both address families to be 347 provided by a single source. 349 A PvD-aware node that provides an API to use, enumerate and inspect 350 PvDs and / or their properties shall provide the ability to filter 351 PvDs and / or their properties by address family. 353 3. Conveying PvD information using DHCPv6 and Router Advertisements 355 DHCPv6 and Router Advertisements are the two most common methods of 356 configuring hosts. To support the architecture described in this 357 document, these protocols would need to be extended to convey 358 explicit PvD information. The following sections describe topic 359 which must be considered before finalizing a mechanism to augment 360 DHCPv6 and RAs with PvD information. 362 3.1. Separate Messages or One Message? 364 When information related to several PvDs is available from the same 365 configuration source, there are two possible ways of distributing 366 this information: One way is to send information from each different 367 provisioning domain in separate messages. The second method is 368 combining the information from multiple PvDs into a single message. 369 The latter method has the advantage of being more efficient but could 370 have problems with to authentication and authorization, as well as 371 potential issues with accommodating information not tagged with any 372 PvD information. 374 3.2. Securing PvD Information 376 DHCPv6 and RAs both provide some form of authentication to ensure the 377 identity of the source as well as the integrity of the secured 378 message content. While this is useful, determining authenticity does 379 tell a node whether the configuration source is actually allowed to 380 provide information from a given PvD. To resolve this, there must be 381 a mechanism for the PvD owner to attach some form of authorization 382 token to the configuration information that is delivered. 384 3.3. Backward Compatibility 386 The extensions to RAs and DHCPv6 should be defined in such a manner 387 than unmodified hosts (i.e. hosts not aware of PvDs) will continue 388 to function as well as they did prior to PvD information being added. 389 This could imply that some information may need to be duplicated in 390 order to be conveyed to legacy hosts. Similarly, PvD aware hosts 391 need to be able to correctly utiilize legacy configuration sources 392 which do not provide PvD information. There are also several 393 initiatives that are aimed at adding some form of additional 394 information to prefixes [I-D.bhandari-dhc-class-based-prefix] and 395 [I-D.korhonen-dmm-prefix-properties] and any new mechanism should try 396 to consider co-existence with such deployed mechanisms. 398 3.4. Selective Propagation 399 When a configuration source has information regarding several PvDs, 400 it is currently unclear whether the source should provide information 401 about all PvDs to any host that requests this information. While 402 this may be reasonable in some cases, it might become an unreasonable 403 burden once the number of PvDs starts increasing. One way to 404 restrict the propagation of information which is of no use to a 405 specific host is for the host to indicate the PvD information they 406 require within their configuration request. One way this could be 407 accomplished is by using a DHCPv6 ORO containing the PvDs that are of 408 interest. The configuration source can then respond with only the 409 requested information. 411 By default, a configuration source SHOULD provide information related 412 to all provisioning domains without expecting the client to request 413 the PvD(s) it requires. This is necessary to ensure that hosts that 414 do not support a selective PvD information request mechanism will 415 work. Also, note that IPv6 neighbor discovery does not provide any 416 functionality analogous to the DHCPv6 ORO. 418 In this case, when a host receives superfluous PvD information, it 419 can simply be discarded. Also, in constrained networks such as LLNs, 420 the amount of configuration information needs to be restricted to 421 ensure that the load on the hosts is bearable while keeping the 422 information identical across all the hosts. 424 If selective propagation is required, some form of PvD discovery 425 mechanism needs to be specified so that hosts / applications can be 426 pre-provisioned to request a specific PvD. Alternately, the set of 427 PvDs that the network can provide to the host can be propagated to 428 the host using RAs or stateless DHCPv6. The discovery mechanism may 429 potentially support the discovery of available PvDs on a per-host 430 basis. 