idnits 2.17.1 draft-ietf-karp-ops-model-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 12, 2012) is 4299 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-10) exists of draft-ietf-karp-crypto-key-table-03 -- Obsolete informational reference (is this intentional?): RFC 4572 (Obsoleted by RFC 8122) Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group S. Hartman 3 Internet-Draft Painless Security 4 Intended status: Informational D. Zhang 5 Expires: January 13, 2013 Huawei 6 July 12, 2012 8 Operations Model for Router Keying 9 draft-ietf-karp-ops-model-03.txt 11 Abstract 13 Developing an operational and management model for routing protocol 14 security that works across protocols will be critical to the success 15 of routing protocol security efforts. This document discusses issues 16 and begins to consider development of these models. 18 Status of this Memo 20 This Internet-Draft is submitted in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at http://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on January 13, 2013. 35 Copyright Notice 37 Copyright (c) 2012 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (http://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 2. Requirements notation . . . . . . . . . . . . . . . . . . . . 4 54 3. Breakdown of KARP configuration . . . . . . . . . . . . . . . 5 55 3.1. Integrity of the Key Table . . . . . . . . . . . . . . . . 6 56 3.2. Management of Key Table . . . . . . . . . . . . . . . . . 6 57 3.3. Interactions with Automated Key Management . . . . . . . . 7 58 3.4. VRFs . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 59 4. Credentials and Authorization . . . . . . . . . . . . . . . . 8 60 4.1. Preshared Keys . . . . . . . . . . . . . . . . . . . . . . 9 61 4.2. Asymmetric Keys . . . . . . . . . . . . . . . . . . . . . 11 62 4.3. Public Key Infrastructure . . . . . . . . . . . . . . . . 11 63 4.4. The role of Central Servers . . . . . . . . . . . . . . . 12 64 5. Grouping Peers Together . . . . . . . . . . . . . . . . . . . 13 65 6. Administrator Involvement . . . . . . . . . . . . . . . . . . 15 66 6.1. Enrollment . . . . . . . . . . . . . . . . . . . . . . . . 15 67 6.2. Handling Faults . . . . . . . . . . . . . . . . . . . . . 15 68 7. Upgrade Considerations . . . . . . . . . . . . . . . . . . . . 17 69 8. Security Considerations . . . . . . . . . . . . . . . . . . . 18 70 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 71 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 20 72 10.1. Normative References . . . . . . . . . . . . . . . . . . . 20 73 10.2. Informative References . . . . . . . . . . . . . . . . . . 20 74 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 21 76 1. Introduction 78 The KARP working group is designing improvements to the cryptographic 79 authentication of IETF routing protocols. These improvements include 80 improvements to how integrity functions are handled within each 81 protocol as well as designing an automated key management solution. 83 This document discusses issues to consider when thinking about the 84 operational and management model for KARP. Each implementation will 85 take its own approach to management; this is one area for vendor 86 differentiation. However, it is desirable to have a common baseline 87 for the management objects allowing administrators, security 88 architects and protocol designers to understand what management 89 capabilities they can depend on in heterogeneous environments. 90 Similarly, designing and deploying the protocol will be easier with 91 thought paid to a common operational model. This will also help with 92 the design of NetConf schemas or MIBs later. 94 2. Requirements notation 96 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 97 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 98 document are to be interpreted as described in [RFC2119]. 100 3. Breakdown of KARP configuration 102 There are multiple ways of structuring configuration information. 103 One factor to consider is the scope of the configuration information. 104 Several protocols are peer-to-peer routing protocols where a 105 different key could potentially be used for each neighbor. Other 106 protocols require the same group key to be used for all nodes in an 107 administrative domain or routing area. In other cases, the same 108 group key needs to be used for all routers on an interface, but 109 different group keys can be used for each interface. 111 Within situations where a per-interface, per-area or per-peer key can 112 be used for manually configured long-term keys, that flexibility may 113 not be desirable from an operational standpoint. For example 114 consider OSPF [RFC2328]. Each OSPF link needs to use the same 115 authentication configuration, including the set of keys used for 116 reception and the set of keys used for transmission, but may use 117 different keys for different links. The most general management 118 model would be to configure keys per link. However for deployments 119 where the area uses the same key it would be strongly desirable to 120 configure the key as a property of the area. If the keys are 121 configured per-link, they can get out of sync. In order to support 122 generality of configuration and common operational situations, it 123 would be desirable to have some sort of inheritance where default 124 configurations are made per-area unless overridden per-interface. 