idnits 2.17.1 draft-ietf-ipsecme-failure-detection-05.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 18, 2011) is 4810 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'IDr' is mentioned on line 306, but not defined == Missing Reference: 'KEi' is mentioned on line 365, but not defined == Missing Reference: 'KEr' is mentioned on line 367, but not defined == Missing Reference: 'CERTREQ' is mentioned on line 586, but not defined ** Obsolete normative reference: RFC 5996 (Obsoleted by RFC 7296) Summary: 1 error (**), 0 flaws (~~), 5 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IPsecME Working Group Y. Nir, Ed. 3 Internet-Draft Check Point 4 Intended status: Standards Track D. Wierbowski 5 Expires: August 22, 2011 IBM 6 F. Detienne 7 P. Sethi 8 Cisco 9 February 18, 2011 11 A Quick Crash Detection Method for IKE 12 draft-ietf-ipsecme-failure-detection-05 14 Abstract 16 This document describes an extension to the IKEv2 protocol that 17 allows for faster detection of SA desynchronization using a saved 18 token. 20 When an IPsec tunnel between two IKEv2 peers is disconnected due to a 21 restart of one peer, it can take as much as several minutes for the 22 other peer to discover that the reboot has occurred, thus delaying 23 recovery. In this text we propose an extension to the protocol, that 24 allows for recovery immediately following the restart. 26 Status of this Memo 28 This Internet-Draft is submitted in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF). Note that other groups may also distribute 33 working documents as Internet-Drafts. The list of current Internet- 34 Drafts is at http://datatracker.ietf.org/drafts/current/. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 This Internet-Draft will expire on August 22, 2011. 43 Copyright Notice 45 Copyright (c) 2011 IETF Trust and the persons identified as the 46 document authors. All rights reserved. 48 This document is subject to BCP 78 and the IETF Trust's Legal 49 Provisions Relating to IETF Documents 50 (http://trustee.ietf.org/license-info) in effect on the date of 51 publication of this document. Please review these documents 52 carefully, as they describe your rights and restrictions with respect 53 to this document. Code Components extracted from this document must 54 include Simplified BSD License text as described in Section 4.e of 55 the Trust Legal Provisions and are provided without warranty as 56 described in the Simplified BSD License. 58 Table of Contents 60 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 61 1.1. Conventions Used in This Document . . . . . . . . . . . . 4 62 2. RFC 5996 Crash Recovery . . . . . . . . . . . . . . . . . . . 5 63 3. Protocol Outline . . . . . . . . . . . . . . . . . . . . . . . 6 64 4. Formats and Exchanges . . . . . . . . . . . . . . . . . . . . 7 65 4.1. Notification Format . . . . . . . . . . . . . . . . . . . 7 66 4.2. Passing a Token in the AUTH Exchange . . . . . . . . . . 8 67 4.3. Replacing Tokens After Rekey or Resumption . . . . . . . 9 68 4.4. Replacing the Token for an Existing SA . . . . . . . . . 10 69 4.5. Presenting the Token in an Unprotected Message . . . . . 10 70 5. Token Generation and Verification . . . . . . . . . . . . . . 11 71 5.1. A Stateless Method of Token Generation . . . . . . . . . 11 72 5.2. A Stateless Method with IP addresses . . . . . . . . . . 12 73 5.3. Token Lifetime . . . . . . . . . . . . . . . . . . . . . 12 74 6. Backup Gateways . . . . . . . . . . . . . . . . . . . . . . . 12 75 7. Interaction with Session Resumption . . . . . . . . . . . . . 13 76 8. Operational Considerations . . . . . . . . . . . . . . . . . . 14 77 8.1. Who should implement this specification . . . . . . . . . 14 78 8.2. Response to unknown child SPI . . . . . . . . . . . . . . 15 79 9. Security Considerations . . . . . . . . . . . . . . . . . . . 16 80 9.1. QCD Token Generation and Handling . . . . . . . . . . . . 16 81 9.2. QCD Token Transmission . . . . . . . . . . . . . . . . . 17 82 9.3. QCD Token Enumeration . . . . . . . . . . . . . . . . . . 17 83 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 84 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 18 85 12. Change Log . . . . . . . . . . . . . . . . . . . . . . . . . . 18 86 12.1. Changes from draft-ietf-ipsecme-failure-detection-04 . . 18 87 12.2. Changes from draft-ietf-ipsecme-failure-detection-03 . . 19 88 12.3. Changes from draft-ietf-ipsecme-failure-detection-02 . . 19 89 12.4. Changes from draft-ietf-ipsecme-failure-detection-01 . . 19 90 12.5. Changes from draft-ietf-ipsecme-failure-detection-00 . . 19 91 12.6. Changes from draft-nir-ike-qcd-07 . . . . . . . . . . . . 19 92 12.7. Changes from draft-nir-ike-qcd-03 and -04 . . . . . . . . 20 93 12.8. Changes from draft-nir-ike-qcd-02 . . . . . . . . . . . . 20 94 12.9. Changes from draft-nir-ike-qcd-01 . . . . . . . . . . . . 20 95 12.10. Changes from draft-nir-ike-qcd-00 . . . . . . . . . . . . 20 96 12.11. Changes from draft-nir-qcr-00 . . . . . . . . . . . . . . 20 97 13. References . . . . . . . . . . . . . . . . . . . . . . . . . . 20 98 13.1. Normative References . . . . . . . . . . . . . . . . . . 20 99 13.2. Informative References . . . . . . . . . . . . . . . . . 21 100 Appendix A. The Path Not Taken . . . . . . . . . . . . . . . . . 21 101 A.1. Initiating a new IKE SA . . . . . . . . . . . . . . . . . 21 102 A.2. SIR . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 103 A.3. Birth Certificates . . . . . . . . . . . . . . . . . . . 22 104 A.4. Reducing Liveness Check Length . . . . . . . . . . . . . 