432 3.5. Retracting / Updating PvD Information 434 After PvD information is provisioned to a host, it may become 435 outdated or superseded by updated information before the hosts would 436 normally request updates. To resolve this requires that the 437 mechanism be able to update and / or withdraw all (or some subset) of 438 the information related to a given PvD. For efficiency reasons, there 439 should be a way to specify that all information from the PvD needs to 440 be reconfigured instead of individually updating each item associated 441 with the PvD. 443 3.6. Conveying Configuration Information using IKEv2 444 Internet Key Exchange protocol version 2 (IKEv2) [RFC5996] [RFC5739] 445 is another widely used method of configuring host IP information. 446 For IKEv2, the provisioning domain could be implicitly learned from 447 the Identification - Responder (IDr) payloads that the IKEv2 448 initiator and responder inject during their IKEv2 exchange. The IP 449 configuration may depend on the named IDr. Another possibility could 450 be adding a specific provisioning domain identifying payload 451 extensions to IKEv2. All of the considerations for DHCPv6 and RAs 452 listed above potentially apply to IKEv2 as well. 454 4. Example Network Configurations 456 4.1. A Mobile Node 458 Consider a mobile node with two network interfaces: one to the mobile 459 network, the other to the Wi-Fi network. When the mobile node is 460 only connected to the mobile network, it will typically have one PvD, 461 implicit or explicit. When the mobile node discovers and connects to 462 a Wi-Fi network, it will have zero or more (typically one) additional 463 PvD(s). 465 Some existing OS implementations only allow one active network 466 connection. In this case, only the PvD(s) associated with the active 467 interface can be used at any given time. 469 As an example, the mobile network can explicitly deliver PvD 470 information through the PDP context activation process. Then, the 471 PvD aware mobile node will treat the mobile network as an explicit 472 PvD. Conversely, the legacy Wi-Fi network may not explicitly 473 communicate PvD information to the mobile node. The PvD aware mobile 474 node will associate network configuration for the Wi-Fi network with 475 an implicit PvD in this case. 477 The following diagram illustrates the use of different PvDs in this 478 scenario: 480 <----------- Wi-Fi 'Internet' PvD -------> 481 +--------+ 482 | +----+ | +----+ _ __ _ _ 483 | |WiFi| | | | ( ` ) ( ` )_ 484 | |-IF +-|----+ |--------------------------( `) 485 | | | | |WiFi| ( ) (_ Internet _) 486 | +----+ | | AP | ( ) `- __ _) - 487 | | | | ( Service ) 488 | | +----+ ( Provider's ) 489 | | ( Networks - 490 | +----+ | `_ ) _ _ 491 | |CELL| | ( ) ( ` )_ 492 | |-IF +-|-------------------------------------( `) 493 | | | | (_ __) (_ Internet _) 494 | +----+ | `- -- `- __ _) - 495 +--------+ 496 <------- Mobile 'Internet' PvD -----------> 498 An example of PvD use with Wi-Fi and mobile interfaces. 500 4.2. A Node with a VPN Connection 502 If the node has established a VPN connection, zero or more (typically 503 one) additional PvD(s) will be created. These may be implicit or 504 explicit. The routing to IP addresses reachable within this PvD will 505 be set up via the VPN connection, and the routing of packets to 506 addresses outside the scope of this PvD will remain unaffected. If a 507 node already has N connected PvDs, after the VPN session has been 508 established typically there will be N+1 connected PvDs. 510 The following diagram illustrates the use of different PvDs in this 511 scenario: 513 <----------- 'Internet' PvD ------------> 514 +--------+ 515 | +----+ | +----+ _ __ _ _ 516 | |Phy | | | | ( ` ) ( ` )_ 517 | |-IF +-|----+ |--------------------------( `) 518 | | | | | | ( ) (_ Internet _) 519 | +----+ | | | ( ) `- __ _) - 520 | | |Home| ( Service ) || 521 | | |gate| ( Provider's ) || 522 | | |-way| ( Network - || 523 | +----+ | | | `_ ) +---------+ +------------+ 524 | |VPN | | | | ( ) | VPN | | | 525 | |-IF +-|----+ |---------------------------+ Gateway |--+ Private | 526 | | | | | | (_ __) | | | Services | 527 | +----+ | +----+ `- -- +---------+ +------------+ 528 +--------+ 529 <-------------- Explicit 'VPN' PvD -----------> 531 An example of PvD use with VPN. 533 4.3. A Home Network and a Network Operator with Multiple PvDs 535 An operator may use separate PvDs for individual services which they 536 offer to their customers. These may be used so that services can be 537 designed and provisioned to be completely independent of each other, 538 allowing for complete flexibility in combinations of services which 539 are offered to customers. 