126 As described in [I-D.ietf-karp-crypto-key-table], the cryptographic 127 keys are separated from the interface configuration into their own 128 configuration store. Each routing protocol is responsible for 129 defining the form of the Peer specification used by that protocol. 130 Thus each routing protocol needs to define the scope of keys. For 131 group keying, the Peer specification names the group. A protocol 132 could define a Peer specification indicating the key had a link scope 133 and also a Peer specification for scoping a key to a specific area. 134 For link-scoped keys it is generally best to define a single Peer 135 specification indicating the key has a link scope and to use 136 interface restrictions to restrict the key to the appropriate link. 138 Operational Requirements: KARP MUST support configuration of keys at 139 the most general scope for the underlying protocol; protocols 140 supporting per-peer keys MUST permit configuration of per-peer keys, 141 protocols supporting per-interface keys MUST support configuration of 142 per-interface keys, and so on. KARP MUST NOT permit configuration of 143 an inappropriate key scope. For example, configuration of separate 144 keys per interface MUST NOT be supported for a protocol requiring 145 per-area keys. This restriction can be enforced by rules specified 146 by each routing protocol for validating key table entries. 148 3.1. Integrity of the Key Table 150 The routing key table [I-D.ietf-karp-crypto-key-table] provides a 151 very general mechanism to abstract the storage of keys for routing 152 protocols. To avoid misconfiguration and simplify problem 153 determination, the router MUST verify the internal consistency of 154 entries added to the table. Routing protocols describe how their 155 protocol interacts with the key table including what validation MUSt 156 be performed. At a minimum, the router MUST verify: 158 o The cryptographic algorithms are valid for the protocol. 160 o The key derivation function is valid for the protocol. 162 o The direction is valid for the protocol; for example protocols 163 that require the same session key be used in both directions MUST 164 have a direction of both. 166 o The peer specification is consistent with the protocol. 168 Other checks are possible. For example the router could verify that 169 if a key is associated with a peer, that peer is a configured peer 170 for the specified protocol. However, this may be undesirable. It 171 may be desirable to load a key table when some peers have not yet 172 been configured. Also, it may be desirable to share portions of a 173 key table across devices even when their current configuration does 174 not require an adjacency with a particular peer in the interest of 175 uniform configuration or preparing for fail-over. 177 3.2. Management of Key Table 179 Several management operations will be quite common. For service 180 provider deployments the configuration management system can simply 181 update the key table. However, for smaller deployments, efficient 182 management operations are important. 184 As part of adding a new key it is typically desirable to set an 185 expiration time for an old key. The management interface SHOULD 186 provide a mechanism to easily update the expiration time for a 187 current key used with a given peer or interface. Also when adding a 188 key it is desirable to push the key out to nodes that will need it, 189 allowing use for receiving packets then later enabling transmit. 190 This can be accomplished automatically by providing a delay between 191 when a key becomes valid for reception and transmission. However, 192 some environments may not be able to predict when all the necessary 193 changes will be made. In these cases having a mechanism to enable a 194 key for sending is desirable. 196 The key table's schema supports these operations. However equipment 197 can improve usability by providing convenient functions to effect 198 these common changes. 200 3.3. Interactions with Automated Key Management 202 Consideration is required for how an automated key management 203 protocol will assign key IDs for group keys. All members of the 204 group may need to use the same key ID. This requires careful 205 coordination of global key IDs. Interactions with the peer key ID 206 field may make this easier; this requires additional study. 208 Automated key management protocols also assign keys for single peers. 209 If the key ID is global and needs to be coordinated between the 210 receiver and transmitter, then there is complexity in key management 211 protocols. 213 3.4. VRFs 215 Many core and enterprise routers support multiple routing instances. 