22 105 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 22 107 1. Introduction 109 IKEv2, as described in [RFC5996] and its predecessor RFC 4306, has a 110 method for recovering from a reboot of one peer. As long as traffic 111 flows in both directions, the rebooted peer should re-establish the 112 tunnels immediately. However, in many cases the rebooted peer is a 113 VPN gateway that protects only servers, or else the non-rebooted peer 114 has a dynamic IP address. In such cases, the rebooted peer will not 115 be able to re-establish the tunnels. Section 2 describes how 116 recovery works under RFC 5996, and explains why it may take several 117 minutes. 119 The method proposed here, is to send an octet string, called a "QCD 120 token" in the IKE_AUTH exchange that establishes the tunnel. That 121 token can be stored on the peer as part of the IKE SA. After a 122 reboot, the rebooted implementation can re-generate the token, and 123 send it to the peer, so as to delete the IKE SA. Deleting the IKE SA 124 results is a quick establishment of new IPsec tunnels. This is 125 described in Section 3. 127 1.1. Conventions Used in This Document 129 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 130 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 131 document are to be interpreted as described in [RFC2119]. 133 The term "token" refers to an octet string that an implementation can 134 generate using only the properties of a protected IKE message (such 135 as IKE SPIs) as input. A conforming implementation MUST be able to 136 generate the same token from the same input even after rebooting. 138 The term "token maker" refers to an implementation that generates a 139 token and sends it to the peer as specified in this document. 141 The term "token taker" refers to an implementation that stores such a 142 token or a digest thereof, in order to verify that a new token it 143 receives is identical to the old token it has stored. 145 The term "non-volatile storage" in this document refers to a data 146 storage module, that persists across restarts of the token maker. 147 Examples of such a storage module include an internal disk, an 148 internal flash memory module, an external disk and an external 149 database. A small non-volatile storage module is required for a 150 token maker, but a larger one can be used to enhance performance, as 151 described in Section 8.2. 153 2. RFC 5996 Crash Recovery 155 When one peer loses state or reboots, the other peer does not get any 156 notification, so unidirectional IPsec traffic can still flow. The 157 rebooted peer will not be able to decrypt it, however, and the only 158 remedy is to send an unprotected INVALID_SPI notification as 159 described in section 3.10.1 of [RFC5996]. That section also 160 describes the processing of such a notification: 162 "If this Informational Message is sent outside the 163 context of an IKE_SA, it should be used by the recipient 164 only as a "hint" that something might be wrong (because it 165 could easily be forged)." 167 Since the INVALID_SPI can only be used as a hint, the non-rebooted 168 peer has to determine whether the IPsec SA, and indeed the parent IKE 169 SA are still valid. The method of doing this is described in section 170 2.4 of [RFC5996]. This method, called "liveness check" involves 171 sending a protected empty INFORMATIONAL message, and awaiting a 172 response. This procedure is sometimes referred to as "Dead Peer 173 Detection" or DPD. 175 Section 2.4 does not mandate how many times the liveness check 176 message should be retransmitted, or for how long, but does recommend 177 the following: 179 "It is 180 suggested that messages be retransmitted at least a dozen times over 181 a period of at least several minutes before giving up on an SA..." 183 Those "at least several minutes" are a time during part of which both 184 peers are active, but IPsec cannot be used. 186 Especially in the case of a reboot (rather than fail-over or 187 administrative clearing of state), the peer does not recover 188 immediately. Reboot, depending on the system, may take from a few 189 seconds to a few minutes. This means that at first the peer just 190 goes silent, i.e., does not send or respond to any messages. IKEv2 191 implementations can detect this situation and follow the rules given 192 in section 2.4: 194 If there has only been outgoing traffic on all of 195 the SAs associated with an IKE SA, it is essential to confirm 196 liveness of the other endpoint to avoid black holes. If no 197 cryptographically protected messages have been received on an IKE 198 SA or any of its Child SAs recently, the system needs to perform a 199 liveness check in order to prevent sending messages to a dead peer. 201 [RFC5996] does not mandate any time limits, but it is possible that 202 the peer will start liveness checks even before the other end is 203 sending INVALID_SPI notification, as it detected that the other end 204 is not sending any packets anymore while it is still rebooting or 205 recovering from the situation. 207 This means that the several minutes recovery period is overlaping the 208 actual recover time of the other peer, i.e., if the security gateway 209 requires several minutes to boot up from the crash then the other 210 peers have already finished their liveness checks before the crashing 211 peer even has a chance to send INVALID_SPI notifications. 213 There are cases where the peer loses state and is able to recover 214 immediately; in those cases it might take several minutes to recover. 216 Note that the IKEv2 specification specifically gives no guidance for 217 the number of retries or the length of timeouts, as these do not 218 affect interoperability. This means that implementations are allowed 219 to use the hints provided by the INVALID_SPI messages to shorten 220 those timeouts (i.e., different environment and situation requiring 221 different rules). 