541 From the perspective of the home network and the node, this model is 542 functionally very similar to being multihomed to multiple upstream 543 operators: Each of the different services offered by the service 544 provider is its own PvD with associated PvD information. In this 545 case, the operator may provide a generic / default PvD (explicit or 546 implicit), which provides Internet access to the customer. 547 Additional services would then be provisioned as explicit PvDs for 548 subscribing customers. 550 The following diagram illustrates this, using video-on-demand as a 551 service-specific PvD: 553 <------ Implicit 'Internet' PvD ------> 554 +----+ +----+ _ __ _ _ 555 | | | | ( ` ) ( ` )_ 556 | PC +-----+ |--------------------------( `) 557 | | | | ( ) (_ Internet _) 558 +----+ | | ( ) `- __ _) - 559 |Home| ( Service ) 560 |gate| ( Provider's ) 561 |-way| ( Network - 562 +-----+ | | `_ ) +---------+ 563 | Set | | | ( ) |ISP Video| 564 | Top +----+ |---------------------------+on Demand| 565 | Box | | | (_ __) | Service | 566 +-----+ +----+ `- -- +---------+ 567 <-- Explicit 'Video-on-Demand' PvD --> 569 An example of PvD use within a home network. 571 In this case, the number of PvDs that a single operator could 572 provision is based on the number of independently provisioned 573 services which they offer. Some examples may include: 575 o Real-time packet voice 577 o Streaming video 579 o Interactive video (n-way video conferencing) 581 o Interactive gaming 583 o Best effort / Internet access 585 5. Reference Model for the PvD-aware Node 586 5.1. Constructions and Maintenance of Separate PvDs 588 It is assumed that normally, the configuration information contained 589 in a single PvD shall be sufficient for a node to fulfill a network 590 connection request by an application, and hence there should be no 591 need to attempt to merge information across different PvDs. 593 Nevertheless, even when a PvD lacks some necessary configuration 594 information, merging of information associated with different PvD(s) 595 shall not be done automatically as this will typically lead to the 596 issues described in [RFC6418]. 598 A node may use other sources, for example: node local policy, user 599 input or other mechanisms not defined by the IETF for any of the 600 following: 602 o Construction of a PvD in its entirety (analogous to statically 603 configuring IP on an interface) 605 o Supplementing some, or all learned PvDs with particular 606 configuration elements 608 o Merging of information from different PvDs (if this is explicitly 609 allowed by policy) 611 As an example, a node administrator could inject a DNS server which 612 is not ISP-specific into PvDs for use on any of the networks that the 613 node could attach to. Such creation / augmentation of PvD(s) could 614 be static or dynamic. The specific mechanism(s) for implementing 615 this are outside of scope of this document. 617 5.2. Consistent use of PvDs for Network Connections 619 PvDs enable PvD-aware nodes to consistently use the correct set of 620 configuration elements to serve specific network requests from 621 beginning to end. This section provides examples of such use. 623 5.2.1. Name Resolution 625 When a PvD-aware node needs to resolve the name of the destination 626 for use by a connection request, the node could use one, or multiple 627 PvDs for a given name lookup. 629 The node shall chose a single PvD if, for example, the node policy 630 required the use of a particular PvD for a specific purpose (e.g. to 631 download an MMS message using a specific APN over a cellular 632 connection). To make this selection, the node could use a match 633 between the PvD DNS suffix and an FQDN which is being resolved or 634 match of PvD ID, as determined by the node policy. 