idnits 2.17.1 draft-ietf-nsis-rsvp-sec-properties-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1.a on line 18. -- Found old boilerplate from RFC 3978, Section 5.5 on line 2027. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 2004. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 2011. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 2017. ** The document seems to lack an RFC 3978 Section 5.1 IPR Disclosure Acknowledgement. ** This document has an original RFC 3978 Section 5.4 Copyright Line, instead of the newer IETF Trust Copyright according to RFC 4748. ** This document has an original RFC 3978 Section 5.5 Disclaimer, instead of the newer disclaimer which includes the IETF Trust according to RFC 4748. ** The document uses RFC 3667 boilerplate or RFC 3978-like boilerplate instead of verbatim RFC 3978 boilerplate. After 6 May 2005, submission of drafts without verbatim RFC 3978 boilerplate is not accepted. The following non-3978 patterns matched text found in the document. That text should be removed or replaced: This document is an Internet-Draft and is subject to all provisions of Section 3 of RFC 3667. By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** There is 1 instance of lines with control characters in the document. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the RFC 3978 Section 5.4 Copyright Line does not match the current year -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (February 20, 2005) is 6977 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: '16' is defined on line 1736, but no explicit reference was found in the text == Unused Reference: '17' is defined on line 1739, but no explicit reference was found in the text == Unused Reference: '49' is defined on line 1856, but no explicit reference was found in the text == Unused Reference: '50' is defined on line 1861, but no explicit reference was found in the text -- Possible downref: Non-RFC (?) normative reference: ref. '1' -- Possible downref: Non-RFC (?) normative reference: ref. '2' -- Possible downref: Non-RFC (?) normative reference: ref. '3' -- Possible downref: Non-RFC (?) normative reference: ref. '4' -- Possible downref: Non-RFC (?) normative reference: ref. '5' -- Possible downref: Non-RFC (?) normative reference: ref. '6' -- Possible downref: Non-RFC (?) normative reference: ref. '7' -- Possible downref: Non-RFC (?) normative reference: ref. '8' ** Obsolete normative reference: RFC 3588 (ref. '9') (Obsoleted by RFC 6733) -- Possible downref: Non-RFC (?) normative reference: ref. '10' -- Possible downref: Non-RFC (?) normative reference: ref. '11' -- Possible downref: Non-RFC (?) normative reference: ref. '13' -- Possible downref: Non-RFC (?) normative reference: ref. '14' == Outdated reference: A later version (-34) exists of draft-ietf-cat-kerberos-pk-init-24 == Outdated reference: A later version (-11) exists of draft-ietf-kink-kink-06 == Outdated reference: A later version (-33) exists of draft-ietf-pkix-scvp-11 == Outdated reference: A later version (-04) exists of draft-manner-lrsvp-00 -- No information found for draft-ietf-rsvp-lpm-arch - is the name correct? -- Duplicate reference: draft-ietf-rap-new-rsvp-ext, mentioned in '49', was also mentioned in '18'. Summary: 7 errors (**), 0 flaws (~~), 10 warnings (==), 21 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 NSIS H. Tschofenig 3 Internet-Draft Siemens 4 Expires: August 21, 2005 R. Graveman 5 RFG Security 6 February 20, 2005 8 RSVP Security Properties 9 draft-ietf-nsis-rsvp-sec-properties-06.txt 11 Status of this Memo 13 This document is an Internet-Draft and is subject to all provisions 14 of section 3 of RFC 3667. By submitting this Internet-Draft, each 15 author represents that any applicable patent or other IPR claims of 16 which he or she is aware have been or will be disclosed, and any of 17 which he or she become aware will be disclosed, in accordance with 18 RFC 3668. 20 Internet-Drafts are working documents of the Internet Engineering 21 Task Force (IETF), its areas, and its working groups. Note that 22 other groups may also distribute working documents as 23 Internet-Drafts. 25 Internet-Drafts are draft documents valid for a maximum of six months 26 and may be updated, replaced, or obsoleted by other documents at any 27 time. It is inappropriate to use Internet-Drafts as reference 28 material or to cite them other than as "work in progress." 30 The list of current Internet-Drafts can be accessed at 31 http://www.ietf.org/ietf/1id-abstracts.txt. 33 The list of Internet-Draft Shadow Directories can be accessed at 34 http://www.ietf.org/shadow.html. 36 This Internet-Draft will expire on August 21, 2005. 38 Copyright Notice 40 Copyright (C) The Internet Society (2005). 42 Abstract 44 This document summarizes the security properties of RSVP. The goal 45 of this analysis is to benefit from previous work done on RSVP and to 46 capture knowledge about past activities. 48 Table of Contents 50 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 51 2. Terminology and Architectural Assumptions . . . . . . . . . . 4 52 3. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 53 3.1 The RSVP INTEGRITY Object . . . . . . . . . . . . . . . . 6 54 3.2 Security Associations . . . . . . . . . . . . . . . . . . 8 55 3.3 RSVP Key Management Assumptions . . . . . . . . . . . . . 9 56 3.4 Identity Representation . . . . . . . . . . . . . . . . . 9 57 3.5 RSVP Integrity Handshake . . . . . . . . . . . . . . . . . 13 58 4. Detailed Security Property Discussion . . . . . . . . . . . . 15 59 4.1 Network Topology . . . . . . . . . . . . . . . . . . . . . 15 60 4.2 Host/Router . . . . . . . . . . . . . . . . . . . . . . . 15 61 4.3 User to PEP/PDP . . . . . . . . . . . . . . . . . . . . . 19 62 4.4 Communication between RSVP-Aware Routers . . . . . . . . . 26 63 5. Miscellaneous Issues . . . . . . . . . . . . . . . . . . . . . 29 64 5.1 First Hop Issue . . . . . . . . . . . . . . . . . . . . . 29 65 5.2 Next-Hop Problem . . . . . . . . . . . . . . . . . . . . . 29 66 5.3 Last-Hop Issue . . . . . . . . . . . . . . . . . . . . . . 32 67 5.4 RSVP and IPsec protected data traffic . . . . . . . . . . 33 68 5.5 End-to-End Security Issues and RSVP . . . . . . . . . . . 35 69 5.6 IPsec protection of RSVP signaling messages . . . . . . . 35 70 5.7 Authorization . . . . . . . . . . . . . . . . . . . . . . 36 71 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 37 72 7. Security Considerations . . . . . . . . . . . . . . . . . . . 39 73 8. IANA considerations . . . . . . . . . . . . . . . . . . . . . 40 74 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 41 75 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 42 76 10.1 Normative References . . . . . . . . . . . . . . . . . . . . 42 77 10.2 Informative References . . . . . . . . . . . . . . . . . . . 43 78 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 45 79 A. Dictionary Attacks and Kerberos . . . . . . . . . . . . . . . 47 80 B. Example of User-to-PDP Authentication . . . . . . . . . . . . 48 81 C. Literature on RSVP Security . . . . . . . . . . . . . . . . . 49 82 Intellectual Property and Copyright Statements . . . . . . . . 50 84 1. Introduction 86 As the work of the NSIS working group has begun, there are also 87 concerns about security and its implications for the design of a 88 signaling protocol. In order to understand the security properties 89 and available options of RSVP a number of documents have to be read. 90 This document summarizes the security properties of RSVP and is part 91 of the overall process of analyzing other signaling protocols and 92 learning from their design considerations. This document should also 93 provide a starting point for further discussions. 95 The content of this document is organized as follows: 97 Section 3 provides an overview of the security mechanisms provided by 98 RSVP including the INTEGRITY object, a description of the identity 99 representation within the POLICY_DATA object (i.e., user 100 authentication), and the RSVP Integrity Handshake mechanism. Section 101 4 provides a more detailed discussion of the mechanisms used and 102 tries to describe in detail the mechanisms provided. 104 RSVP also supports multicast but this document does not address 105 security aspects for supporting multicast QoS signaling. Multicast 106 is currently outside the scope of the NSIS working group. 108 Although a variation of RSVP, namely RSVP-TE, is used in the context 109 of MPLS to distribute labels for a label switched path its usage is 110 different than the usage scenarios envisioned for NSIS. Hence, this 111 document does not address RSVP-TE and the security properties of it. 113 2. Terminology and Architectural Assumptions 115 This section describes some important terms and explains some 116 architectural assumptions: 118 Chain-of-Trust: 120 The security mechanisms supported by RSVP [1] heavily rely on 121 optional hop-by-hop protection using the built-in INTEGRITY 122 object. Hop-by-hop security with the INTEGRITY object inside the 123 RSVP message thereby refers to the protection between 124 RSVP-supporting network elements. Additionally, there is the 125 notion of policy-aware network elements that understand the 126 POLICY_DATA element within the RSVP message. Because this element 127 also includes an INTEGRITY object, there is an additional 128 hop-by-hop security mechanism that provides security between 129 policy-aware nodes. Policy-ignorant nodes are not affected by the 130 inclusion of this object in the POLICY_DATA element, because they 131 do not try to interpret it. 133 To protect signaling messages that are possibly modified by each 134 RSVP router along the path, it must be assumed that each incoming 135 request is authenticated, integrity protected, and replay 136 protected. This provides protection against unauthorized nodes' 137 injecting bogus messages. Furthermore, each RSVP-aware router is 138 assumed to behave in the expected manner. Outgoing messages 139 transmitted to the next hop network element receive protection 140 according RSVP security processing. 142 Using the above described mechanisms, a chain-of-trust is created 143 whereby a signaling message transmitted by router A via router B 144 and received by router C is supposed to be secure if routers A and 145 B and routers B and C share security associations and all routers 146 behave as expected. Hence router C trusts router A although 147 router C does not have a direct security association with router 148 A. We can therefore conclude that the protection achieved with 149 this hop-by-hop security for the chain-of-trust is no better than 150 the weakest link in the chain. 152 If one router is malicious (for example because an adversary has 153 control over this router), then it can arbitrarily modify 154 messages, cause unexpected behavior, and mount a number of attacks 155 not limited only to QoS signaling. Additionally, it must be 156 mentioned that some protocols demand more protection than others 157 (which depends in part on which nodes are executing these 158 protocols). For example, edge devices, where end-users are 159 attached, may more likely be attacked in comparison with the more 160 secure core network of a service provider. In some cases a 161 network service provider may choose not to use the RSVP-provided 162 security mechanisms inside the core network because a different 163 security protection is deployed. 165 Section 6 of [2] mentions the term chain-of-trust in the context 166 of RSVP integrity protection. In Section 6 of [18] the same term 167 is used in the context of user authentication with the INTEGRITY 168 object inside the POLICY_DATA element . Unfortunately the term is 169 not explained in detail and the assumptions behind it are not 170 clearly specified. 172 Host and User Authentication: 174 The presence of RSVP protection and a separate user identity 175 representation leads to the fact that both user-identity and 176 host-identity are used for RSVP protection. Therefore, user-based 177 security and host-based security are covered separately, because 178 of the different authentication mechanisms provided. To avoid 179 confusion about the different concepts, Section 3.4 describes the 180 concept of user authentication in more detail. 