223 Some existing IKEv2 implementations already do that (i.e., both 224 shorten timeouts or limit number of retries) based on these kind of 225 hints and also start liveness checks quickly after the other end goes 226 silent. However, see Appendix A.4 for a discussion of why this may 227 not be enough. 229 3. Protocol Outline 231 Supporting implementations will send a notification, called a "QCD 232 token", as described in Section 4.1 in the first IKE_AUTH exchange 233 messages. These are the first IKE_AUTH request and final IKE_AUTH 234 response that contain the AUTH payloads. The generation of these 235 tokens is a local matter for implementations, but considerations are 236 described in Section 5. Implementations that send such a token will 237 be called "token makers". 239 A supporting implementation receiving such a token MUST store it (or 240 a digest thereof) along with the IKE SA. Implementations that 241 support this part of the protocol will be called "token takers". 242 Section 8.1 has considerations for which implementations need to be 243 token takers, and which should be token makers. Implementations that 244 are not token takers will silently ignore QCD tokens. 246 When a token maker receives a protected IKE request message with 247 unknown IKE SPIs, it SHOULD generate a new token that is identical to 248 the previous token, and send it to the requesting peer in an 249 unprotected IKE message as described in Section 4.5. 251 When a token taker receives the QCD token in an unprotected 252 notification, it MUST verify that the TOKEN_SECRET_DATA matches the 253 token stored with the matching IKE SA. If the verification fails, or 254 if the IKE SPIs in the message do not match any existing IKE SA, it 255 SHOULD log the event. If it succeeds, it MUST silently delete the 256 IKE SA associated with the IKE_SPI fields, and all dependent child 257 SAs. This event MAY also be logged. The token taker MUST accept 258 such tokens from any IP address and port combination, so as to allow 259 different kinds of high-availability configurations of the token 260 maker. 262 A supporting token taker MAY immediately create new SAs using an 263 Initial exchange, or it may wait for subsequent traffic to trigger 264 the creation of new SAs. 266 See Section 7 for a short discussion about this extensions's 267 interaction with IKEv2 Session Resumption ([RFC5723]). 269 4. Formats and Exchanges 271 4.1. Notification Format 273 The notification payload called "QCD token" is formatted as follows: 275 1 2 3 276 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 277 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 278 ! Next Payload !C! RESERVED ! Payload Length ! 279 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 280 ! Protocol ID ! SPI Size ! QCD Token Notify Message Type ! 281 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 282 ! ! 283 ~ TOKEN_SECRET_DATA ~ 284 ! ! 285 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 287 o Protocol ID (1 octet) MUST be 1, as this message is related to an 288 IKE SA. 289 o SPI Size (1 octet) MUST be zero, in conformance with section 3.10 290 of [RFC5996]. 291 o QCD Token Notify Message Type (2 octets) - MUST be xxxxx, the 292 value assigned for QCD token notifications. TBA by IANA. 294 o TOKEN_SECRET_DATA (variable) contains a generated token as 295 described in Section 5. 297 4.2. Passing a Token in the AUTH Exchange 299 For brevity, only the EAP version of an AUTH exchange will be 300 presented here. The non-EAP version is very similar. The figures 301 below are based on appendix C.3 of [RFC5996]. 303 first request --> IDi, 304 [N(INITIAL_CONTACT)], 305 [[N(HTTP_CERT_LOOKUP_SUPPORTED)], CERTREQ+], 306 [IDr], 307 [N(QCD_TOKEN)] 308 [CP(CFG_REQUEST)], 309 [N(IPCOMP_SUPPORTED)+], 310 [N(USE_TRANSPORT_MODE)], 311 [N(ESP_TFC_PADDING_NOT_SUPPORTED)], 312 [N(NON_FIRST_FRAGMENTS_ALSO)], 313 SA, TSi, TSr, 314 [V+] 316 first response <-- IDr, [CERT+], AUTH, 317 EAP, 318 [V+] 320 / --> EAP 321 repeat 1..N times | 322 \ <-- EAP 324 last request --> AUTH 326 last response <-- AUTH, 327 [N(QCD_TOKEN)] 328 [CP(CFG_REPLY)], 329 [N(IPCOMP_SUPPORTED)], 330 [N(USE_TRANSPORT_MODE)], 331 [N(ESP_TFC_PADDING_NOT_SUPPORTED)], 332 [N(NON_FIRST_FRAGMENTS_ALSO)], 333 SA, TSi, TSr, 334 [N(ADDITIONAL_TS_POSSIBLE)], 335 [V+] 337 Note that the QCD_TOKEN notification is marked as optional because it 338 is not required by this specification that every implementation be 339 both token maker and token taker. If only one peer sends the QCD 340 token, then a reboot of the other peer will not be recoverable by 341 this method. This may be acceptable if traffic typically originates 342 from the other peer. 344 In any case, the lack of a QCD_TOKEN notification MUST NOT be taken 345 as an indication that the peer does not support this standard. 346 Conversely, if a peer does not understand this notification, it will 347 simply ignore it. Therefore a peer may send this notification 348 freely, even if it does not know whether the other side supports it. 350 The QCD_TOKEN notification is related to the IKE SA and should follow 351 the AUTH payload and precede the Configuration payload and all 352 payloads related to the child SA. 354 4.3. Replacing Tokens After Rekey or Resumption 356 After rekeying an IKE SA, the IKE SPIs are replaced, so the new SA 357 also needs to have a token. If only the responder in the rekey 358 exchange is the token maker, this can be done within the 359 CREATE_CHILD_SA exchange. If the initiator is a token maker, then we 360 need an extra informational exchange. 362 The following figure shows the CREATE_CHILD_SA exchange for rekeying 363 the IKE SA. Only the responder sends a QCD token. 