636 The node may pick multiple PvDs, if for example, the PvDs are for 637 general purpose Internet connectivity, and the node is attempting to 638 maximize the probability of connectivity similar to the Happy 639 Eyeballs [RFC6555] approach. In this case, the node could perform 640 DNS lookups in parallel, or in sequence. Alternatively, the node may 641 use only one PvD for the lookup, based on the PvD connectivity 642 properties, user configuration of preferred Internet PvD, etc. 644 If an application implements an API that provides a way of explicitly 645 specifying the desired interface or PvD, that interface or PvD should 646 be used for name resolution (and the subsequent connection attempt), 647 provided that the host's configuration permits this. 649 In either case, by default a node uses information obtained via a 650 name service lookup to establish connections only within the same PvD 651 as the lookup results were obtained. 653 For clarification, when it is written that the name service lookup 654 results were obtained "from a PvD", it should be understood to mean 655 that the name service query was issued against a name service which 656 is configured for use in a particular PvD. In that sense, the 657 results are "from" that particular PvD. 659 Some nodes may support transports and / or APIs which provide an 660 abstraction of a single connection, aggregating multiple underlying 661 connections. MPTCP [RFC6182] is an example of such a transport 662 protocol. For connections provided by such transports/APIs, a PvD- 663 aware node may use different PvDs for servicing that logical 664 connection, provided that all operations on the underlying 665 connections are performed consistently within their corresponding 666 PvD(s). 668 5.2.2. Next-hop and Source Address Selection 670 For the purpose of this example, let us assume that the preceding 671 name lookup succeeded in a particular PvD. For each obtained 672 destination address, the node shall perform a next-hop lookup among 673 routers associated with that PvD. As an example, the node could 674 determine such associations via matching the source address prefixes/ 675 specific routes advertized by the router against known PvDs, or 676 receiving an explicit PvD affiliation advertized through a new Router 677 Discovery [RFC4861] option. 679 For each destination, once the best next-hop is found, the node 680 selects the best source address according to rules defined in 681 [RFC6724], but with the constraint that the source address must 682 belong to a range associated with the used PvD. If needed, the node 683 would use prefix policy from the same PvD for selecting the best 684 source address from multiple candidates. 686 When destination / source pairs are identified, they are sorted using 687 the [RFC6724] destination sorting rules and prefix policy table from 688 the used PvD. 690 5.2.3. Listening Applications 692 Consider a host connected to several PvDs, running an application 693 that opens a listening socket / transport API object. The 694 application is authorized by the host policy to use a subset of 695 connected PvDs that may or may not be equal to the complete set of 696 the connected PvDs. As an example, in the case where there are 697 different PvDs on the Wi-Fi and cellular interfaces, for general 698 Internet traffic the host could use only one, preferred PvD at a time 699 (and accordingly, advertise to remote peers the host name and 700 addresses associated with that PvD), or it could use one PvD as the 701 default for outgoing connections, while still allowing use of the 702 other PvDs simultaneously. 704 Another example is a host with an established VPN connection. Here, 705 security policy could be used to permit or deny application's access 706 to the VPN (and other) PvD(s). 708 For non-PvD aware applications, the operating system has policies 709 that determine the authorized set of PvDs and the preferred outgoing 710 PvD. For PvD-aware applications, both the authorized set of PvDs and 711 the default outgoing PvD can be determined as the common subset 712 produced between the OS policies and the set of PvD IDs or 713 characteristics provided by the application. 715 Application input could be provided on per-application, per- 716 transport-API-object or per-transport-API-call basis. The API for 717 application input may have an option for specifying whether the input 718 should be treated as a preference instead of a requirement. 720 5.2.3.1. Processing of Incoming Traffic 722 Unicast IP packets are received on a specific IP address associated 723 with a PvD. For multicast packets, the host can derive the PvD 724 association from other configuration information, such as an explicit 725 PvD property or local policy. 