182 Key Management: 184 It is assumed that most of the security associations required for 185 the protection of RSVP signaling messages are already available, 186 and hence key management was done in advance. There is, however, 187 an exception with respect to support for Kerberos. Using 188 Kerberos, an entity is able to distribute a session key used for 189 RSVP signaling protection. 191 RSVP INTEGRITY and POLICY_DATA INTEGRITY Objects: 193 RSVP uses an INTEGRITY object in two places in a message. The 194 first is in the RSVP message itself and covers the entire RSVP 195 message as defined in [1]. The second is included in the 196 POLICY_DATA object and defined in [2]. To differentiate the two 197 objects regarding their scope of protection, the two terms RSVP 198 INTEGRITY and POLICY_DATA INTEGRITY object are used, respectively. 199 The data structure of the two objects, however, is the same. 201 Hop versus Peer: 203 In the past, the terminology for nodes addressed by RSVP has been 204 discussed considerably. In particular, two favorite terms have 205 been used: hop and peer. This document uses the term hop, which 206 is different from an IP hop. Two neighboring RSVP nodes 207 communicating with each other are not necessarily neighboring IP 208 nodes (i.e., they may be more than one IP hop away). 210 3. Overview 212 This section describes the security mechanisms provided by RSVP. 213 Although use of IPsec is mentioned in Section 10 of [1], the security 214 mechanisms primarily envisioned for RSVP are described. 216 3.1 The RSVP INTEGRITY Object 218 The RSVP INTEGRITY object is the major component of RSVP security 219 protection. This object is used to provide integrity and replay 220 protection for the content of the signaling message between two RSVP 221 participating routers or between an RSVP router and host. 222 Furthermore, the RSVP INTEGRITY object provides data origin 223 authentication. The attributes of the object are briefly described: 225 Flags field: 227 The Handshake Flag is the only defined flag. It is used to 228 synchronize sequence numbers if the communication gets out of sync 229 (e.g., it allows a restarting host to recover the most recent 230 sequence number). Setting this flag to one indicates that the 231 sender is willing to respond to an Integrity Challenge message. 232 This flag can therefore be seen as a negotiation capability 233 transmitted within each INTEGRITY object. 235 Key Identifier: 237 The Key Identifier selects the key used for verification of the 238 Keyed Message Digest field and, hence, must be unique for the 239 sender. It has a fixed 48-bit length. The generation of this Key 240 Identifier field is mostly a decision of the local host. [1] 241 describes this field as a combination of an address, sending 242 interface, and key number. We assume that the Key Identifier is 243 simply a (keyed) hash value computed over a number of fields with 244 the requirement to be unique if more than one security association 245 is used in parallel between two hosts (e.g., as is the case with 246 security associations having overlapping lifetimes). A receiving 247 system uniquely identifies a security association based on the Key 248 Identifier and the sender's IP address. The sender's IP address 249 may be obtained from the RSVP_HOP object or from the source IP 250 address of the packet if the RSVP_HOP object is not present. The 251 sender uses the outgoing interface to determine which security 252 association to use. The term outgoing interface may be confusing. 253 The sender selects the security association based on the 254 receiver's IP address (i.e., the address of the next RSVP-capable 255 router). The process of determining which node is the next 256 RSVP-capable router is not further specified and is likely to be 257 statically configured. 259 Sequence Number: 261 The sequence number used by the INTEGRITY object is 64 bits in 262 length, and the starting value can be selected arbitrarily. The 263 length of the sequence number field was chosen to avoid exhaustion 264 during the lifetime of a security association as stated in Section 265 3 of [1]. In order for the receiver to distinguish between a new 266 and a replayed message, the sequence number must be monotonically 267 incremented modulo 2^64 for each message. We assume that the 268 first sequence number seen (i.e., the starting sequence number) is 269 stored somewhere. The modulo-operation is required because the 270 starting sequence number may be an arbitrary number. The receiver 271 therefore only accepts packets with a sequence number larger 272 (modulo 2^64) than the previous packet. As explained in [1] this 273 process is started by handshaking and agreeing on an initial 274 sequence number. If no such handshaking is available then the 275 initial sequence number must be part of the establishment of the 276 security association. 278 The generation and storage of sequence numbers is an important 279 step in preventing replay attacks and is largely determined by the 280 capabilities of the system in presence of system crashes, failures 281 and restarts. Section 3 of [1] explains some of the most 282 important considerations. However, the description of how the 283 receiver distinguishes proper from improper sequence numbers is 284 incomplete--it implicitly assumes that gaps large enough to cause 285 the sequence number to wrap around cannot occur. 287 If delivery in order were guaranteed, the following procedure 288 would work: The receiver keeps track of the first sequence number 289 received, INIT-SEQ, and most recent sequence number received, 290 LAST-SEQ, for each key identifier in a security association. When 291 the first message is received, set INIT-SEQ = LAST-SEQ = value 292 received and accept. When a subsequent message is received, if 293 its sequence number is strictly between LAST-SEQ and INIT-SEQ, 294 modulo 2^64, accept and update LAST-SEQ with the value just 295 received. If it is between INIT-SEQ and LAST-SEQ, inclusive, 296 modulo 2^64, reject and leave the value of LAST-SEQ unchanged. 297 Because delivery in order is not guaranteed, the above rules need 298 to be combined with a method of allowing a fixed sized window in 299 the neighborhood of LAST-SEQ for out-of-order delivery, for 300 example, as described in Appendix C of [3]. 302 Keyed Message Digest: 304 The Keyed Message Digest is a security mechanism built into RSVP 305 and used to provide integrity protection of a signaling message 306 (including its sequence number). Prior to computing the value for 307 the Keyed Message Digest field, the Keyed Message Digest field 308 itself must be set to zero and a keyed hash computed over the 309 entire RSVP packet. The Keyed Message Digest field is variable in 310 length but must be a multiple of four octets. If HMAC-MD5 is 311 used, then the output value is 16 bytes long. The keyed hash 312 function HMAC-MD5 [4] is required for a RSVP implementation as 313 noted in Section 1 of [1]. Hash algorithms other than MD5 [5] 314 like SHA-1 [19] may also be supported. 316 The key used for computing this Keyed Message Digest may be 317 obtained from the pre-shared secret, which is either manually 318 distributed or the result of a key management protocol. No key 319 management protocol, however, is specified to create the desired 320 security associations. Also, no guidelines for key length are 321 given. It should be recommended that HMAC-MD5 keys be 128 bits 322 and SHA-1 key 160 bits, as in IPsec AH [20] and ESP [21]. 324 3.2 Security Associations 326 Different attributes are stored for security associations of sending 327 and receiving systems (i.e., unidirectional security associations). 328 The sending system needs to maintain the following attributes in such 329 a security association [1]: 331 o Authentication algorithm and algorithm mode 332 o Key 333 o Key Lifetime 334 o Sending Interface 335 o Latest sequence number (received with this key identifier) 337 The receiving system has to store the following fields: 339 o Authentication algorithm and algorithm mode 340 o Key 341 o Key Lifetime 342 o Source address of the sending system 343 o List of last n sequence numbers (received with this key 344 identifier) 346 Note that the security associations need to have additional fields to 347 indicate their state. It is necessary to have an overlapping 348 lifetime of security associations to avoid interrupting an ongoing 349 communication because of expired security associations. During such 350 a period of overlapping lifetime it is necessary to authenticate 351 either one or both active keys. As mentioned in [1], a sender and a 352 receiver may have multiple active keys simultaneously.If more than 353 one algorithm is supported then the algorithm used must be specified 354 for a security association. 356 3.3 RSVP Key Management Assumptions 358 [6] assumes that security associations are already available. An 359 implementation must support manual key distribution as noted in 360 Section 5.2 of [1]. Manual key distribution, however, has different 361 requirements for key storage - a simple plaintext ASCII file may be 362 sufficient in some cases. If multiple security associations with 363 different lifetimes need to be supported at the same time, then a key 364 engine would be more appropriate. Further security requirements 365 listed in Section 5.2 of [1] are the following: 367 o The manual deletion of security associations must be supported. 368 o The key storage should persist a system restart. 369 o Each key must be assigned a specific lifetime and a specific Key 370 Identifier. 372 3.4 Identity Representation 374 In addition to host-based authentication with the INTEGRITY object 375 inside the RSVP message, user-based authentication is available as 376 introduced in [2]. Section 2 of [7] states that "Providing policy 377 based admission control mechanism based on user identities or 378 application is one of the prime requirements." To identify the user 379 or the application, a policy element called AUTH_DATA, which is 380 contained in the POLICY_DATA object, is created by the RSVP daemon at 381 the user's host and transmitted inside the RSVP message. The 382 structure of the POLICY_DATA element is described in [2]. Network 383 nodes like the policy decision point (PDP) then use the information 384 contained in the AUTH_DATA element to authenticate the user and to 385 allow policy-based admission control to be executed. As mentioned in 386 [7], the policy element is processed and the PDP replaces the old 387 element with a new one for forwarding to the next hop router. 389 A detailed description of the POLICY_DATA element can be found in 390 [2]. The attributes contained in the authentication data policy 391 element AUTH_DATA, which is defined in [7], are briefly explained in 392 this Section. Figure 1 shows the abstract structure of the RSVP 393 message with its security-relevant objects and the scope of 394 protection. The RSVP INTEGRITY object (outer object) covers the 395 entire RSVP message, whereas the POLICY_DATA INTEGRITY object only 396 covers objects within the POLICY_DATA element. 398 +--------------------------------------------------------+ 399 | RSVP Message | 400 +--------------------------------------------------------+ 401 | Object |POLICY_DATA Object || 402 | +-------------------------------------------+| 403 | | INTEGRITY +------------------------------+|| 404 | | Object | AUTH_DATA Object ||| 405 | | +------------------------------+|| 406 | | | Various Authentication ||| 407 | | | Attributes ||| 408 | | +------------------------------+|| 409 | +-------------------------------------------+| 410 +--------------------------------------------------------+ 412 Figure 1: Security Relevant Objects and Elements within the RSVP 413 Message 415 The AUTH_DATA object contains information for identifying users and 416 applications together with credentials for those identities. The 417 main purpose of these identities seems to be usage for policy-based 418 admission control and not authentication and key management. As 419 noted in Section 6.1 of [7], an RSVP message may contain more than 420 one POLICY_DATA object and each of them may contain more than one 421 AUTH_DATA object. As indicated in Figure 1 and in [7], one AUTH_DATA 422 object may contain more than one authentication attribute. A typical 423 configuration for Kerberos-based user authentication includes at 424 least the Policy Locator and an attribute containing the Kerberos 425 session ticket. 