216 For example a router serving multiple VPNs is likely to have a 217 forwarding/routing instance for each of these VPNs. We need to 218 decide how the key table and other configuration information for KARP 219 interacts with this. The obvious first-order answer is that each 220 routing instance gets its own key table. However, we need to 221 consider how these instances interact with each other and confirm 222 this makes sense. 224 4. Credentials and Authorization 226 Several methods for authentication have been proposed for KARP. The 227 simplest is preshared keys used directly as traffic keys. In this 228 mode, the traffic integrity keys are directly configured. This is 229 the mode supported by most of today's routing protocols. 231 As discussed in [I-D.polk-saag-rtg-auth-keytable], preshared keys can 232 be used as the input to a key derivation function (KDF) to generate 233 traffic keys. For example the TCP Authentication Option (TCP-AO) 234 [RFC5925] derives keys based on the initial TCP session state. 235 Typically a KDF will combine a long-term key with public inputs 236 exchanged as part of the protocol to form fresh session keys. a KDF 237 could potentially be used with some inputs that are configured along 238 with the long-term key. Also, it's possible that inputs to a KDF 239 will be private and exchanged as part of the protocol, although this 240 will be uncommon in KARP's uses of KDFs. 242 Preshared keys could also be used by an automated key management 243 protocol. In this mode, preshared keys would be used for 244 authentication. However traffic keys would be generated by some key 245 agreement mechanism or transported in a key encryption key derived 246 from the preshared key. This mode may provide better replay 247 protection. Also, in the absence of active attackers, key agreement 248 strategies such as Diffie-Hellman can be used to produce high-quality 249 traffic keys even from relatively weak preshared keys. 251 Public keys can be used for authentication. The design guide 252 [I-D.ietf-karp-design-guide] describes a mode in which routers have 253 the hashes of peer routers' public keys. In this mode, a traditional 254 public-key infrastructure is not required. The advantage of this 255 mode is that a router only contains its own keying material, limiting 256 the scope of a compromise. The disadvantage is that when a router is 257 added or deleted from the set of authorized routers, all routers that 258 peer need to be updated. Note that self-signed certificates are a 259 common way of communicating public-keys in this style of 260 authentication. 262 Certificates signed by a certification authority or some other PKI 263 could be used. The advantage of this approach is that routers may 264 not need to be directly updated when peers are added or removed. The 265 disadvantage is that more complexity and cost is required. 267 Each of these approaches has a different set of management and 268 operational requirements. Key differences include how authorization 269 is handled and how identity works. This section discusses these 270 differences. 272 4.1. Preshared Keys 274 In the protocol, manual preshared keys are either unnamed or named by 275 a small integer (typically 16 or 32 bits) key ID. Implementations 276 that support multiple keys for protocols that have no names for keys 277 need to try all possible keys before deciding a packet cannot be 278 validated [RFC4808]. Typically key IDs are names used by one group 279 or peer. 281 Manual preshared keys are often known by a group of peers rather than 282 just oneother peer. This is an interesting security property: unlike 283 with digitally signed messages or protocols where symmetric keys are 284 known only to two parties, it is impossible to identify the peer 285 sending a message cryptographically. However, it is possible to show 286 that the sender of a message is one of the parties who knows the 287 preshared key. Within the routing threat model the peer sending a 288 message can be identified only because peers are trusted and thus can 289 be assumed to correctly label the packets they send. This contrasts 290 with a protocol where cryptographic means such as digital signatures 291 are used to verify the origin of a message. As a consequence, 292 authorization is typically based on knowing the preshared key rather 293 than on being a particular peer. Note that once an authorization 294 decision is made, the peer can assert its identity; this identity is 295 trusted just as the routing information from the peer is trusted. 296 Doing an additional check for authorization based on the identity 297 included in the packet would provide little value: an attacker who 298 somehow had the key could claim the identity of an authorized peer 299 and an attacker without the key should be unable to claim the 300 identity of any peer. Such a check is not required by the KARP 301 threat model: inside attacks are not in scope. 