727 The node OS or middleware may apply more advanced techniques for 728 determining the resultant PvD and / or authorization of the incoming 729 traffic. Those techniques are outside of scope of this document. 731 If the determined receiving PvD of a packet is not in the allowed 732 subset of PvDs for the particular application / transport API object, 733 the packet should be handled in the same way as if there were no 734 listener. 736 5.2.3.1.1. Connection-oriented APIs 738 For connection-oriented APIs, when the initial incoming packet is 739 received, the packet PvD is remembered for the established connection 740 and used for handling of outgoing traffic for that connection. While 741 typically, connection-oriented APIs use a connection-oriented 742 transport protocol, such as TCP, it is possible to have a connection- 743 oriented API that uses a generally connectionless transport protocol, 744 such as UDP. 746 For APIs/protocols that support multiple IP traffic flows associated 747 with a single transport API connection object (for example, multi 748 path TCP), the processing rules may be adjusted accordingly. 750 5.2.3.1.2. Connectionless APIs 752 For connectionless APIs, the host should provide an API that PvD- 753 aware applications can use to query the PvD associated with the 754 packet. For outgoing traffic on this transport API object, the OS 755 should use the selected outgoing PvDs, determined as described above. 757 5.2.4. Enforcement of Security Policies 759 By themselves, PvDs do not define, and cannot be used for 760 communication of, security policies. When implemented in a network, 761 this architecture provides the host with information about connected 762 networks. The actual behavior of the host then depends on the host's 763 policies (provisioned through mechanisms out of scope of this 764 document), applied taking received PvD information into account. In 765 some scenarios, e.g. a VPN, such policies could require the host to 766 use only a particular VPN PvD for some / all of the application's 767 traffic (VPN 'disable split tunneling' also known as 'force 768 tunneling' behavior), or apply such restrictions only to selected 769 applications and allow the simultaneous use of the VPN PvD together 770 with the other connected PvDs by the other or all applications (VPN 771 'split tunneling' behavior). 773 5.3. Connectivity Tests 775 Although some PvDs may appear as valid candidates for PvD selection 776 (e.g. good link quality, consistent connection parameters, etc.), 777 they may provide limited or no connectivity to the desired network or 778 the Internet. For example, some PvDs provide limited IP connectivity 779 (e.g., scoped to the link or to the access network), but require the 780 node to authenticate through a web portal to get full access to the 781 Internet. This may be more likely to happen for PvDs which are not 782 trusted by a given PvD-aware node. 784 An attempt to use such a PvD may lead to limited network connectivity 785 or application connection failures. To prevent the latter, a PvD- 786 aware node may perform a connectivity test for the PvD before using 787 it to serve application network connection requests. In current 788 implementations, some nodes already implement this e.g., by trying to 789 reach a dedicated web server (see [RFC6419]). 791 Section 5.2 describes how a PvD-aware node shall maintain and use 792 multiple PvDs separately. The PvD-aware node shall perform a 793 connectivity test and, only after validation of the PvD, consider 794 using it to serve application connections requests. Ongoing 795 connectivity tests are also required, since during the IP session, 796 the end-to-end connectivity could be disrupted for various reasons 797 (e.g. L2 problems, IP QoS issues); hence, a connectivity monitoring 798 function is needed to check the connectivity status and remove the 799 PvD from the set of usable PvDs if necessary. 801 There may be cases where a connectivity test for PvD selection may 802 not be appropriate and should be complemented, or replaced, by PvD 803 selection based on other factors. For example, this could be 804 realized by leveraging some 3GPP and IEEE mechanisms, which would 805 allow the exposure of some PvD characteristics to the node (e.g. 806 3GPP Access Network Discovery and Selection Function (ANDSF) 807 [TS23.402], IEEE 802.11u [IEEE802.11u]/ANQP). 