427 Successful user authentication is the basis for executing 428 policy-based admission control. Additionally, other information such 429 as time-of-day , application type, location information, group 430 membership, etc. may be relevant to implement an access control 431 policy. 433 The following attributes are defined for the usage in the AUTH_DATA 434 object: 436 1. Policy Locator 437 * ASCII_DN 438 * UNICODE_DN 439 * ASCII_DN_ENCRYPT 440 * UNICODE_DN_ENCRYPT 441 The policy locator string that is an X.500 distinguished name 442 (DN) used to locate user or application specific policy 443 information. The following types of X.500 DNs are listed: 444 The first two types are the ASCII and the Unicode representation 445 of the user or application DN identity. The two "encrypted" 446 distinguished name types are either encrypted with the Kerberos 447 session key or with the private key of the user's digital 448 certificate (i.e., digitally signed). The term encrypted 449 together with a digital signature is easy to misconceive. If 450 user identity confidentiality is provided, then the policy 451 locator has to be encrypted with the public key of the recipient. 452 How to obtain this public key is not described in the document. 453 Such an issue may be specified in a concrete architecture where 454 RSVP is used. 455 2. Credentials 456 Two cryptographic credentials are currently defined for a user: 457 Authentication with Kerberos V5 [8], and authentication with the 458 help of digital signatures based on X.509 [22] and PGP [23]. The 459 following list contains all defined credential types currently 460 available and defined in [7]: 462 +--------------+--------------------------------+ 463 | Credential | Description | 464 | Type | | 465 +===============================================| 466 | ASCII_ID | User or application identity | 467 | | encoded as an ASCII string | 468 +--------------+--------------------------------+ 469 | UNICODE_ID | User or application identity | 470 | | encoded as a Unicode string | 471 +--------------+--------------------------------+ 472 | KERBEROS_TKT | Kerberos V5 session ticket | 473 +--------------+--------------------------------+ 474 | X509_V3_CERT | X.509 V3 certificate | 475 +--------------+--------------------------------+ 476 | PGP_CERT | PGP certificate | 477 +--------------+--------------------------------+ 479 Figure 2: Credentials Supported in RSVP 481 The first two credentials contain only a plaintext string, and 482 therefore they do not provide cryptographic user authentication. 483 These plaintext strings may be used to identify applications, 484 which are included for policy-based admission control. Note that 485 these plain-text identifiers may, however, be protected if either 486 the RSVP INTEGRITY or the INTEGRITY object of the POLICY_DATA 487 element is present. Note that the two INTEGRITY objects can 488 terminate at different entities depending on the network 489 structure. The digital signature may also provide protection of 490 application identifiers. A protected application identity (and 491 the entire content of the POLICY_DATA element) cannot be modified 492 as long as no policy ignorant nodes are encountered in between. 493 A Kerberos session ticket, as previously mentioned, is the ticket 494 of a Kerberos AP_REQ message [8] without the Authenticator. 495 Normally, the AP_REQ message is used by a client to authenticate 496 to a server. The INTEGRITY object (e.g., of the POLICY_DATA 497 element) provides the functionality of the Kerberos 498 Authenticator, namely protecting against replay and showing that 499 the user was able to retrieve the session key following the 500 Kerberos protocol. This is, however, only the case if the 501 Kerberos session was used for the keyed message digest field of 502 the INTEGRITY object. Section 7 of [1] discusses some issues for 503 establishment of keys for the INTEGRITY object. The 504 establishment of the security association for the RSVP INTEGRITY 505 object with the inclusion of the Kerberos Ticket within the 506 AUTH_DATA element may be complicated by the fact that the ticket 507 can be decrypted by node B whereas the RSVP INTEGRITY object 508 terminates at a different host C. The Kerberos session ticket 509 contains, among many other fields, the session key. The Policy 510 Locator may also be encrypted with the same session key. The 511 protocol steps that need to be executed to obtain such a Kerberos 512 service ticket are not described in [7] and may involve several 513 roundtrips depending on many Kerberos-related factors. The 514 Kerberos ticket does not need to be included in every RSVP 515 message as an optimization, as described in Section 7.1 of [1]. 516 Thus the receiver must store the received service ticket. If the 517 lifetime of the ticket has expired, then a new service ticket 518 must be sent. If the receiver lost its state information 519 (because of a crash or restart) then it may transmit an Integrity 520 Challenge message to force the sender to re-transmit a new 521 service ticket. 522 If either the X.509 V3 or the PGP certificate is included in the 523 policy element, then a digital signature must be added. The 524 digital signature computed over the entire AUTH_DATA object 525 provides authentication and integrity protection. The SubType of 526 the digital signature authentication attribute is set to zero 527 before computing the digital signature. Whether or not a 528 guarantee of freshness with replay protection (either timestamps 529 or sequence numbers) is provided by the digital signature is an 530 open issue as discussed in Section 4.3 531 3. Digital Signature 532 The digital signature computed over the data of the AUTH_DATA 533 object must be the last attribute. The algorithm used to compute 534 the digital signature depends on the authentication mode listed 535 in the credential. This is only partially true, because, for 536 example, PGP again allows different algorithms to be used for 537 computing a digital signature. The algorithm identifier used for 538 computing the digital signature is not included in the 539 certificate itself. The algorithm identifier included in the 540 certificate only serves the purpose of allowing the verification 541 of the signature computed by the certificate authority (except 542 for the case of self-signed certificates). 543 4. Policy Error Object 544 The Policy Error Object is used in the case of a failure of 545 policy-based admission control or other credential verification. 546 Currently available error messages allow notification if the 547 credentials are expired (EXPIRED_CREDENTIALS), if the 548 authorization process disallowed the resource request 549 (INSUFFICIENT_PRIVILEGES), or if the given set of credentials is 550 not supported (UNSUPPORTED_CREDENTIAL_TYPE). The last error 551 message returned by the network allows the user's host to 552 discover the type of credentials supported. Particularly for 553 mobile environments this might be quite inefficient. 554 Furthermore, it is unlikely that a user supports different types 555 of credentials. The purpose of the error message 556 IDENTITY_CHANGED is unclear. Also, the protection of the error 557 message is not discussed in [7]. 559 3.5 RSVP Integrity Handshake 561 The Integrity Handshake protocol was designed to allow a crashed or 562 restarted host to obtain the latest valid challenge value stored at 563 the receiving host. Due to the absence of key management, it must be 564 guaranteed that two messages do not use the same sequence number with 565 the same key. A host stores the latest sequence number of a 566 cryptographically verified message. An adversary can replay 567 eavesdropped packets if the crashed host has lost its sequence 568 numbers. A signaling message from the real sender with a new 569 sequence number would therefore allow the crashed host to update the 570 sequence number field and prevent further replays. Hence, if there 571 is a steady flow of RSVP protected messages between the two hosts, an 572 attacker may find it difficult to inject old messages, because new, 573 authenticated messages with higher sequence numbers arrive and get 574 stored immediately. 576 The following description explains the details of a RSVP Integrity 577 Handshake that is started by Node A after recovering from a 578 synchronization failure: 580 Integrity Challenge 582 (1) Message (including 583 +----------+ a Cookie) +----------+ 584 | |-------------------------->| | 585 | Node A | | Node B | 586 | |<--------------------------| | 587 +----------+ Integrity Response +----------+ 588 (2) Message (including 589 the Cookie and the 590 INTEGRITY object) 592 Figure 3: RSVP Integrity Handshake 594 The details of the messages are as follows: 596 CHALLENGE:=(Key Identifier, Challenge Cookie) 597 Integrity Challenge Message:=(Common Header, CHALLENGE) 598 Integrity Response Message:=(Common Header, INTEGRITY, CHALLENGE) 600 The "Challenge Cookie" is suggested to be a MD5 hash of a local 601 secret and a timestamp [1]. 603 The Integrity Challenge message is not protected with an INTEGRITY 604 object as shown in the protocol flow above. As explained in Section 605 10 of [1] this was done to avoid problems in situations where both 606 communicating parties do not have a valid starting sequence number. 608 Using the RSVP Integrity Handshake protocol is recommended although 609 it is not mandatory (since it may not be needed in all network 610 environments). 612 4. Detailed Security Property Discussion 614 The purpose of this section is to describe the protection of the 615 RSVP-provided mechanisms individually for authentication, 616 authorization, integrity and replay protection, user identity 617 confidentiality, and confidentiality of the signaling messages. 619 4.1 Network Topology 621 The main purpose of this paragraph is to show the basic interfaces in 622 a simple RSVP network architecture. The architecture below assumes 623 that there is only a single domain and that two routers are RSVP and 624 policy aware. These assumptions are relaxed in the individual 625 paragraphs as necessary. Layer 2 devices between the clients and 626 their corresponding first hop routers are not shown. Other network 627 elements like a Kerberos Key Distribution Center and for example a 628 LDAP server, from which the PDP retrieves its policies are also 629 omitted. The security of various interfaces to the individual 630 servers (KDC, PDP, etc.) depends very much on the security policy of 631 a specific network service provider. 633 +--------+ 634 | Policy | 635 +----|Decision| 636 | | Point +---+ 637 | +--------+ | 638 | | 639 | | 640 +------+ +-+----+ +---+--+ +------+ 641 |Client| |Router| |Router| |Client| 642 | A +-------+ 1 +--------+ 2 +----------+ B | 643 +------+ +------+ +------+ +------+ 645 Figure 4: Simple RSVP Architecture 647 4.2 Host/Router 649 When considering authentication in RSVP it is important to make a 650 distinction between user and host authentication of the signaling 651 messages . By using the RSVP INTEGRITY object the host is 652 authenticated while credentials inside the AUTH_DATA object can be 653 used to authenticate the user. In this section the focus is on host 654 authentication whereas the next section covers user authentication. 