303 Preshared keys used with key derivation function similarly to manual 304 preshared keys. However to form the actual traffic keys, session or 305 peer specific information is combined with the key. From an 306 authorization standpoint, the derivation key works the same as a 307 manual key. An additional routing protocol step or transport step 308 forms the key that is actually used. 310 Preshared keys that are used via automatic key management have not 311 been specified for KARP. Their naming and authorization may differ 312 from existing uses of preshared keys in routing protocols. In 313 particular, such keys may end up being known only by two peers. 314 Alternatively they may also be known by a group of peers. 315 Authorization could potentially be based on peer identity, although 316 it is likely that knowing the right key will be sufficient. There 317 does not appear to be a compelling reason to decouple the 318 authorization of a key for some purpose from authorization of peers 319 holding that key to perform the authorized function. 321 Care needs to be taken when symmetric keys are used for multiple 322 purposes. Consider the implications of using the same preshared key 323 for two interfaces: it becomes impossible to cryptographically 324 distinguish a router on one interface from a router on another 325 interface. So, a router that is trusted to participate in a routing 326 protocol on one interface becomes implicitly trusted for the other 327 interfaces that share the key. For many cases, such as link-state 328 routers in the same routing area, there is no significant advantage 329 that an attacker could gain from this trust within the KARP threat 330 model. However, distance-vector protocols, such as BGP and RIP, 331 permit routes to be filtered across a trust boundary. For these 332 protocols, participation in one interface might be more advantageous 333 than another. Operationally, when this trust distinction is 334 important to a deployment, different keys need to be used on each 335 side of the trust boundary. Key derivation can help prevent this 336 problem in cases of accidental misconfiguration. However, key 337 derivation cannot protect against a situation where a system was 338 incorrectly trusted to have the key used to perform the derivation. 339 To the extent that there are multiple zones of trust and a routing 340 protocol is determining whether a particular router is within a 341 certain zone, the question of untrusted actors is within the scope of 342 the routing threat model. 344 Key derivation can be part of a management solution to a desire to 345 have multiple keys for different zones of trust. A master key could 346 be combined with peer, link or area identifiers to form a router- 347 specific preshared key that is loaded onto routers. Provided that 348 the master key lives only on the management server and not the 349 individual routers, trust is preserved. However in many cases, 350 generating independent keys for the routers and storing the result is 351 more practical. If the master key were somehow compromised, all the 352 resulting keys would need to be changed. However if independent keys 353 are used, the scope of a compromise may be more limited. 355 More subtle problems with key separation can appear in protocol 356 design. Two protocols that use the same traffic keys may work 357 together in unintended ways permitting one protocol to be used to 358 attack the other. Consider two hypothetical protocols. Protocol A 359 starts its messages with a set of extensions that are ignored if not 360 understood. Protocol B has a fixed header at the beginning of its 361 messages but ends messages with extension information. It may be 362 that the same message is valid both as part of protocol A and 363 protocol B. An attacker may be able to gain an advantage by getting a 364 router to generate this message with one protocol under situations 365 where the other protocol would not generate the message. This 366 hypothetical example is overly simplistic; real-world attacks 367 exploiting key separation weaknesses tend to be complicated and 368 involve specific properties of the cryptographic functions involved. 370 The key point is that whenever the same key is used in multiple 371 protocols, attacks may be possible. All the involved protocols need 372 to be analyzed to understand the scope of potential attacks. 374 Key separation attacks interact with the KARP operational model in a 375 number of ways. Administrators need to be aware of situations where 376 using the same manual traffic key with two different protocols (or 377 the same protocol in different contexts) creates attack 378 opportunities. Design teams should consider how their protocol might 379 interact with other routing protocols and describe any attacks 380 discovered so that administrators can understand the operational 381 implications. When designing automated key management or new 382 cryptographic authentication within routing protocols, we need to be 383 aware that administrators expect to be able to use the same preshared 384 keys in multiple contexts. As a result, we should use appropriate 385 key derivation functions so that different cryptographic keys are 386 used even when the same initial input key is used. 388 4.2. Asymmetric Keys 390 Outside of a PKI, public keys are expected to be known by the hash of 391 a key or (potentially self-signed) certificate. The Session 392 Description Protocol provides a standardized mechanism for naming 393 keys (in that case certificates) based on hashes (section 5 394 [RFC4572]). KARP SHOULD adopt this approach or another approach 395 already standardized within the IETF rather than inventing a new 396 mechanism for naming public keys. 398 A public key is typically expected to belong to one peer. As a peer 399 generates new keys and retires old keys, its public key may change. 