365 request --> SA, Ni, [KEi] 367 response <-- SA, Nr, [KEr], N(QCD_TOKEN) 369 If the initiator is also a token maker, it SHOULD initiate an 370 INFORMATIONAL exchange immediately after the CREATE_CHILD_SA exchange 371 as follows: 373 request --> N(QCD_TOKEN) 375 response <-- 377 For session resumption, as specified in [RFC5723], the situation is 378 similar. The responder, which is necessarily the peer that has 379 crashed, SHOULD send a new ticket within the protected payload of the 380 IKE_SESSION_RESUME exchange. If the Initiator is also a token maker, 381 it needs to send a QCD_TOKEN in a separate INFORMATIONAL exchange. 383 The INFORMATIONAL exchange described in this section can also be used 384 if QCD tokens need to be replaced due to a key rollover. However, 385 since token takers are required to verify at least 4 QCD tokens, this 386 is only necessary if secret QCD keys are rolled over more than four 387 times as often as IKE SAs are rekeyed. See Section 5.1 for an 388 example method that uses secret keys which may require rollover. 390 4.4. Replacing the Token for an Existing SA 392 With some token generation methods, such as that described in 393 Section 5.2, a QCD token may sometimes become invalid, although the 394 IKE SA is still perfectly valid. 396 In such a case, the token maker MUST send the new token in a 397 protected message under that IKE SA. That exchange could be a simple 398 INFORMATIONAL, such as in the last figure in the previous section, or 399 else it can be part of a MOBIKE INFORMATIONAL exchange such as in the 400 following figure taken from section 2.2 of [RFC4555] and modified by 401 adding a QCD_TOKEN notification: 403 (IP_I2:4500 -> IP_R1:4500) 404 HDR, SK { N(UPDATE_SA_ADDRESSES), 405 N(NAT_DETECTION_SOURCE_IP), 406 N(NAT_DETECTION_DESTINATION_IP) } --> 408 <-- (IP_R1:4500 -> IP_I2:4500) 409 HDR, SK { N(NAT_DETECTION_SOURCE_IP), 410 N(NAT_DETECTION_DESTINATION_IP) } 412 <-- (IP_R1:4500 -> IP_I2:4500) 413 HDR, SK { N(COOKIE2), [N(QCD_TOKEN)] } 415 (IP_I2:4500 -> IP_R1:4500) 416 HDR, SK { N(COOKIE2), [N(QCD_TOKEN)] } --> 418 A token taker MUST accept such gratuitous QCD_TOKEN notifications as 419 long as they are carried in protected exchanges. A token maker 420 SHOULD NOT generate them unless it is no longer able to generate the 421 old QCD_TOKEN. 423 4.5. Presenting the Token in an Unprotected Message 425 This QCD_TOKEN notification is unprotected, and is sent as a response 426 to a protected IKE request, which uses an IKE SA that is unknown. 428 message --> N(INVALID_IKE_SPI), N(QCD_TOKEN)+ 430 If child SPIs are persistently mapped to IKE SPIs as described in 431 Section 8.2, a token taker may get the following unprotected message 432 in response to an ESP or AH packet. 434 message --> N(INVALID_SPI), N(QCD_TOKEN)+ 436 The QCD_TOKEN and INVALID_IKE_SPI notifications are sent together to 437 support both implementations that conform to this specification and 438 implementations that don't. Similar to the description in section 439 2.21 of [RFC5996], the IKE SPI and message ID fields in the packet 440 headers are taken from the protected IKE request. 442 To support a periodic rollover of the secret used for token 443 generation, the token taker MUST support at least four QCD_TOKEN 444 notifications in a single packet. The token is considered verified 445 if any of the QCD_TOKEN notifications matches. The token maker MAY 446 generate up to four QCD_TOKEN notifications, based on several 447 generations of keys. 449 If the QCD_TOKEN verifies OK, the receiver MUST silently discard the 450 IKE SA and all associated child SAs. If the QCD_TOKEN cannot be 451 validated, a response MUST NOT be sent, and the event may be logged. 452 Section 5 defines token verification. 454 5. Token Generation and Verification 456 No token generation method is mandated by this document. Two methods 457 are documented in the following sub-sections, but they only serve as 458 examples. 460 The following lists the requirements for a token generation 461 mechanism: 462 o Tokens MUST be at least 16 octets long, and no more than 128 463 octets long, to facilitate storage and transmission. Tokens 464 SHOULD be indistinguishable from random data. 465 o It should not be possible for an external attacker to guess the 466 QCD token generated by an implementation. Cryptographic 467 mechanisms such as PRNG and hash functions are RECOMMENDED. 468 o The token maker MUST be able to re-generate or retrieve the token 469 based on the IKE SPIs even after it reboots. 470 o The method of token generation MUST be such that a collision of 471 QCD tokens between different pairs of IKE SPI will be highly 472 unlikely. 474 5.1. A Stateless Method of Token Generation 476 This describes a stateless method of generating a token: 477 o At installation or immediately after the first boot of the token 478 maker, 32 random octets are generated using a secure random number 479 generator or a PRNG. 480 o Those 32 bytes, called the "QCD_SECRET", are stored in non- 481 volatile storage on the machine, and kept indefinitely. 482 o If key rollover is required by policy, the implementation MAY 483 periodically generate a new QCD_SECRET and keep up to 3 previous 484 generations. When sending an unprotected QCD_TOKEN, as many as 4 485 notification payloads may be sent, each from a different 486 QCD_SECRET. 487 o The TOKEN_SECRET_DATA is calculated as follows: 489 TOKEN_SECRET_DATA = HASH(QCD_SECRET | SPI-I | SPI-R) 491 5.2. A Stateless Method with IP addresses 493 This method is similar to the one in the previous section, except 494 that the IP address of the token taker is also added to the block 495 being hashed. This has the disadvantage that the token needs to be 496 replaced (as described in Section 4.4) whenever the token taker 497 changes its address. 