656 1. Authentication 657 The term host authentication is used above, because the selection 658 of the security association is bound to the host's IP address as 659 mentioned in Section 3.1. and Section 3.2. Depending on the key 660 management protocol used to create this security association and 661 the identity used, it is also possible to bind a user identity to 662 this security association. Because the key management protocol 663 is not specified, it is difficult to evaluate this part and hence 664 we speak about data origin authentication based on the host's 665 identity for RSVP INTEGRITY objects. The fact that the host 666 identity is used for selecting the security association has 667 already been described in Section 3.1. 668 Data origin authentication is provided with the keyed hash value 669 computed over the entire RSVP message excluding the keyed message 670 digest field itself. The security association used between the 671 user's host and the first-hop router is, as previously mentioned, 672 not established by RSVP and must therefore be available before 673 signaling is started. 674 * Kerberos for the RSVP INTEGRITY object 675 As described in Section 7 of [1], Kerberos may be used to create 676 the key for the RSVP INTEGRITY object. How to learn the 677 principal name (and realm information) of the other node is 678 outside the scope of [1]. [24] describes a way to distribute 679 principal and realm information via DNS, which can be used for 680 this purpose (assuming that the FQDN or the IP address of the 681 other node for which this information is desired is known). All 682 that is required is to encapsulate the Kerberos ticket inside the 683 policy element. It is furthermore mentioned that Kerberos 684 tickets with expired lifetime must not be used and the initiator 685 is responsible for requesting and exchanging a new service ticket 686 before expiration. 687 RSVP multicast processing in combination with Kerberos requires 688 additional considerations: 689 Section 7 of [1] states that in the multicast case all receivers 690 must share a single key with the Kerberos Authentication Server, 691 i.e., a single principal used for all receivers). From a 692 personal discussion with Rodney Hess it seems that there is 693 currently no other solution available in the context of Kerberos. 694 Multicast handling therefore leaves some open questions in this 695 context. 696 In the case where one entity crashed, the established security 697 association is lost and therefore the other node must retransmit 698 the service ticket . The crashed entity can use an Integrity 699 Challenge message to request a new Kerberos ticket to be 700 retransmitted by the other node. If a node receives such a 701 request, then a reply message must be returned. 702 2. Integrity protection 703 Integrity protection between the user's host and the first hop 704 router is based on the RSVP INTEGRITY object. HMAC-MD5 is 705 preferred, although other keyed hash functions may also be used 706 within the RSVP INTEGRITY object. In any case, both 707 communicating entities must have a security association that 708 indicates the algorithm to use. This may, however, be difficult, 709 because no negotiation protocol is defined to agree on a specific 710 algorithm. Hence, if RSVP is used in a mobile environment, it is 711 likely that HMAC-MD5 is the only usable algorithm for the RSVP 712 INTEGRITY object. Only in local environments may it be useful to 713 switch to a different keyed hash algorithm. The other possible 714 alternative is that every implementation must support the most 715 important keyed hash algorithms for example MD5, SHA-1, 716 RIPEMD-160, etc. HMAC-MD5 was mainly chosen because of its 717 performance characteristics. The weaknesses of MD5 [25] are 718 known and described in [26]. Other algorithms like SHA-1 [19] 719 and RIPEMD-160 [25] have stronger security properties. 720 3. Replay Protection 721 The main mechanism used for replay protection in RSVP is based on 722 sequence numbers, whereby the sequence number is included in the 723 RSVP INTEGRITY object. The properties of this sequence number 724 mechanism are described in Section 3.1. The fact that the 725 receiver stores a list of sequence numbers is an indicator for a 726 window mechanism. This somehow conflicts with the requirement 727 that the receiver only has to store the highest number given in 728 Section 3 of [1]. We assume that this is a typo. Section 4.2 of 729 [1] gives a few comments about the out-of-order delivery and the 730 ability of an implementation to specify the replay window. 731 Appendix C of [3] describes a window mechanism for handling 732 out-of-sequence delivery. 733 4. Integrity Handshake 734 The mechanism of the Integrity Handshake is explained in Section 735 Section 3.5. The Cookie value is suggested to be hash of a local 736 secret and a timestamp. The Cookie value is not verified by the 737 receiver. The mechanism used by the Integrity Handshake is a 738 simple Challenge/Response message, which assumes that the key 739 shared between the two hosts survives the crash. If, however, 740 the security association is dynamically created, then this 741 assumption may not be true. 742 In Section 10 of [1] the authors note that an adversary can 743 create a faked Integrity Handshake message including challenge 744 cookies. Subsequently it could store the received response and 745 later try to replay these responses while a responder recovers 746 from a crash or restart. If this replayed Integrity Response 747 value is valid and has a lower sequence number than actually 748 used, then this value is stored at the recovering host. In order 749 for this attack to be successful the adversary must either have 750 collected a large number of challenge/response value pairs or 751 have "discovered" the cookie generation mechanism (for example by 752 knowing the local secret). The collection of Challenge/Response 753 pairs is even more difficult, because they depend on the Cookie 754 value, the sequence number included in the response message, and 755 the shared key used by the INTEGRITY object. 756 5. Confidentiality 757 Confidentiality is not considered to be a security requirement 758 for RSVP. Hence it is not supported by RSVP, except as described 759 in paragraph d) of Section 4.3. This assumption may not hold, 760 however, for enterprises or carriers who want to protect, in 761 addition to users' identities, also billing data, network usage 762 patterns, or network configurations from eavesdropping and 763 traffic analysis. Confidentiality may also help make certain 764 other attacks more difficult. For example, the PathErr attack 765 described in Section 5.2 is harder to carry out if the attacker 766 cannot observe the Path message to which the PathErr corresponds. 767 6. Authorization 768 The task of authorization consists of two subcategories: network 769 access authorization and RSVP request authorization. Access 770 authorization is provided when a node is authenticated to the 771 network, e.g., using EAP [27] in combination with AAA protocols 772 (for example using RADIUS [28] or DIAMETER [9]). Issues related 773 to network access authentication and authorization are outside 774 the scope of RSVP. 775 The second authorization refers to RSVP itself. Depending on the 776 network configuration: 777 * the router either forwards the received RSVP request to the 778 policy decision point, e.g., by using COPS [10] and [11],to 779 request that an admission control procedure be executed or 780 * the router supports the functionality of a PDP and therefore 781 there is no need to forward the request or 782 * the router may already be configured with the appropriate 783 policy information to decide locally whether to grant this 784 request or not 785 Based on the result of the admission control, the request may be 786 granted or rejected. Information about the resource-requesting 787 entity must be available to provide policy-based admission 788 control. 789 7. Performance 790 The computation of the keyed message digest for a RSVP INTEGRITY 791 object does not represent a performance problem. The protection 792 of signaling messages is usually not a problem, because these 793 messages are transmitted at a low rate. Even a high volume of 794 messages does not cause performance problems for a RSVP routers 795 due to the efficiency of the keyed message digest routine. 796 Dynamic key management, which is computationally more demanding, 797 is more important for scalability. Because RSVP does not specify 798 a particular key exchange protocol, it is difficult to estimate 799 the effort to create the required security associations. 800 Furthermore, the number of key exchanges to be triggered depends 801 on security policy issues like lifetime of a security 802 association, required security properties of the key exchange 803 protocol, authentication mode used by the key exchange protocol, 804 etc. In a stationary environment with a single administrative 805 domain, manual security association establishment may be 806 acceptable and may provide the best performance characteristics. 807 In a mobile environment, asymmetric authentication methods are 808 likely to be used with a key exchange protocol, and some sort of 809 public key or certificate verification needs to be supported. 811 4.3 User to PEP/PDP 813 As noted in the previous section, both user-based and host-based 814 authentication are supported by RSVP. Using RSVP, a user may 815 authenticate to the first hop router or to the PDP as specified in 816 [1], depending on the infrastructure provided by the network domain 817 or the architecture used (e.g., the integration of RSVP and Kerberos 818 V5 into the Windows 2000 Operating System [29]. Another architecture 819 in which RSVP is tightly integrated is the one specified by the 820 PacketCable organization. The interested reader is referred to [30] 821 for a discussion of their security architecture. 823 1. Authentication 824 When a user sends a RSVP PATH or RESV message, this message may 825 include some information to authenticate the user. [7] describes 826 how user and application information is embedded into the RSVP 827 message (AUTH_DATA object) and how to protect it. A router 828 receiving such a message can use this information to authenticate 829 the client and forward the user or application information to the 830 policy decision point (PDP). Optionally the PDP itself can 831 authenticate the user, which is described in the next section. 832 To be able to authenticate the user, to verify the integrity, and 833 to check for replays, the entire POLICY_DATA element has to be 834 forwarded from the router to the PDP, e.g., by including the 835 element into a COPS message. It is assumed, although not clearly 836 specified in [7], that the INTEGRITY object within the 837 POLICY_DATA element is sent to the PDP along with all other 838 attributes. 839 * Certificate Verification 840 Using the policy element as described in [7] it is not possible 841 to provide a certificate revocation list or other information to 842 prove the validity of the certificate inside the policy element. 843 A specific mechanism for certificate verification is not 844 discussed in [7] and hence a number of them can be used for this 845 purpose. For certificate verification, the network element (a 846 router or the policy decision point), which has to authenticate 847 the user, could frequently download certificate revocation lists 848 or use a protocol like the Online Certificate Status Protocol 849 (OCSP) [31] and the Simple Certificate Validation Protocol (SCVP) 851 [32] to determine the current status of a digital certificate. 852 * User Authentication to the PDP 853 This alternative authentication procedure uses the PDP to 854 authenticate the user instead of the first hop router. In 855 Section 4.2.1 of [7] the choice is given for the user to obtain a 856 session ticket either for the next hop router or for the PDP. As 857 noted in the same Section, the identity of the PDP or the next 858 hop router is statically configured or dynamically retrieved. 859 Subsequently, user authentication to the PDP is considered. 