809 5.4. Relationship to Interface Management and Connection Managers 811 Current devices, such as mobile handsets make use of proprietary 812 mechanisms and custom applications to manage connectivity in 813 environments with multiple interfaces and multiple sets of network 814 configuration. These mechanisms or applications are commonly known 815 as connection managers [RFC6419]. 817 Connection managers sometimes rely on policy servers to allow a node 818 that is connected to multiple networks to perform network selection. 819 They can also make use of routing guidance from the network (e.g. 820 3GPP ANDSF [TS23.402]). Although connection managers solve some 821 connectivity problems, they rarely address network selection problems 822 in a comprehensive manner. With proprietary solutions, it is 823 challenging to present coherent behavior to the end user of the 824 device, as different platforms present different behaviors even when 825 connected to the same network, with the same type of interface, and 826 for the same purpose. The architecture described in this document 827 should improve the hosts behavior by providing the hosts with tools 828 and guidance to make informed network selection decisions. 830 6. PvD support in APIs 831 For all levels of PvD support in APIs described in this chapter, it 832 is expected that the notifications about changes in the set of 833 available PvDs are exposed as part of the API surface. 835 6.1. Basic 837 Applications are not PvD-aware in any manner and only submit 838 connection requests. The node performs PvD selection implicitly, 839 without any application participation, based purely on node-specific 840 administrative policies and / or choices made by the user from a user 841 interface provided by the operating environment, not by the 842 application. 844 As an example, PvD selection can be done at the name service lookup 845 step by using the relevant configuration elements, such as those 846 described in [RFC6731]. As another example, PvD selection could be 847 made based on application identity or type (i.e., a node could always 848 use a particular PvD for a VOIP application). 850 6.2. Intermediate 852 Applications indirectly participate in PvD selection by specifying 853 hard requirements and soft preferences. As an example, a real time 854 communication application intending to use the connection for the 855 exchange of real time audio / video data may indicate a preference or 856 a requirement for connection quality, which could affect PvD 857 selection (different PvDs could correspond to Internet connections 858 with different loss rates and latencies). 860 Another example is the connection of an infrequently executed 861 background activity, which checks for application updates and 862 performs large downloads when updates are available. For such 863 connections, a cheaper or zero cost PvD may be preferable, even if 864 such a connection has a higher relative loss rate or lower bandwidth. 865 The node performs PvD selection based on applications' inputs and 866 policies and / or user preferences. Some / all properties of the 867 resultant PvD may be exposed to applications. 869 6.3. Advanced 871 PvDs are directly exposed to applications for enumeration and 872 selection. Node polices and / or user choices may still override the 873 applications' preferences and limit which PvD(s) can be enumerated 874 and / or used by the application, irrespective of any preferences 875 which the application may have specified. Depending on the 876 implementation, such restrictions (imposed by node policy and / or 877 user choice) may or may not be visible to the application. 879 7. PvD Trust for PvD-Aware Node 881 7.1. Untrusted PvDs 882 Implicit and explicit PvDs for which no trust relationship exists are 883 considered untrusted. Only PvDs which meet the requirements in 884 Section 7.2 are trusted; any other PvD is untrusted. 886 In order to avoid the various forms of misinformation that could 887 occur when PvDs are untrusted, nodes that implement PvD separation 888 cannot assume that two explicit PvDs with the same identifier are 889 actually the same PvD. A node that makes this assumption will be 890 vulnerable to attacks where, for example, an open Wifi hotspot might 891 assert that it was part of another PvD and thereby attempt to draw 892 traffic intended for that PvD onto its own network. 