499 See Section 9.2 for a discussion of a use-case for this method. When 500 using this method, the TOKEN_SECRET_DATA field is calculated as 501 follows: 503 TOKEN_SECRET_DATA = HASH(QCD_SECRET | SPI-I | SPI-R | IPaddr-T) 505 The IPaddr-T field specifies the IP address of the token taker. 506 Secret rollover considerations are similar to those in the previous 507 section. 509 5.3. Token Lifetime 511 The token is associated with a single IKE SA, and SHOULD be deleted 512 by the token taker when the SA is deleted or expires. More formally, 513 the token is associated with the pair (SPI-I, SPI-R). 515 6. Backup Gateways 517 Making crash detection and recovery quick is a worthy goal, but since 518 rebooting a gateway takes a non-zero amount of time, many 519 implementations choose to have a stand-by gateway ready to take over 520 as soon as the primary gateway fails for any reason. [cluster] 521 describes considerations for such clusters of gateways with 522 synchronized state, but the rest of this section is relevant even 523 when there is no synchronized state. 525 If such a configuration is available, it is RECOMMENDED that the 526 stand-by gateway be able to generate the same token as the active 527 gateway. if the method described in Section 5.1 is used, this means 528 that the QCD_SECRET field is identical in both gateways. This has 529 the effect of having the crash recovery available immediately. 531 Note that this refers to "high availability" configurations, where 532 only one gateway is active at any given moment. This is different 533 from "load sharing" configurations where more than one gateway is 534 active at the same time. For load sharing configurations, please see 535 Section 9.2 for security considerations. 537 7. Interaction with Session Resumption 539 Session resumption, specified in [RFC5723], allows the setting up of 540 a new IKE SA to consume less computing resources. This is 541 particularly useful in the case of a remote access gateway that has 542 many tunnels. A failure of such a gateway requires all these many 543 remote access clients to establish an IKE SA either with the rebooted 544 gateway or with a backup. This tunnel re-establishment occurs within 545 a short period of time, creating a burden on the remote access 546 gateway. Session resumption addresses this problem by having the 547 clients store an encrypted derivative of the IKE SA for quick re- 548 establishment. 550 What Session Resumption does not help is the problem of detecting 551 that the peer gateway has failed. A failed gateway may go undetected 552 for an arbitrarily long time, because IPsec does not have packet 553 acknowledgement, and applications cannot signal the IPsec layer that 554 the tunnel "does not work". Section 2.4 of RFC 5996 does not specify 555 how long an implementation needs to wait before beginning a liveness 556 check, and only says "not recently" (see full quote in Section 2). 557 In practice some mobile devices wait a very long time before 558 beginning liveness check, in order to extend battery life by allowing 559 parts of the device to remain in low-power modes. 561 QCD tokens provide a way to detect the failure of the peer in the 562 case where liveness check has not yet ended (or begun). 564 A remote access client conforming to both specifications will store 565 QCD tokens, as well as the Session Resumption ticket, if provided by 566 the gateway. A remote access gateway conforming to both 567 specifications will generate a QCD token for the client. When the 568 gateway reboots, the client will discover this in either of two ways: 569 1. The client does regular liveness checks, or else the time for 570 some other IKE exchange has come. Since the gateway is still 571 down, the IKE exchange times out after several minutes. In this 572 case QCD does not help. 574 2. Either the primary gateway or a backup gateway (see Section 6) is 575 ready and sends a QCD token to the client. In that case the 576 client will quickly re-establish the IPsec tunnel, either with 577 the rebooted primary gateway or the backup gateway as described 578 in this document. 580 The full combined protocol looks like this: 582 Initiator Responder 583 ----------- ----------- 584 HDR, SAi1, KEi, Ni --> 586 <-- HDR, SAr1, KEr, Nr, [CERTREQ] 588 HDR, SK {IDi, [CERT,] 589 [CERTREQ,] [IDr,] 590 AUTH, N(QCD_TOKEN) 591 SAi2, TSi, TSr, 592 N(TICKET_REQUEST)} --> 593 <-- HDR, SK {IDr, [CERT,] AUTH, 594 N(QCD_TOKEN), SAr2, TSi, TSr, 595 N(TICKET_LT_OPAQUE) } 597 ---- Reboot ----- 599 HDR, {} --> 600 <-- HDR, N(QCD_TOKEN) 602 HDR, [N(COOKIE),] 603 Ni, N(TICKET_OPAQUE) 604 [,N+] --> 605 <-- HDR, Nr [,N+] 607 8. Operational Considerations 609 8.1. Who should implement this specification 611 Throughout this document, we have referred to reboot time 612 alternatingly as the time that the implementation crashes and the 613 time when it is ready to process IPsec packets and IKE exchanges. 614 Depending on the hardware and software platforms and the cause of the 615 reboot, rebooting may take anywhere from a few seconds to several 616 minutes. If the implementation is down for a long time, the benefit 617 of this protocol extension is reduced. For this reason critical 618 systems should implement backup gateways as described in Section 6. 620 Implementing the "token maker" side of QCD makes sense for IKE 621 implementation where protected connections originate from the peer, 622 such as inter-domain VPNs and remote access gateways. Implementing 623 the "token taker" side of QCD makes sense for IKE implementations 624 where protected connections originate, such as inter-domain VPNs and 625 remote access clients. 