860 * Kerberos-based Authentication to the PDP 861 If Kerberos is used to authenticate the user, then a session 862 ticket for the PDP needs to be requested first. A user who roams 863 between different routers in the same administrative domain does 864 not need to request a new service ticket, because the PDP is 865 likely to be used by most or all first-hop routers within the 866 same administrative domain. This is different from the case in 867 which a session ticket for a router has to be obtained and 868 authentication to a router is required. The router therefore 869 plays a passive role of forwarding the request only to the PDP 870 and executing the policy decision returned by the PDP. 871 Appendix B describes one example of user-to-PDP authentication. 872 User authentication with the policy element only provides 873 unilateral authentication whereby the client authenticates to the 874 router or to the PDP. If a RSVP message is sent to the user's 875 host and public key based authentication is used, then the 876 message does not contain a certificate and digital signature. 877 Hence no mutual authentication can be assumed. In case of 878 Kerberos, mutual authentication may be accomplished if the PDP or 879 the router transmits a policy element with an INTEGRITY object 880 computed with the session key retrieved from the Kerberos ticket 881 or if the Kerberos ticket included in the policy element is also 882 used for the RSVP INTEGRITY object as described in Section 4.2. 883 This procedure only works if a previous message was transmitted 884 from the end host to the network and such key is already 885 established. [7] does not discuss this issue and therefore there 886 is no particular requirement dealing with transmitting 887 network-specific credentials back to the end-user's host. 888 2. Integrity Protection 889 Integrity protection is applied separately to the RSVP message 890 and the POLICY_DATA element as shown in Figure 1. In case of a 891 policy-ignorant node along the path, the RSVP INTEGRITY object 892 and the INTEGRITY object inside the policy element terminate at 893 different nodes. Basically, the same is true for the user 894 credentials if they are verified at the policy decision point 895 instead of the first hop router. 896 * Kerberos 897 If Kerberos is used to authenticate the user to the first hop 898 router, then the session key included in the Kerberos ticket may 899 be used to compute the INTEGRITY object of the policy element. 900 It is the keyed message digest that provides the authentication. 901 The existence of the Kerberos service ticket inside the AUTH_DATA 902 object does not provide authentication and a guarantee of 903 freshness for the receiving host. Authentication and guarantee 904 of freshness are provided by the keyed hash value of the 905 INTEGRITY object inside the POLICY_DATA element. This shows that 906 the user actively participated in the Kerberos protocol and was 907 able to obtain the session key to compute the keyed message 908 digest. The Authenticator used in the Kerberos V5 protocol 909 provides similar functionality, but replay protection is based on 910 timestamps (or on a sequence number if the optional seq-number 911 field inside the Authenticator is used for KRB_PRIV/KRB_SAFE 912 messages as described in Section 5.3.2 of [8]). 913 * Digital Signature 914 If public key based authentication is provided, then user 915 authentication is accomplished with a digital signature. As 916 explained in Section 3.3.3 of [7], the DIGITAL_SIGNATURE 917 attribute must be the last attribute in the AUTH_DATA object, and 918 the digital signature covers the entire AUTH_DATA object. Which 919 hash algorithm and public key algorithm are used for the digital 920 signature computation is described in [23] in the case of PGP. 921 In the case of X.509 credentials the situation is more complex, 922 because different mechanisms like CMS [33] or PKCS#7 [34] may be 923 used for digitally signing the message element. X.509 only 924 provides the standard for the certificate layout, which seems to 925 provide insufficient information for this purpose. Therefore, 926 X.509 certificates are supported for example by CMS and PKCS#7. 927 [7], however, does not make any statements about the usage of CMS 928 and PKCS#7. Currently there is no support for CMS or PKCS#7 929 described in [7], which provides more than just public key based 930 authentication (e.g., CRL distribution, key transport, key 931 agreement, etc.). Furthermore, the use of PGP in RSVP is vaguely 932 defined, because there are different versions of PGP (including 933 OpenPGP [23]), and no indication is given as to which should be 934 used. 935 Supporting public key based mechanisms in RSVP might increase the 936 risks of denial of service attacks. Additionally, the large 937 processing, memory, and bandwidth utilization should be 938 considered. Fragmentation might also be an issue here. 939 If the INTEGRITY object is not included in the POLICY_DATA 940 element or not sent to the PDP, then we have to make the 941 following observations: 942 3. For the digital signature case, only the replay protection 943 provided by the digital signature algorithm can be used. It 944 is not clear, however, whether this usage was anticipated or 945 not. Hence, we might assume that replay protection is based 946 on the availability of the RSVP INTEGRITY object used with a 947 security association that is established by other means. 948 4. Including only the Kerberos session ticket is insufficient, 949 because freshness is not provided (since the Kerberos 950 Authenticator is missing). Obviously there is no guarantee 951 that the user actually followed the Kerberos protocol and was 952 able to decrypt the received TGS_REP (or in rare cases the 953 AS_REP if a session ticket is requested with the initial 954 AS_REQ). 955 5. Replay Protection 956 Figure 5 shows the interfaces relevant for replay protection 957 of signaling messages in a more complicated architecture. In 958 this case, the client uses the policy data element with PEP2, 959 because PEP1 is not policy aware. The interfaces between the 960 client and PEP1 and between PEP1 and PEP2 are protected with 961 the RSVP INTEGRITY object. The link between the PEP2 and the 962 PDP is protected, for example, by using the COPS built-in 963 INTEGRITY object. The dotted line between the Client and the 964 PDP indicates the protection provided by the AUTH_DATA 965 element, which has no RSVP INTEGRITY object included. 967 AUTH_DATA +----+ 968 +---------------------------------------------------+PDP +-+ 969 | +----+ | 970 | | 971 | | 972 | COPS | 973 | INTEGRITY| 974 | | 975 | | 976 | | 977 +--+---+ RSVP INTEGRITY +----+ RSVP INTEGRITY +----+ | 978 |Client+-------------------+PEP1+----------------------+PEP2+-+ 979 +--+---+ +----+ +-+--+ 980 | | 981 +-----------------------------------------------------+ 982 POLICY_DATA INTEGRITY 984 Figure 5: Replay Protection 986 Host authentication with the RSVP INTEGRITY object and user 987 authentication with the INTEGRITY object inside the 988 POLICY_DATA element both use the same anti-replay mechanism. 989 The length of the Sequence Number field, sequence number 990 rollover, and the Integrity Handshake have already been 991 explained in Section 3.1. 992 Section 9 of [7] states: "RSVP INTEGRITY object is used to 993 protect the policy object containing user identity 994 information from security (replay) attacks." When using 995 public key based authentication, RSVP based replay protection 996 is not supported, because the digital signature does not 997 cover the POLICY_DATA INTEGRITY object with its Sequence 998 Number field. The digital signature covers only the entire 999 AUTH_DATA object. 1000 The use of public key cryptography within the AUTH_DATA 1001 object complicates replay protection. Digital signature 1002 computation with PGP is described in [35] and in [23]. The 1003 data structure preceding the signed message digest includes 1004 information about the message digest algorithm used and a 1005 32-bit timestamp of when the signature was created 1006 ("Signature creation time"). The timestamp is included in 1007 the computation of the message digest. The IETF standardized 1008 OpenPGP version [23] contains more information and describes 1009 the different hash algorithms (MD2, MD5, SHA-1, RIPEMD-160) 1010 supported. [7] does not make any statements as to whether 1011 the "Signature creation time" field is used for replay 1012 protection. Using timestamps for replay protection requires 1013 different synchronization mechanisms in the case of 1014 clock-skew. Traditionally, these cases assume "loosely 1015 synchronized" clocks but also require specifying a 1016 replay-window. 1017 If the "Signature creation time" is not used for replay 1018 protection, then a malicious, policy-ignorant node can use 1019 this weakness to replace the AUTH_DATA object without 1020 destroying the digital signature. If this was not simply an 1021 oversight, it is therefore assumed that replay protection of 1022 the user credentials was not considered an important security 1023 requirement, because the hop-by-hop processing of the RSVP 1024 message protects the message against modification by an 1025 adversary between two communicating nodes. 1026 The lifetime of the Kerberos ticket is based on the fields 1027 starttime and endtime of the EncTicketPart structure in the 1028 ticket, as described in Section 5.3.1 of [8]. Because the 1029 ticket is created by the KDC located at the network of the 1030 verifying entity, it is not difficult to have the clocks 1031 roughly synchronized for the purpose of lifetime 1032 verification. Additional information about 1033 clock-synchronization and Kerberos can be found in [36]. 1034 If the lifetime of the Kerberos ticket expires, then a new 1035 ticket must be requested and used. Rekeying is implemented 1036 with this procedure. 1037 3. (User Identity) Confidentiality 1038 This section discusses privacy protection of identity information 1039 transmitted inside the policy element. User identity 1040 confidentiality is of particular interest because there is no 1041 built-in RSVP mechanism for encrypting the POLICY_DATA object or 1042 the AUTH_DATA elements. Encryption of one of the attributes 1043 inside the AUTH_DATA element, the POLICY_LOCATOR attribute, is 1044 discussed. 1045 To protect the user's privacy it is important not to reveal the 1046 user's identity to an adversary located between the user's host 1047 and the first-hop router (e.g., on a wireless link). User 1048 identities should furthermore not be transmitted outside the 1049 domain of the visited network provider, i.e., the user identity 1050 information inside the policy data element should be removed or 1051 modified by the PDP to prevent revealing its contents to other 1052 (non-authorized) entities along the signaling path. It is not 1053 possible (with the offered mechanisms) to hide the user's 1054 identity in such a way that it is not visible to the first 1055 policy-aware RSVP node (or to the attached network in general). 1056 The ASCII or Unicode distinguished name of user or application 1057 inside the POLICY_LOCATOR attribute of the AUTH_DATA element may 1058 be encrypted as specified in Section 3.3.1 of [7]. The user (or 1059 application) identity is then encrypted with either the Kerberos 1060 session key or with the private key in case of public key based 1061 authentication. When the private key is used, we usually speak 1062 of a digital signature that can be verified by everyone 1063 possessing the public key. Because the certificate with the 1064 public key is included in the message itself, decryption is no 1065 obstacle. Furthermore, the included certificate together with 1066 the additional (unencrypted) information in the RSVP message 1067 provides enough identity information for an eavesdropper. Hence, 1068 the possibility of encrypting the policy locator in case of 1069 public key based authentication is problematic. To encrypt the 1070 identities using asymmetric cryptography, the user's host must be 1071 able somehow to retrieve the public key of the entity verifying 1072 the policy element (i.e., the first policy aware router or the 1073 PDP). Then, this public key could be used to encrypt a symmetric 1074 key, which in turn encrypts the user's identity and certificate, 1075 as is done, e.g., by PGP. Currently no such mechanism is defined 1076 in [7]. 