894 Since implicit PvD identifiers are synthesized by the node, this 895 issue cannot arise with implicit PvDs. 897 Mechanisms exist (for example, [RFC6731]) whereby a PvD can provide 898 configuration information that asserts special knowledge about the 899 reachability of resources through that PvD. Such assertions cannot be 900 validated unless the node has a trust relationship with the PvD; 901 therefore, assertions of this type must be ignored by nodes that 902 receive them from untrusted PvDs. Failure to ignore such assertions 903 could result in traffic being diverted from legitimate destinations 904 to spoofed destinations. 906 7.2. Trusted PvDs 908 Trusted PvDs are PvDs for which two conditions apply: First, a trust 909 relationship must exist between the node that is using the PvD 910 configuration and the source that provided that configuration; this 911 is the authorization portion of the trust relationship. Second, 912 there must be some way to validate the trust relationship. This is 913 the authentication portion of the trust relationship. Two mechanisms 914 for validating the trust relationship are defined. 916 It shall be possible to validate the trust relationship for all 917 advertised elements of a trusted PvD, irrespective of whether the PvD 918 elements are communicated as a whole, e.g., in a single DHCP option, 919 or separately, e.g., in supplementary RA options. The feasibility of 920 mechanisms to implement a trust relationship for all PvD elements 921 will be determined in the respective companion design documents. 923 7.2.1. Authenticated PvDs 925 One way to validate the trust relationship between a node and the 926 source of a PvD is through the combination of cryptographic 927 authentication and an identifier configured on the node. In some 928 cases, the two could be the same; for example, if authentication is 929 by a shared secret, the secret would have to be associated with the 930 PvD identifier. Without a PvD Identifier / shared key tuple, 931 authentication would be impossible, and hence authentication and 932 authorization are combined. 934 However, if authentication is done using a public key mechanism such 935 as a TLS certificate or DANE, authentication by itself is not enough 936 since theoretically any PvD could be authenticated in this way. In 937 addition to authentication, the node would need configuration to 938 trust the identifier being authenticated. Validating the 939 authenticated PvD name against a list of PvD names configured as 940 trusted on the node would constitute the authorization step in this 941 case. 943 7.2.2. PvDs Trusted by Attachment 945 In some cases, a trust relationship may be validated by some means 946 other than those described in Section 7.2.1 simply by virtue of the 947 connection through which the PvD was obtained. For instance, a 948 handset connected to a mobile network may know through the mobile 949 network infrastructure that it is connected to a trusted PvD. 950 Whatever mechanism was used to validate that connection constitutes 951 the authentication portion of the PvD trust relationship. 952 Presumably, such a handset would be configured from the factory (or 953 else through mobile operator or user preference settings) to trust 954 the PvD, and this would constitute the authorization portion of this 955 type of trust relationship. 957 8. Acknowledgments 959 The following people contributed to this document (in no specific 960 order): Alper Yegin, Aaron Yi Ding, Zhen Cao, Dan York, Dapeng Liu, 961 Dave Thaler, Dmitry Anipko, Fred Baker, Hui Deng, Ian Farrer, Jouni 962 Korhonen, Juan Carlos Zuniga, Konstantinos Pentikousis, Marc 963 Blanchet, Marcus Stenberg, Margaret Wasserman, Mikael Abrahamsson, 964 Pierrick Seite, Suresh Krishnan, Teemu Savolainen, Ted Lemon and Tim 965 Chown. 967 9. IANA Considerations 969 This memo does not include any IANA requests. 971 10. Security Considerations 973 There are at least three different forms of attacks that can be 974 performed using configuration sources that support multiple 975 provisioning domains. 977 Tampering with provided configuration information: An attacker may 978 attempt to modify information provided inside the PvD container 979 option. These attacks can easily be prevented by using message 980 integrity features provided by the underlying protocol used to 981 carry the configuration information. E.g. SEND [RFC3971] would 982 detect any form of tampering with the RA contents and the DHCPv6 983 [RFC3315] AUTH option that would detect any form of tampering with 984 the DHCPv6 message contents. This attack can also be performed by 985 a compromised configuration source by modifying information inside 986 a specific PvD, in which case the mitigations proposed in the next 987 subsection may be helpful. 