627 To clarify the this discussion: 628 o For remote-access clients it makes sense to implement the token 629 taker role. 630 o For remote-access gateways it makes sense to implement the token 631 maker role. 632 o For inter-domain VPN gateways it makes sense to implement both 633 roles, because it can't be known in advance where the traffic 634 originates. 635 o It is perfectly valid to implement both roles in any case, for 636 example when using a single library or a single gateway to perform 637 several roles. 639 In order to limit the effects of DoS attacks, a token taker SHOULD 640 limit the rate of QCD_TOKENs verified from a particular source. 642 If excessive amounts of IKE requests protected with unknown IKE SPIs 643 arrive at a token maker, the IKE module SHOULD revert to the behavior 644 described in section 2.21 of [RFC5996] and either send an 645 INVALID_IKE_SPI notification, or ignore it entirely. 647 8.2. Response to unknown child SPI 649 After a reboot, it is more likely that an implementation receives 650 IPsec packets than IKE packets. In that case, the rebooted 651 implementation will send an INVALID_SPI notification, triggering a 652 liveness check. The token will only be sent in a response to the 653 liveness check, thus requiring an extra round-trip. 655 To avoid this, an implementation that has access to enough non- 656 volatile storage MAY store a mapping of child SPIs to owning IKE 657 SPIs, or to generated tokens. If such a mapping is available and 658 persistent across reboots, the rebooted implementation SHOULD respond 659 to the IPsec packet with an INVALID_SPI notification, along with the 660 appropriate QCD_Token notifications. A token taker SHOULD verify the 661 QCD token that arrives with an INVALID_SPI notification the same as 662 if it arrived with the IKE SPIs of the parent IKE SA. 664 However, a persistent storage module might not be updated in a timely 665 manner, and could be populated with tokens relating to IKE SPIs that 666 have already been rekeyed. A token taker MUST NOT take an invalid 667 QCD Token sent along with an INVALID_SPI notification as evidence 668 that the peer is either malfunctioning or attacking, but it SHOULD 669 limit the rate at which such notifications are processed. 671 9. Security Considerations 673 The extension described in this document must not reduce the security 674 of IKEv2 or IPsec. Specifically, an eavesdropper must not learn any 675 non-public information about the peers. 677 The proposed mechanism should be secure against attacks by a passive 678 MITM (eavesdropper). Such an attacker must not be able to disrupt an 679 existing IKE session, either by resetting the session or by 680 introducing significant delays. This requirement is especially 681 significant, because this document introduces a new way to reset an 682 IKE SA. 684 The mechanism need not be similarly secure against an active MITM, 685 since this type of attacker is already able to disrupt IKE sessions. 687 9.1. QCD Token Generation and Handling 689 Tokens MUST be hard to guess. This is critical, because if an 690 attacker can guess the token associated with an IKE SA, she can tear 691 down the IKE SA and associated tunnels at will. When the token is 692 delivered in the IKE_AUTH exchange, it is encrypted. When it is sent 693 again in an unprotected notification, it is not, but that is the last 694 time this token is ever used. 696 An aggregation of some tokens generated by one maker together with 697 the related IKE SPIs MUST NOT give an attacker the ability to guess 698 other tokens. Specifically, if one taker does not properly secure 699 the QCD tokens and an attacker gains access to them, this attacker 700 MUST NOT be able to guess other tokens generated by the same maker. 701 This is the reason that the QCD_SECRET in Section 5.1 needs to be 702 sufficiently long. 704 The token taker MUST store the token in a secure manner. No attacker 705 should be able to gain access to a stored token. 707 The QCD_SECRET MUST be protected from access by other parties. 708 Anyone gaining access to this value will be able to delete all the 709 IKE SAs for this token maker. 711 The QCD token is sent by the rebooted peer in an unprotected message. 712 A message like that is subject to modification, deletion and replay 713 by an attacker. However, these attacks will not compromise the 714 security of either side. Modification is meaningless because a 715 modified token is simply an invalid token. Deletion will only cause 716 the protocol not to work, resulting in a delay in tunnel re- 717 establishment as described in Section 2. Replay is also meaningless, 718 because the IKE SA has been deleted after the first transmission. 720 9.2. QCD Token Transmission 722 A token maker MUST NOT send a valid QCD token in an unprotected 723 message for an existing IKE SA. 725 This requirement is obvious and easy in the case of a single gateway. 726 However, some implementations use a load balancer to divide the load 727 between several physical gateways. It MUST NOT be possible even in 728 such a configuration to trick one gateway into sending a valid QCD 729 token for an IKE SA which is valid on another gateway. This is true 730 whether the attempt to trick the gateway uses the token taker's IP 731 address or a different IP address. 733 IPsec Failure Detection is not applicable to deployments where the 734 QCD secret is shared by multiple gateways and the gateways cannot 735 assess whether the token can be legitimately sent in the clear while 736 another gateway may actually still own the SA's. Load balancer 737 configurations typically fall in this category. In order for a load 738 balancing configuration of IPsec gateways to support this 739 specification, all members MUST be able to tell whether a particular 740 IKE SA is active anywhere in the cluster. One way to do it is to 741 synchronize a list of active IKE SPIs among all the cluster members. 