1077 The algorithm used to encrypt the POLICY_LOCATOR with the 1078 Kerberos session key is assumed to be the same as the one used 1079 for encrypting the service ticket. The information about the 1080 algorithm used is available in the etype field of the 1081 EncryptedData ASN.1 encoded message part. Section 6.3 of [8] 1082 lists the supported algorithms. [12] defines new encryption 1083 algorithms (Rijndael, Serpent, and Twofish). 1084 Evaluating user identity confidentiality requires also looking at 1085 protocols executed outside of RSVP (for example, the Kerberos 1086 protocol). The ticket included in the CREDENTIAL attribute may 1087 provide user identity protection by not including the optional 1088 cname attribute inside the unencrypted part of the Ticket. 1089 Because the Authenticator is not transmitted with the RSVP 1090 message, the cname and the crealm of the unencrypted part of the 1091 Authenticator are not revealed. In order for the user to request 1092 the Kerberos session ticket for inclusion in the CREDENTIAL 1093 attribute, the Kerberos protocol exchange must be executed. Then 1094 the Authenticator sent with the TGS_REQ reveals the identity of 1095 the user. The AS_REQ must also include the user's identity to 1096 allow the Kerberos Authentication Server to respond with an 1097 AS_REP message that is encrypted with the user's secret key. 1098 Using Kerberos, it is therefore only possible to hide the content 1099 of the encrypted policy locator, which is only useful if this 1100 value differs from the Kerberos principal name. Hence using 1101 Kerberos it is not "entirely" possible to provide user identity 1102 confidentiality. 1103 It is important to note that information stored in the policy 1104 element may be changed by a policy-aware router or by the policy 1105 decision point. Which parts are changed depends upon whether 1106 multicast or unicast is used, how the policy server reacts, where 1107 the user is authenticated, whether the user needs to be 1108 re-authenticated in other network nodes, etc. Hence, user and 1109 application specific information can leak after the messages 1110 leave the first hop within the network where the user's host is 1111 attached. As mentioned at the beginning of this section, this 1112 information leakage is assumed to be intentional. 1113 4. Authorization 1114 In addition to the description of the authorization steps of the 1115 Host-to-Router interface, user-based authorization is performed 1116 with the policy element providing user credentials. The 1117 inclusion of user and application specific information enables 1118 policy-based admission control with special user policies that 1119 are likely to be stored at a dedicated server. Hence a Policy 1120 Decision Point can query, for example, a LDAP server for a 1121 service level agreement stating the amount of resources a certain 1122 user is allowed to request. In addition to the user identity 1123 information, group membership and other non-security-related 1124 information may contribute to the evaluation of the final policy 1125 decision . If the user is not registered to the currently 1126 attached domain, then there is the question of how much 1127 information the home domain of the user is willing to exchange. 1128 This also impacts the user's privacy policy. In general, the 1129 user may not want to distribute much of this policy information. 1130 Furthermore, the lack of a standardized authorization data format 1131 may create interoperability problems when exchanging policy 1132 information. Hence, we can assume that the policy decision point 1133 may use information from an initial authentication and key 1134 agreement protocol, which may have already required cross-realm 1135 communication with the user's home domain if only to assume that 1136 the home domain knows the user and that the user is entitled to 1137 roam and to be able to forward accounting messages to this 1138 domain. This represents the traditional subscriber-based 1139 accounting scenario. Non-traditional or alternative means of 1140 access might be deployed in the near future that do not require 1141 any type of inter-domain communication. 1142 Additional discussions are required to determine the expected 1143 authorization procedures. [37] and [38] discuss authorization 1144 issues for QoS signaling protocols. Furthermore, a number of 1145 mobililty implications for policy handling in RSVP are described 1146 in [39] 1147 5. Performance 1148 If Kerberos is used for user authentication, then a Kerberos 1149 ticket must be included in the CREDENTIAL Section of the 1150 AUTH_DATA element. The Kerberos ticket has a size larger than 1151 500 bytes but only needs to be sent once, because a performance 1152 optimization allows the session key to be cached as noted in 1153 Section 7.1 of [1]. It is assumed that subsequent RSVP messages 1154 only include the POLICY_DATA INTEGRITY object with a keyed 1155 message digest that uses the Kerberos session key. This, 1156 however, assumes that the security association required for the 1157 POLICY_DATA INTEGRITY object is created (or modified) to allow 1158 the selection of the correct key. Otherwise, it difficult to say 1159 which identifier is used to index the security association. 1160 When Kerberos is used as an authentication system then, from a 1161 performance perspective, the message exchange to obtain the 1162 session key needs to be considered, although the exchange only 1163 needs to be done once in the lifetime of the session ticket. 1164 This is particularly true in a mobile environment with a fast 1165 roaming user's host. 1166 Public key based authentication usually provides the best 1167 scalability characteristics for key distribution, but the 1168 protocols are performance demanding. A major disadvantage of the 1169 public key based user authentication in RSVP is the lack of a 1170 method to derive a session key. Hence every RSVP PATH or RESV 1171 message includes the certificate and a digital signature, which 1172 is a huge performance and bandwidth penalty. For a mobile 1173 environment with low power devices, high latency, channel noise, 1174 and low bandwidth links, this seems to be less encouraging. Note 1175 that a public key infrastructure is required to allow the PDP (or 1176 the first-hop router) to verify the digital signature and the 1177 certificate. To check for revoked certificates, certificate 1178 revocation lists or protocols like the Online Certificate Status 1179 Protocol [31] and the Simple Certificate Validation Protocol [32] 1180 are needed. Then the integrity of the AUTH_DATA object via the 1181 digital signature can be verified. 1183 4.4 Communication between RSVP-Aware Routers 1185 1. Authentication 1186 RSVP signaling messages are data origin authenticated and 1187 protected against modification and replay using the RSVP 1188 INTEGRITY object. The RSVP message flow between routers is 1189 protected based on the chain of trust and hence each router only 1190 needs to have a security association with its neighboring 1191 routers. This assumption was made because of performance 1192 advantages and because of special security characteristics of the 1193 core network where no user hosts are directly attached. In the 1194 core network the network structure does not change frequently and 1195 the manual distribution of shared secrets for the RSVP INTEGRITY 1196 object may be acceptable. The shared secrets may be either 1197 manually configured or distributed by using appropriately secured 1198 network management protocols like SNMPv3. 1199 Independent of the key distribution mechanism, host 1200 authentication with RSVP built-in mechanisms is accomplished with 1201 the keyed message digest in the RSVP INTEGRITY object computed 1202 using the previously exchanged symmetric key. 1203 2. Integrity Protection 1204 Integrity protection is accomplished with the RSVP INTEGRITY 1205 object with the variable length Keyed Message Digest field. 1206 3. Replay Protection 1207 Replay protection with the RSVP INTEGRITY object is extensively 1208 described in previous sections. To enable crashed hosts to learn 1209 the latest sequence number used, the Integrity Handshake 1210 mechanism is provided in RSVP. 1211 4. Confidentiality 1212 Confidentiality is not provided by RSVP. 1213 5. Authorization 1214 Depending on the RSVP network, QoS resource authorization at 1215 different routers may need to contact the PDP again. Because the 1216 PDP is allowed to modify the policy element, a token may be added 1217 to the policy element to increase the efficiency of the 1218 re-authorization procedure. This token is used to refer to an 1219 already computed policy decision. The communications interface 1220 from the PEP to the PDP must be properly secured. 1221 6. Performance 1222 The performance characteristics for the protection of the RSVP 1223 signaling messages is largely determined by the key exchange 1224 protocol, because the RSVP INTEGRITY object is only used to 1225 compute a keyed message digest of the transmitted signaling 1226 messages. 1227 The security associations within the core network, i.e., between 1228 individual routers (in comparison with the security association 1229 between the user's host and the first-hop router or with the 1230 attached network in general) can be established more easily 1231 because of the normally strong trust assumptions. Furthermore, 1232 it is possible to use security associations with an increased 1233 lifetime to avoid frequent rekeying. Hence, there is less impact 1234 on the performance compared with the user-to-network interface. 1236 The security association storage requirements are also less 1237 problematic. 1239 5. Miscellaneous Issues 1241 This section describes a number of issues that illustrate some of the 1242 shortcomings of RSVP with respect to security. 1244 5.1 First Hop Issue 1246 In case of end-to-end signaling, an end host starts signaling to its 1247 attached network. The first-hop communication is often more 1248 difficult to secure because of the different requirements and a 1249 missing trust relationship. An end host must therefore obtain some 1250 information to start RSVP signaling: 1252 o Does this network support RSVP signaling? 1253 o Which node supports RSVP signaling? 1254 o To which node is authentication required? 1255 o Which security mechanisms are used for authentication? 1256 o Which algorithms have to be used? 1257 o Where should the keys and security association come from? 1258 o Should a security association be established? 1260 RSVP, as specified today, is used as a building block. Hence, these 1261 questions have to be answered as part of overall architectural 1262 considerations. Without giving an answer to this question, ad hoc 1263 RSVP communication by an end host roaming to an unknown network is 1264 not possible. A negotiation of security mechanisms and algorithms is 1265 not supported for RSVP. 1267 5.2 Next-Hop Problem 1269 Throughout the document it was assumed that the next RSVP node along 1270 the path is always known. Knowing your next hop is important to be 1271 able to select the correct key for the RSVP Integrity object and to 1272 apply the proper protection. In case in which an RSVP node assumes 1273 it knows which node is the next hop the following protocol exchange 1274 can occur: 1276 Integrity 1277 (A<->C) +------+ 1278 (3) | RSVP | 1279 +------------->+ Node | 1280 | | B | 1281 Integrity | +--+---+ 1282 (A<->C) | | 1283 +------+ (2) +--+----+ | 1284 (1) | RSVP +----------->+Router | | Error 1285 ----->| Node | | or +<-----------+ (I am B) 1286 | A +<-----------+Network| (4) 1287 +------+ (5) +--+----+ 1288 Error . 1289 (I am B) . +------+ 1290 . | RSVP | 1291 ...............