400 For this reason, from a management standpoint, peers should be 401 thought of as associated with multiple public keys rather than as 402 containing a single public key hash as an attribute of the peer 403 object. 405 Authorization of public keys could be done either by key hash or by 406 peer identity. Performing authorizations by peer identity should 407 make it easier to update the key of a peer without risk of losing 408 authorizations for that peer. However management interfaces need to 409 be carefully designed to avoid making this extra level of indirection 410 complicated for operators. 412 4.3. Public Key Infrastructure 414 When a PKI is used, certificates are used. The certificate binds a 415 key to a name of a peer. The key management protocol is responsible 416 for exchanging certificates and validating them to a trust anchor. 418 Authorization needs to be done in terms of peer identities not in 419 terms of keys. One reason for this is that when a peer changes its 420 key, the new certificate needs to be sufficient for authentication to 421 continue functioning even though the key has never been seen before. 423 Potentially authorization could be performed in terms of groups of 424 peers rather than single peers. An advantage of this is that it may 425 be possible to add a new router with no authentication related 426 configuration of the peers of that router. For example, a domain 427 could decide that any router with a particular keyPurposeID signed by 428 the organization's certificate authority is permitted to join the 429 IGP. Just as in configurations where cryptographic authentication is 430 not used, automatic discovery of this router can establish 431 appropriate adjacencies. 433 Assuming that potentially self-signed certificates are used by 434 routers that wish to use public keys but that do not need a PKI, then 435 PKI and the infrastructureless mode of public-key operation described 436 in the previous section can work well together. One router could 437 identify its peers based on names and use certificate validation. 438 Another router could use hashes of certificates. This could be very 439 useful for border routers between two organizations. Smaller 440 organizations could use public keys and larger organizations could 441 use PKI. 443 4.4. The role of Central Servers 445 An area to explore is the role of central servers like RADIUS or 446 directories. As discussed in the design-guide, a system where keys 447 are pushed by a central management system is undesirable as an end 448 result for KARP. However central servers may play a role in 449 authorization and key rollover. For example a node could send a hash 450 of a public key to a RADIUS server. 452 If central servers do play a role it will be critical to make sure 453 that they are not required during routine operation or a cold-start 454 of a network. They are more likely to play a role in enrollment of 455 new peers or key migration/compromise. 457 Another area where central servers may play a role is for group key 458 agreement. As an example, [I-D.liu-ospfv3-automated-keying-req] 459 discusses the potential need for key agreement servers in OSPF. 460 Other routing protocols that use multicast or broadcast such as IS-IS 461 are likely to need a similar approach. 463 5. Grouping Peers Together 465 One significant management consideration will be the grouping of 466 management objects necessary to determine who is authorized to act as 467 a peer for a given routing action. As discussed previously, the 468 following objects are potentially required: 470 o Key objects are required. Symmetric keys may be preshared. 471 Asymmetric public keys may be used directly for authorization as 472 well. During key transitions more than one key may refer to a 473 given peer. Group preshared keys may refer to multiple peers. 475 o A peer is a router that this router might wish to communicate 476 with. Peers may be identified by names or keys. 478 o Groups of peers may be authorized for a given routing protocol. 480 Establishing a management model is difficult because of the complex 481 relationships between each set of objects. As discussed there may be 482 more than one key for a peer. However in the preshared key case, 483 there may be more than one peer for a key. This is true both for 484 group security association protocols such as an IGP or one-to-one 485 protocols where the same key is used administratively. In some of 486 these situations, it may be undesirable to explicitly enumerate the 487 peers in the configuration; for example IGP peers are auto-discovered 488 for broadcast links but not for non-broadcast multi-access links. 