743 Because it includes the token taker's IP address in the token 744 generation, the method in Section 5.2 can (under certain conditions) 745 prevent revealing the QCD token for an existing pair of IKE SPIs to 746 an attacker who is using a different IP address, even in a load- 747 sharing cluster without state synchronization. This method does not 748 prevent revealing the QCD token to an active attacker who is spoofing 749 the token taker's IP address. Such an attacker may attempt to direct 750 messages to a cluster member other than the member responsible for 751 the IKE SA in an attempt to trick that gateway into sending a QCD 752 token for a valid IKE SA. This method should not be used unless the 753 load balancer guarantees that IKE packets from the same source IP 754 address always go to the same cluster member. 756 9.3. QCD Token Enumeration 758 An attacker may try to attack QCD if the generation algorithm 759 described in Section 5.1 is used. The attacker will send several 760 fake IKE requests to the gateway under attack, receiving and 761 recording the QCD Tokens in the responses. This will allow the 762 attacker to create a dictionary of IKE SPIs to QCD Tokens, which can 763 later be used to tear down any IKE SA. 765 Three factors mitigate this threat: 766 o The space of all possible IKE SPI pairs is huge: 2^128, so making 767 such a dictionary is impractical. Even if we assume that one 768 implementation always generates predictable IKE SPIs, the space is 769 still at least 2^64 entries, so making the dictionary is extremely 770 hard. To ensure this, token makers MUST generate unpredictable 771 IKE SPIs by using a cryptographically strong pseudo-random number 772 generator. 773 o Throttling the amount of QCD_TOKEN notifications sent out, as 774 discussed in Section 8.1, especially when not soon after a crash 775 will limit the attacker's ability to construct a dictionary. 776 o The methods in Section 5.1 and Section 5.2 allow for a periodic 777 change of the QCD_SECRET. Any such change invalidates the entire 778 dictionary. 780 10. IANA Considerations 782 IANA is requested to assign a notify message type from the status 783 types range (16406-40959) of the "IKEv2 Notify Message Types" 784 registry with name "QUICK_CRASH_DETECTION". 786 11. Acknowledgements 788 We would like to thank Hannes Tschofenig and Yaron Sheffer for their 789 comments about Session Resumption. 791 Others who have contributed valuable comments are, in alphabetical 792 order, Lakshminath Dondeti, Paul Hoffman, Tero Kivinen, Scott C 793 Moonen, and Keith Welter. 795 12. Change Log 797 This section lists all changes in this document 799 NOTE TO RFC EDITOR : Please remove this section in the final RFC 801 12.1. Changes from draft-ietf-ipsecme-failure-detection-04 803 o Some more rephrasing of section 9.2 based on suggestions by Tero 804 Kivinen and Dave Wierbowski. 806 12.2. Changes from draft-ietf-ipsecme-failure-detection-03 808 o Merged section 9.4 into section 9.2. 809 o Multiple typos discovered by Scott Moonen, Keith Welter and Yaron. 811 12.3. Changes from draft-ietf-ipsecme-failure-detection-02 813 o Moved section 7 to Appendix A. Also changed some wording. 814 o Fixed some language in the "interaction with session resumption" 815 section to say that although liveness check MUST be done, there 816 are no time limits to how long an implementation takes before 817 starting liveness check, or ending it. 819 12.4. Changes from draft-ietf-ipsecme-failure-detection-01 821 o Fixed the language requiring random IKE SPIs. 822 o Some better explanation of the reasons to choose the methods in 823 Section 5.2 and the method in Section 5.1, to close issue #193. 824 o Added text to the beginning of Section 9 to accomodate issue #194. 826 12.5. Changes from draft-ietf-ipsecme-failure-detection-00 828 o Nits pointed out by Scott and Yaron. 829 o Pratima and Frederic are back on board. 830 o Changed IKEv2bis draft reference to RFC 5996. 831 o Resolved issues #189, #190, #191, and #192: 832 * Renamed section 4.5 and removed the requirement to send an 833 acknowledgement for the unprotected message. 834 * Moved the QCD token from the last to the first IKE_AUTH 835 request. 836 * Added a MUST to Section 9.3 to require that IKE SPIs be 837 randomly generated. 838 * Changed the language in Section 8.1, to not use RFC 2119 839 terminology. 840 * Moved the section describing why one would want the method 841 dependant on IP addresses (in Section 5.2 from operational 842 considerations to security considerations. 844 12.6. Changes from draft-nir-ike-qcd-07 846 o First WG version. 847 o Addressed Scott C Moonen's concern about collisions of QCD tokens. 848 o Updated references to point to IKEv2bis instead of RFC 4306 and 849 4718. Also converted draft reference for resumption to RFC 5723. 850 o Added Dave Wiebrowski as author, and removed Pratima and Frederic. 852 12.7. Changes from draft-nir-ike-qcd-03 and -04 854 Mostly editorial changes and cleaning up. 856 12.8. Changes from draft-nir-ike-qcd-02 858 o Described QCD token enumeration, following a question by 859 Lakshminath Dondeti. 860 o Added the ability to replace the QCD token for an existing IKE SA. 861 o Added tokens dependent on peer IP address and their interaction 862 with MOBIKE. 864 12.9. Changes from draft-nir-ike-qcd-01 866 o Removed stateless method. 867 o Added discussion of rekeying and resumption. 868 o Added discussion of non-synchronized load-balanced clusters of 869 gateways in the security considerations. 