+ Node | 1292 | C | 1293 +------+ 1295 Figure 6: Next-Hop Issue 1297 When RSVP node A in Figure 6 receives an incoming RSVP Path message, 1298 standard RSVP message processing takes place. Node A then has to 1299 decide which key to select to protect the signaling message. We 1300 assume that some unspecified mechanism is used to make this decision. 1301 In this example node A assumes that the message will travel to RSVP 1302 node C. However, because of some reasons (e.g. a route change, 1303 inability to learn the next RSVP hop along the path, etc.) the 1304 message travels to node B via a non-RSVP supporting router that 1305 cannot verify the integrity of the message (or cannot decrypt the 1306 Kerberos service ticket). The processing failure causes a PathErr 1307 message to be returned to the originating sender of the Path message. 1308 This error message also contains information about the node 1309 recognizing the error. In many cases a security association might 1310 not be available. Node A receiving the PathErr message might use the 1311 information returned with the PathErr message to select a different 1312 security association (or to establish one). 1314 Figure 6 describes a behavior that might help node A learn that an 1315 error occurred. However, the description of Section 4.2 of [1] 1316 describes in step (5) that a signaling message is silently discarded 1317 if the receiving host cannot properly verify the message: "If the 1318 calculated digest does not match the received digest, the message is 1319 discarded without further processing." For RSVP Path and similar 1320 messages this functionality is not really helpful. 1322 The RSVP Path message therefore provides a number of functions: path 1323 discovery, detecting route changes, learning of QoS capabilities 1324 along the path using the Adspec object, (with some interpretation) 1325 next-hop discovery, and possibly security association establishment 1326 (for example, in the case of Kerberos). 1328 From a security point of view there is a conflict between 1330 o Idempotent message delivery and efficiency 1332 The RSVP Path message especially performs a number of functions. 1333 Supporting idempotent message delivery somehow contradicts with 1334 security association establishment, efficient message delivery, 1335 and message size. For example, a "real" idempotent signaling 1336 message would contain enough information to perform security 1337 processing without depending on a previously executed message 1338 exchange. Adding a Kerberos ticket with every signaling message 1339 is, however, inefficient. Using public key based mechanisms is 1340 even more inefficient when included in every signaling message. 1341 With public key based protection for idempotent messages, there is 1342 additionally a risk of introducing denial of service attacks. 1344 o RSVP Path message functionality and next-hop discovery 1346 To protect an RSVP signaling message (and a RSVP Path message in 1347 particular) it is necessary to know the identity of the next 1348 RSVP-aware node (and some other parameters). Without a mechanism 1349 for next-hop discovery, an RSVP Path message is also responsible 1350 for this task. Without knowing the identity of the next hop, the 1351 Kerberos principal name is also unknown. The so-called Kerberos 1352 user-to-user authentication mechanism, which would allow the 1353 receiver to trigger the process of establishing Kerberos 1354 authentication, is not supported. This issue will again be 1355 discussed in relationship with the last-hop problem. 1357 It is fair to assume that a RSVP-supporting node might not have 1358 security associations with all immediately neighboring RSVP nodes. 1359 Especially for inter-domain signaling, IntServ over DiffServ, or 1360 some new applications such as firewall signaling, the next 1361 RSVP-aware node might not be known in advance. The number of next 1362 RSVP nodes might be considerably large if they are separated by a 1363 large number of non-RSVP aware nodes. Hence, a node transmitting 1364 a RSVP Path message might experience difficulties in properly 1365 protecting the message if it serves as a mechanism to detect both 1366 the next RSVP node (i.e., Router Alert Option added to the 1367 signaling message and addressed to the destination address) and to 1368 detect route changes. It is fair to note that in an intra-domain 1369 case with a dense distribution of RSVP nodes this might be 1370 possible with manual configuration. 1372 Nothing prevents an adversary from continuously flooding an RSVP 1373 node with bogus PathErr messages, although it might be possible to 1374 protect the PathErr message with an existing, available security 1375 association. A legitimate RSVP node would believe that a change 1376 in the path took place. Hence, this node might try to select a 1377 different security association or try to create one with the 1378 indicated node. If an adversary is located somewhere along the 1379 path and either authentication or authorization is not performed 1380 with the necessary strength and accuracy, then it might also be 1381 possible to act as a man-in-the-middle. One method of reducing 1382 susceptibility to this attack is as follows: when a PathErr 1383 message is received from a node with which no security association 1384 exists, attempt to establish a security association and then 1385 repeat the action that led to the PathErr message. 1387 5.3 Last-Hop Issue 1389 This section tries to address practical difficulties when 1390 authentication and key establishment are accomplished with a 1391 two-party protocol that shows some asymmetry in message processing. 1392 Kerberos is such a protocol and also the only supported protocol that 1393 provides dynamic session key establishment for RSVP. For first-hop 1394 communication, authentication is typically done between a user and 1395 some router (for example the access router). Especially in a mobile 1396 environment, it is not feasible to authenticate end hosts based on 1397 their IP or MAC address. To illustrate this problem, the typical 1398 processing steps for Kerberos are shown for first-hop communication: 1400 1. The end host A learns the identity (i.e., Kerberos principal 1401 name) of some entity B. This entity B is either the next RSVP 1402 node, a PDP, or the next policy-aware RSVP node. 1403 2. Entity A then requests a ticket granting ticket for the network 1404 domain. This assumes that the identity of the network domain is 1405 known. 1406 3. Entity A then requests a service ticket for entity B, whose name 1407 was learned in step (a). 1408 4. Entity A includes the service ticket with the RSVP signaling 1409 message (inside the policy object). The Kerberos session key is 1410 used to protect the integrity of the entire RSVP signaling 1411 message. 1413 For last-hop communication this processing step theoretically has to 1414 be reversed; entity A is then a node in the network (for example the 1415 access router) and entity B is the other end host (under the 1416 assumption that RSVP signaling is accomplished between two end hosts 1417 and not between an end host and a application server). The access 1418 router might, however, in step (a) not be able to learn the user's 1419 principal name, because this information might not be available. 1421 Entity A could reverse the process by triggering an IAKERB exchange. 1422 This would cause entity B to request a service ticket for A as 1423 described above. IAKERB is however not supported in RSVP. 1425 5.4 RSVP and IPsec protected data traffic 1427 QoS signaling requires flow information to be established at routers 1428 along a path. This flow identifier installed at each device tells 1429 the router which data packets should receive QoS treatment. RSVP 1430 typically establishes a flow identifier based on the 5-tuple (source 1431 IP address, destination IP address, transport protocol type, source 1432 port, and destination port). If this 5-tuple information is not 1433 available, then other identifiers have to be used. IPsec-protected 1434 data traffic is such an example where the transport protocol and the 1435 port numbers are not accessible. Hence the IPsec SPI is used as a 1436 substitute for them. [13] considers these IPsec implications for 1437 RSVP and is based on three assumptions: 1439 1. An end host, which initiates the RSVP signaling message exchange, 1440 has to be able to retrieve the SPI for given flow. This requires 1441 some interaction with the IPsec security association database 1442 (SAD) and security policy database (SPD) [3]. An application 1443 usually does not know the SPI of the protected flow and cannot 1444 provide the desired values. It can provide the signaling 1445 protocol daemon with flow identifiers. The signaling daemon 1446 would then need to query the SAD by providing the flow 1447 identifiers as input parameters and the SPI as an output 1448 parameter. 1449 2. [13] assumes end-to-end IPsec protection of the data traffic. If 1450 IPsec is applied in a nested fashion, then parts of the path do 1451 not experience QoS treatment. This can be treated as a tunneling 1452 problem, but it is initiated by the end host. A figure better 1453 illustrates the problem in the case of enforcing secure network 1454 access: 1456 +------+ +---------------+ +--------+ +-----+ 1457 | Host | | Security | | Router | | Host| 1458 | A | | Gateway (SGW) | | Rx | | B | 1459 +--+---+ +-------+-------+ +----+---+ +--+--+ 1460 | | | | 1461 |IPsec-Data( | | | 1462 | OuterSrc=A, | | | 1463 | OuterDst=SGW, | | | 1464 | SPI=SPI1, | | | 1465 | InnerSrc=A, | | | 1466 | OuterDst=B, | | | 1467 | Protocol=X, |IPsec-Data( | | 1468 | SrcPort=Y, | SrcIP=A, | | 1469 | DstPort=Z) | DstIP=B, | | 1470 |=====================>| Protocol=X, |IPsec-Data( | 1471 | | SrcPort=Y, | SrcIP=A, | 1472 | --IPsec protected-> | DstPort=Z) | DstIP=B, | 1473 | data traffic |------------------>| Protocol=X, | 1474 | | | SrcPort=Y, | 1475 | | | DstPort=Z) | 1476 | | |---------------->| 1477 | | | | 1478 | | --Unprotected data traffic-> | 1479 | | | | 1481 Figure 7: RSVP and IPsec protected data traffic 1483 Host A transmitting data traffic would either indicate a 3-tuple 1484 or a 5-tuple . In any case it is 1485 not possible to make a QoS reservation for the entire path. Two 1486 similar examples are remote access using a VPN and protection of 1487 data traffic between a home agent (or a security gateway in the 1488 home network) and a mobile node. With a nested application of 1489 IPsec (for example, IPsec between A and SGW and between A and B) 1490 the same problem occurs. 1491 One possible solution to this problem is to change the flow 1492 identifier along the path to capture the new flow identifier 1493 after an IPsec endpoint. 1494 IPsec tunnels that neither start nor terminate at one of the 1495 signaling end points (for example between two networks) should be 1496 addressed differently by recursively applying an RSVP signaling 1497 exchange for the IPsec tunnel. RSVP signaling within tunnels is 1498 addressed in [14]. 1499 3. It is assumed that SPIs do not change during the lifetime of the 1500 established QoS reservation. If a new IPsec SA is created, then 1501 a new SPI is allocated for the security association. To reflect 1502 this change, either a new reservation has to be established or 1503 the flow identifier of the existing reservation has to be 1504 updated. Because IPsec SAs usually have a longer lifetime, this 1505 does not seem to be a major issue. IPsec protection of SCTP data 1506 traffic might more often require an IPsec SA (and an SPI) change 1507 to reflect added and removed IP addresses from an SCTP 1508 association. 1510 5.5 End-to-End Security Issues and RSVP 1512 End-to-end security for RSVP has not been discussed throughout the 1513 document. In this context end-to-end security refers to credentials 1514 transmitted between the two end hosts using RSVP. It is obvious that 1515 care must be taken to ensure that routers along the path are able to 1516 process and modify the signaling messages according to prescribed 1517 processing procedures. Some objects or mechanisms, however, could be 1518 used for end-to-end protection. The main question however is what 1519 the benefit of such an end-to-end security is. First, there is the 1520 question of how to establish the required security association. 