490 Peers may be identified either by name or key. If peers are 491 identified by key it is probably strongly desirable from an 492 operational standpoint to consider any peer identifiers or name to be 493 a local matter and not require the names or identifiers to be 494 synchronized. Obviously if peers are identified by names (for 495 example with certificates in a PKI), identifiers need to be 496 synchronized between the authorized peer and the peer making the 497 authorization decision. 499 In many cases peers will explicitly be identified. In these cases it 500 is possible to attach the authorization information (keys or 501 identifiers) to the peer's configuration object. Two cases do not 502 involve enumerating peers. The first is the case where preshared 503 keys are shared among a group of peers. It is likely that this case 504 can be treated from a management standpoint as a single peer 505 representing all the peers that share the keys. The other case is 506 one where certificates in a PKI are used to introduce peers to a 507 router. In this case, rather than configuring peers, , the router 508 needs to be configured with information on what certificates 509 represent acceptable peers. 511 Another consideration is what routing protocols share peers. For 512 example it may be common for LDP peers to also be peers of some other 513 routing protocol. Also, RSVP-TE may be associated with some TE-based 514 IGP. In some of these cases it would be desirable to use the same 515 authorization information for both routing protocols. 517 In order to develop a management model for authorization, the working 518 group needs to consider several questions. What protocols support 519 auto-discovery of peers? What protocols require more configuration 520 of a peer than simply the peer's authorization information and 521 network address? What management operations are going to be common 522 as security information for peers is configured and updated? What 523 operations will be common while performing key transitions or while 524 migrating to new security technologies? 526 6. Administrator Involvement 528 One key operational question is what areas will administrator 529 involvement be required. Likely areas where involvement may be 530 useful includes enrollment of new peers. Fault recovery should also 531 be considered. 533 6.1. Enrollment 535 One area where the management of routing security needs to be 536 optimized is the deployment of a new router. In some cases a new 537 router may be deployed on an existing network where routing to 538 management servers is already available. In other cases, routers may 539 be deployed as part of connecting or creating a site. Here, the 540 router and infrastructure may not be available until the router has 541 securely authenticated. This problem is similar to the problem of 542 getting initial configuration of routing instances onto the router. 543 However, especially in cases where asymmetric keys or per-peer 544 preshared keys are used, the configuration of other routers needs to 545 be modified to bring up the security association. Also, there has 546 been discussion of generating keys on routers and not allowing them 547 to leave devices. This also impacts what strategies are possible. 548 For example this might mean that routers need to be booted in a 549 secure environment where keys can be generated, and public keys 550 copied to a management server to push out the new public key to 551 potential peers. Then, the router needs to be packaged, moved to 552 where it will be deployed and set up.Alternatives are possible; it is 553 critical that we understand how what we propose impacts operators. 555 We need to work through examples with operators familiar with 556 specific real-world deployment practices and understand how proposed 557 security mechanisms will interact with these practices. 559 6.2. Handling Faults 561 Faults may interact with operational practice in at least two ways. 562 First, security solutions may introduce faults. For example if 563 certificates expire in a PKI, previous adjacencies may no longer 564 form. Operational practice will require a way of repairing these 565 errors. This may end up being very similar to deploying a router 566 that is connecting a new site as the security fault may have 567 partitioned the network. However, unlike a new deployment, the event 568 is unplanned. Strategies such as configuring a router and shipping 569 it to a site may not be appropriate for recovering a fault even 570 though they may be more useful for new deployments. 572 Notifications will play a critical role in avoiding security faults. 