870 o Other wording fixes. 872 12.10. Changes from draft-nir-ike-qcd-00 874 o Merged proposal with draft-detienne-ikev2-recovery 875 o Changed the protocol so that the rebooted peer generates the 876 token. This has the effect, that the need for persistent storage 877 is eliminated. 878 o Added discussion of birth certificates. 880 12.11. Changes from draft-nir-qcr-00 882 o Changed name to reflect that this relates to IKE. Also changed 883 from quick crash recovery to quick crash detection to avoid 884 confusion with IFARE. 885 o Added more operational considerations. 886 o Added interaction with IFARE. 887 o Added discussion of backup gateways. 889 13. References 891 13.1. Normative References 893 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 894 Requirement Levels", BCP 14, RFC 2119, March 1997. 896 [RFC4555] Eronen, P., "IKEv2 Mobility and Multihoming Protocol 897 (MOBIKE)", RFC 4555, June 2006. 899 [RFC5996] Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen, 900 "Internet Key Exchange Protocol: IKEv2", RFC 5996, 901 September 2010. 903 13.2. Informative References 905 [RFC5723] Sheffer, Y. and H. Tschofenig, "IKEv2 Session Resumption", 906 RFC 5723, January 2010. 908 [cluster] Nir, Y., Ed., "IPsec Cluster Problem Statement", 909 draft-ietf-ipsecme-ipsec-ha (work in progress), July 2010. 911 [recovery] 912 Detienne, F., Sethi, P., and Y. Nir, "Safe IKE Recovery", 913 draft-detienne-ikev2-recovery (work in progress), 914 January 2010. 916 Appendix A. The Path Not Taken 918 A.1. Initiating a new IKE SA 920 Instead of sending a QCD token, we could have the rebooted 921 implementation start an Initial exchange with the peer, including the 922 INITIAL_CONTACT notification. This would have the same effect, 923 instructing the peer to erase the old IKE SA, as well as establishing 924 a new IKE SA with fewer rounds. 926 The disadvantage here, is that in IKEv2 an authentication exchange 927 MUST have a piggy-backed Child SA set up. Since our use case is such 928 that the rebooted implementation does not have traffic flowing to the 929 peer, there are no good selectors for such a Child SA. 931 Additionally, when authentication is asymmetric, such as when EAP is 932 used, it is not possible for the rebooted implementation to initiate 933 IKE. 935 A.2. SIR 937 Another proposal that was considered for this work item is the SIR 938 extension, which is described in [recovery]. Under that proposal, 939 the non-rebooted peer sends a non-protected query to the possibly 940 rebooted peer, asking whether the IKE SA exists. The peer replies 941 with either a positive or negative response, and the absence of a 942 positive response, along with the existence of a negative response is 943 taken as proof that the IKE SA has really been lost. 945 The working group preferred the QCD proposal to this one. 947 A.3. Birth Certificates 949 Birth Certificates is a method of crash detection that has never been 950 formally defined. Bill Sommerfeld suggested this idea in a mail to 951 the IPsec mailing list on August 7, 2000, in a thread discussing 952 methods of crash detection: 954 If we have the system sign a "birth certificate" when it 955 reboots (including a reboot time or boot sequence number), 956 we could include that with a "bad spi" ICMP error and in 957 the negotiation of the IKE SA. 959 We believe that this method would have some problems. First, it 960 requires Alice to store the certificate, so as to be able to compare 961 the public keys. That requires more storage than does a QCD token. 962 Additionally, the public-key operations needed to verify the self- 963 signed certificates are more expensive for Alice. 965 We believe that a symmetric-key operation such as proposed here is 966 more light-weight and simple than that implied by the Birth 967 Certificate idea. 969 A.4. Reducing Liveness Check Length 971 Some implementations require fewer retransmissions over a shorter 972 period of time for cases of liveness check started because of an 973 INVALID_SPI or INVALID_IKE_SPI notification. 975 We believe that the default retransmission policy should represent a 976 good balance between the need for a timely discovery of a dead peer, 977 and a low probability of false detection. We expect the policy to be 978 set to take the shortest time such that this probability achieves a 979 certain target. Therefore, we believe that reducing the elapsed time 980 and retransmission count may create an unacceptably high probability 981 of false detection, and this can be triggered by a single 982 INVALID_IKE_SPI notification. 984 Additionally, even if the retransmission policy is reduced to, say, 985 one minute, it is still a very noticeable delay from a human 986 perspective, from the time that the gateway has come up (i.e., is 987 able to respond with an INVALID_SPI or INVALID_IKE_SPI notification) 988 and until the tunnels are active, or from the time the backup gateway 989 has taken over until the tunnels are active. The use of QCD tokens 990 can reduce this delay. 992 Authors' Addresses 994 Yoav Nir (editor) 995 Check Point Software Technologies Ltd. 996 5 Hasolelim st. 997 Tel Aviv 67897 998 Israel 1000 Email: ynir@checkpoint.com 1002 David Wierbowski 1003 International Business Machines 1004 1701 North Street 1005 Endicott, New York 13760 1006 United States 1008 Email: wierbows@us.ibm.com 1010 Frederic Detienne 1011 Cisco Systems, Inc. 1012 De Kleetlaan, 7 1013 Diegem B-1831 1014 Belgium 1016 Phone: +32 2 704 5681 1017 Email: fd@cisco.com 1019 Pratima Sethi 1020 Cisco Systems, Inc. 1021 O'Shaugnessy Road, 11 1022 Bangalore, Karnataka 560027 1023 India 1025 Phone: +91 80 4154 1654 1026 Email: psethi@cisco.com