989 Rogue configuration source: A compromised configuration source, such 990 as a router or a DHCPv6 server, may advertise information about 991 PvDs that it is not authorized to advertise. e.g. A coffee shop 992 WLAN may advertise configuration information purporting to be from 993 an enterprise and may try to attract enterprise related traffic. 994 The only real way to prevent this is is for the PvD related 995 configuration container to contain embedded authentication and 996 authorization information from the owner of the PvD. This provides 997 the client with a way of detecting the attack by verifying the 998 authentication and authorization information provided inside the 999 PvD container option, after verifying its trust of the PvD owner 1000 (e.g. a certificate with a well-known / common trust anchor). 1002 Replay attacks: A compromised configuration source or an on-link 1003 attacker may try to capture advertised configuration information 1004 and replay it on a different link, or at a future point in time. 1005 This can be avoided by including a replay protection mechanism 1006 such as a timestamp or a nonce inside the PvD container to ensure 1007 the validity of the provided information. 1009 11. References 1011 11.1. Normative References 1013 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1014 Requirement Levels", BCP 14, RFC 2119, March 1997. 1016 11.2. Informative References 1018 [I-D.bhandari-dhc-class-based-prefix] 1019 Systems, C., Halwasia, G., Gundavelli, S., Deng, H., 1020 Thiebaut, L., Korhonen, J. and I. Farrer, "DHCPv6 class 1021 based prefix", Internet-Draft draft-bhandari-dhc-class- 1022 based-prefix-05, July 2013. 1024 [I-D.korhonen-dmm-prefix-properties] 1025 Korhonen, J., Patil, B., Gundavelli, S., Seite, P. and D. 1026 Liu, "IPv6 Prefix Mobility Management Properties", 1027 Internet-Draft draft-korhonen-dmm-prefix-properties-03, 1028 October 2012. 1030 [IEEE802.11u] 1031 IEEE, "IEEE Standard 802.11u-2011 (Amendment 9: 1032 Interworking with External Networks)", 2011. 1034 [RFC3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and 1035 M. Carney, "Dynamic Host Configuration Protocol for IPv6 1036 (DHCPv6)", RFC 3315, July 2003. 1038 [RFC3971] Arkko, J., Kempf, J., Zill, B. and P. Nikander, "SEcure 1039 Neighbor Discovery (SEND)", RFC 3971, March 2005. 1041 [RFC4861] Narten, T., Nordmark, E., Simpson, W. and H. Soliman, 1042 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 1043 September 2007. 1045 [RFC5739] Eronen, P., Laganier, J. and C. Madson, "IPv6 1046 Configuration in Internet Key Exchange Protocol Version 2 1047 (IKEv2)", RFC 5739, February 2010. 1049 [RFC5996] Kaufman, C., Hoffman, P., Nir, Y. and P. Eronen, "Internet 1050 Key Exchange Protocol Version 2 (IKEv2)", RFC 5996, 1051 September 2010. 1053 [RFC6182] Ford, A., Raiciu, C., Handley, M., Barre, S. and J. 1054 Iyengar, "Architectural Guidelines for Multipath TCP 1055 Development", RFC 6182, March 2011. 1057 [RFC6418] Blanchet, M. and P. Seite, "Multiple Interfaces and 1058 Provisioning Domains Problem Statement", RFC 6418, 1059 November 2011. 1061 [RFC6419] Wasserman, M. and P. Seite, "Current Practices for 1062 Multiple-Interface Hosts", RFC 6419, November 2011. 1064 [RFC6555] Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with 1065 Dual-Stack Hosts", RFC 6555, April 2012. 1067 [RFC6724] Thaler, D., Draves, R., Matsumoto, A. and T. Chown, 1068 "Default Address Selection for Internet Protocol Version 6 1069 (IPv6)", RFC 6724, September 2012. 1071 [RFC6731] Savolainen, T., Kato, J. and T. Lemon, "Improved Recursive 1072 DNS Server Selection for Multi-Interfaced Nodes", RFC 1073 6731, December 2012. 1075 [RFC7078] Matsumoto, A., Fujisaki, T. and T. Chown, "Distributing 1076 Address Selection Policy Using DHCPv6", RFC 7078, January 1077 2014. 1079 [TS23.402] 1080 3GPP, "3GPP TS 23.402; Architecture enhancements for non- 1081 3GPP accesses; release 12", . 1083 Author's Address 1085 Dmitry Anipko, editor 1086 Microsoft Corporation 1087 One Microsoft Way 1088 Redmond, WA 98052 1089 USA 1091 Phone: +1 425 703 7070 1092 Email: dmitry.anipko@microsoft.com