573 Implementations SHOULD use appropriate mechanisms to notify operators 574 as security resources are about to expire. Notifications can include 575 messages to consoles, logged events, SNMP traps, or notifications 576 within a routing protocol. One strategy is to have increasing 577 escalations of notifications. 579 Monitoring will also play a important role in avoiding security 580 faults such as certificate expiration. However, the protocols MUST 581 still have adequate operational mechanisms to recover from these 582 situations. Also, some faults, such as those resulting from a 583 compromise or actual attack on a facility are inherent and may not be 584 prevented. 586 A second class of faults is equipment faults that impact security. 587 For example if keys are stored on a router and never moved from that 588 device, failure of a router implies a need to update security 589 provisioning on the replacement router and its peers. 591 To address these operational considerations, we should identify 592 circumstances surrounding recovery from today's faults and understand 593 how protocols will impact mechanisms used today. 595 7. Upgrade Considerations 597 It needs to be possible to deploy automated key management in an 598 organization without either having to disable existing security or 599 disrupting routing. As a result, it needs to be possible to perform 600 a phased upgrade from manual keying to automated key management. 601 This upgrade procedure needs to be easy and have a very low risk of 602 disrupting routing. Today, many operators do not update keys because 603 the perceived risk of an attack is lower than the complexity of and 604 update and risk of routing disruptions. 606 For peer-to-peer protocols such as BGP, this can be relatively easy. 607 First, code that supports automated key management needs to be loaded 608 on both peers. Then the adjacency can be upgraded. The 609 configuration can be updated to switch to automated key management 610 when the second router reboots. Alternatively, if the key management 611 protocols involved can detect that both peers now support automated 612 key management, then a key can potentially be negotiated for an 613 existing session. 615 The situation is more complicated for multicast protocols. It's 616 probably not reasonable to bring down an entire link to reconfigure 617 it as using automated key management. Two approaches should be 618 considered. One is to support key table rows supporting the 619 automated key management and manually configured keying for the same 620 link at the same time. Coordinating this may be tricky. Another 621 possibility is for the automated key management protocol to actually 622 select the same traffic key that is being used manually. This could 623 potentilaly be accomplished by having an option in the key management 624 protocol to export the current manual group key through the automated 625 key management protocol. Then after all nodes are configured with 626 automated key management, manual key entries can be removed. The 627 next re-key after all nodes have manual entries removed will generate 628 a new fresh key. 630 8. Security Considerations 632 This document does not define a protocol. It does discuss the 633 operational and management implications of several security 634 technologies. 636 9. Acknowledgments 638 Funding for Sam Hartman's work on this memo is provided by Huawei. 640 The authors would like to thank Gregory Lebovitz, Russ White and Bill 641 Atwood for valuable reviews. 643 10. References 645 10.1. Normative References 647 [I-D.ietf-karp-crypto-key-table] 648 Housley, R., Polk, T., Hartman, S., and D. Zhang, 649 "Database of Long-Lived Symmetric Cryptographic Keys", 650 draft-ietf-karp-crypto-key-table-03 (work in progress), 651 June 2012. 653 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 654 Requirement Levels", BCP 14, RFC 2119, March 1997. 656 10.2. Informative References 658 [I-D.ietf-karp-design-guide] 659 Lebovitz, G. and M. Bhatia, "Keying and Authentication for 660 Routing Protocols (KARP) Design Guidelines", 661 draft-ietf-karp-design-guide-10 (work in progress), 662 December 2011. 664 [I-D.liu-ospfv3-automated-keying-req] 665 Liu, Y., "OSPFv3 Automated Group Keying Requirements", 666 draft-liu-ospfv3-automated-keying-req-01 (work in 667 progress), July 2007. 669 [I-D.polk-saag-rtg-auth-keytable] 670 Polk, T. and R. Housley, "Routing Authentication Using A 671 Database of Long-Lived Cryptographic Keys", 672 draft-polk-saag-rtg-auth-keytable-05 (work in progress), 673 November 2010. 675 [RFC2328] Moy, J., "OSPF Version 2", STD 54, RFC 2328, April 1998. 677 [RFC4572] Lennox, J., "Connection-Oriented Media Transport over the 678 Transport Layer Security (TLS) Protocol in the Session 679 Description Protocol (SDP)", RFC 4572, July 2006. 681 [RFC4808] Bellovin, S., "Key Change Strategies for TCP-MD5", 682 RFC 4808, March 2007. 684 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 685 Authentication Option", RFC 5925, June 2010. 687 Authors' Addresses 689 Sam Hartman 690 Painless Security 692 Email: hartmans-ietf@mit.edu 694 Dacheng Zhang 695 Huawei 697 Email: zhangdacheng@huawei.com