1521 Between two arbitrary hosts on the Internet this might turn out to be 1522 quite difficult. Furthermore, te usefulness of end-to-end security 1523 depends on the architecture in which RSVP is deployed. If RSVP is 1524 only used to signal QoS information into the network, and other 1525 protocols have to be executed beforehand to negotiate the parameters 1526 and to decide which entity is charged for the QoS reservation, then 1527 no end-to-end security is likely to be required. Introducing 1528 end-to-end security to RSVP would then cause problems with extensions 1529 like RSVP proxy [40], Localized RSVP [41], and others that terminate 1530 RSVP signaling somewhere along the path without reaching the 1531 destination end host. Such a behavior could then be interpreted as a 1532 man-in-the-middle attack. 1534 5.6 IPsec protection of RSVP signaling messages 1536 It is assumed throughout that RSVP signaling messages can also be 1537 protected by IPsec [3] in a hop-by-hop fashion between two adjacent 1538 RSVP nodes. RSVP, however, uses special processing of signaling 1539 messages, which complicates IPsec protection. As explained in this 1540 section, IPsec should only be used for protection of RSVP signaling 1541 messages in a point-to-point communication environment (i.e., a RSVP 1542 message can only reach one RSVP router and not possibly more than 1543 one). This restriction is caused by the combination of signaling 1544 message delivery and discovery into a single message. Furthermore, 1545 end-to-end addressing complicates IPsec handling considerably. This 1546 section describes at least some of these complications. 1548 RSVP messages are transmitted as raw IP packets with protocol number 1549 46. It might be possible to encapsulate them in UDP as described in 1550 Appendix C of [6]. Some RSVP messages (Path, PathTear, and ResvConf) 1551 must have the Router Alert IP Option set in the IP header. These 1552 messages are addressed to the (unicast or multicast) destination 1553 address and not to the next RSVP node along the path. Hence an IPsec 1554 traffic selector can only use these fields for IPsec SA selection. 1555 If there is only a single path (and possibly all traffic along it is 1556 protected) then there is no problem for IPsec protection of signaling 1557 messages. This type of protection is not common and might only be 1558 used to secure network access between an end host and its first-hop 1559 router. Because the described RSVP messages are addressed to the 1560 destination address instead of the next RSVP node, it is not possible 1561 to use IPsec ESP [21] or AH [20] in transport mode--only IPsec in 1562 tunnel mode is possible. 1564 5.7 Authorization 1566 [37] describes two trust models (NJ Turnpike and NJ Parkway) and two 1567 authorization models (per-session and per-channel financial 1568 settlement). The NJ Turnpike model gives a justification for 1569 hop-by-hop security protection. RSVP focuses on the NJ Turnpike 1570 model although the different trust models are not described in 1571 detail. RSVP supports the NJ Parkway model and per-channel financial 1572 settlement only to a certain extent. Authentication of the user (or 1573 end host) can be provided with the user identity representation 1574 mechanism but authentication might in many cases be insufficient for 1575 authorization. The communication procedures defined for policy 1576 objects [42] can be improved to support the more efficient 1577 per-channel financial settlement model by avoiding policy handling 1578 between inter-domain networks at a signaling message granularity. 1579 Additional information about expected behavior of policy handling in 1580 RSVP can also be obtained from [43]. 1582 [38] and [39] provide additional information on authorization. No 1583 good and agreed mechanism for dealing with authorization of QoS 1584 reservations in roaming environments is provided. Price distribution 1585 mechanisms are only described in papers and never made their way 1586 through standardization. RSVP focuses on receiver-initiated 1587 reservations with authorization for the QoS reservation by the data 1588 receiver which introduces a fair number of complexity for mobility 1589 handling as described, for example, in [39]. 1591 6. Conclusions 1593 RSVP was the first QoS signaling protocol that provided some security 1594 protection. Whether RSVP provides enough security protection heavily 1595 depends on the environment where it is deployed. RSVP as specified 1596 today should be seen as a building block that has to be adapted to a 1597 given architecture. 1599 This document aims to provide more insights into the security of 1600 RSVP. It cannot not be interpreted as a pass or fail evaluation of 1601 the security provided by RSVP. 1603 Certainly this document is not a complete description of all security 1604 issues related to RSVP. Some issues that require further 1605 consideration are RSVP extensions (for example [13]), multicast 1606 issues, and other security properties like traffic analysis. 1607 Additionally, the interaction with mobility protocols (micro- and 1608 macro-mobility) from a security point of view demands further 1609 investigation. 1611 What can be learned from practical protocol experience and from the 1612 increased awareness regarding security is that some of the available 1613 credential types have received more acceptance than others. Kerberos 1614 is a system that is integrated into many IETF protocols today. 1615 Public key based authentication techniques are however still 1616 considered to be too heavy-weight (computationally and from a 1617 bandwidth perspective) to be used for per-flow signaling. The 1618 increased focus on denial of service attacks put additional demands 1619 on the design of public key based authentication. 1621 The following list briefly summarizes a few security or architectural 1622 issues that deserve improvement: 1624 o Discovery and signaling message delivery should be separated. 1625 o For some applications and scenarios it cannot be assumed that 1626 neighboring RSVP-aware nodes know each other. Hence some in-path 1627 discovery mechanism should be provided. 1628 o Addressing for signaling messages should be done in a hop-by-hop 1629 fashion. 1630 o Standard security protocols (IPsec, TLS or CMS) should be used 1631 whenever possible. Authentication and key exchange should be 1632 separated from signaling message protection. In general, it is 1633 necessary to provide key management to establish security 1634 associations dynamically for signaling message protection. 1635 Relying on manually configured keys between neighboring RSVP nodes 1636 is insufficient. A separate, less frequently executed key 1637 management and security association establishment protocol is a 1638 good place to perform entity authentication, security service 1639 negotiation and selection, and agreement on mechanisms, 1640 transforms, and options. 1641 o The use of public key cryptography in authorization tokens, 1642 identity representations, selective object protection, etc. is 1643 likely to cause fragmentation, the need to protect against denial 1644 of service attacks, and other problems. 1645 o Public key authentication and user identity confidentiality 1646 provided with RSVP require some improvement. 1647 o Public key based user authentication only provides entity 1648 authentication. An additional security association is required to 1649 protect signaling messages. 1650 o Data origin authentication should not be provided by non-RSVP 1651 nodes (such as the PDP). Such a procedure could be accomplished 1652 by entity authentication during the authentication and key 1653 exchange phase. 1654 o Authorization and charging should be better integrated into the 1655 base protocol. 1656 o Selective message protection should be provided. A protected 1657 message should be recognizable from a flag in the header. 1658 o Confidentiality protection is missing and should therefore be 1659 added to the protocol. The general principle is that protocol 1660 designers can seldom foresee all of the environments in which 1661 protocols will be run, so they should allow users to select from a 1662 full range of security services, as the needs of different user 1663 communities vary. 1664 o Parameter and mechanism negotiation should be provided. 1666 7. Security Considerations 1668 This document discusses security properties of RSVP and, as such, it 1669 is concerned entirely with security. 1671 8. IANA considerations 1673 This document does not address any IANA considerations. 1675 9. Acknowledgments 1677 We would like to thank Jorge Cuellar, Robert Hancock, Xiaoming Fu, 1678 Guenther Schaefer, Marc De Vuyst, Bob Grillo and Jukka Manner for 1679 their valuable comments. Additionally, we would like to thank Robert 1680 and Jorge for their time to discuss various issues with me. 1682 Finally we would Allison Mankin and John Loughney for their comments. 1684 10. References 1686 10.1 Normative References 1688 [1] Baker, F., Lindell, B. and M. Talwar, "Identity Representation 1689 for RSVP", January 2000. 1691 [2] Herzog, S., "RSVP Extensions for Policy Control", January 2000. 1693 [3] Kent, S., Atkinson, R. and M. Talwar, "Security Architecture 1694 for the Internet Protocol", November 1998. 1696 [4] Krawczyk, H., Bellare, M. and R. Canetti, "HMAC: Keyed-Hashing 1697 for Message Authentication", February 1997. 1699 [5] Rivest, R., "The MD5 Message-Digest Algorithm", April 1992. 1701 [6] Braden, R., Zhang, L., Berson, S., Herzog, S. and S. Jamin, 1702 "Resource ReSerVation Protocol (RSVP) - Version 1 Functional 1703 Specification", September 1997. 1705 [7] Yadav, S., Yavatkar, R., Pabbati, R., Ford, P., Moore, T., 1706 Herzog, S. and R. Hess, "Identity Representation for RSVP", 1707 October 2001. 1709 [8] Kohl, J. and C. Neuman, "The Kerberos Network Authentication 1710 Service (V5)", September 1993. 1712 [9] Calhoun, P., Loughney, J., Guttman, E., Zorn, G. and J. Arkko, 1713 "Diameter Base Protocol", RFC 3588, September 2003. 1715 [10] Boyle, J., Cohen, R., Durham, D., Herzog, S., Rajan, R. and A. 1716 Sastry, "The COPS(Common Open Policy Service) Protocol", 1717 January 2000. 1719 [11] Boyle, J., Cohen, R., Durham, D., Herzog, S., Rajan, R. and A. 1720 Sastry, "COPS usage for RSVP", January 2000. 1722 [12] Raeburn, K., "Encryption and Checksum Specifications for 1723 Kerberos 5", draft-ietf-krb-wg-crypto-07 (work in progress), 1724 February 2004. 1726 [13] Berger, L. and T. O'Malley, "RSVP Extensions for IPSEC Data 1727 Flows", September 1997. 1729 [14] Terzis, A., Krawczyk, J., Wroclawski, J. and L. Zhang, "RSVP 1730 Operation Over IP Tunnels", January 2000. 1732 [15] Tung, B. and L. Zhu, "Public Key Cryptography for Initial 1733 Authentication in Kerberos", draft-ietf-cat-kerberos-pk-init-24 1734 (work in progress), February 2005. 1736 [16] Kaufman, C., "Internet Key Exchange (IKEv2) Protocol", 1737 draft-ietf-ipsec-ikev2-17 (work in progress), October 2004. 1739 [17] Thomas, M. and J. Vilhuber, "Kerberized Internet Negotiation of 1740 Keys (KINK)", draft-ietf-kink-kink-06 (work in progress), July 1741 2004. 1743 10.2 Informative References 1745 [18] Hess, R. and S. Herzog, "RSVP Extensions for Policy Control", 1746 Internet-Draft(Expired) draft-ietf-rap-new-rsvp-ext-00.txt, 1747 June 2001. 1749 [19] "Secure Hash Standard,NIST, FIPS PUB 180-1", April 1995. 1751 [20] Kent, S. and R. Atkinson, "IP Authentication Header", November 1752 1998. 1754 [21] Kent, S. and R. Atkinson, "IP Encapsulating Security Payload 1755 (ESP)", November 1998. 1757 [22] Housley, R., Ford, W., Polk, W. and D. Solo, "Internet X.509 1758 Public Key Infrastructure Certificate and CRL Profile", January 1759 1999. 1761 [23] Callas, J., Donnerhacke, L., Finney, H. and R. Thayer, "OpenPGP 1762 Message Format", November 1998. 1764 [24] Hornstein, K. and J. Altman, "Distributing Kerberos KDC and 1765 Realm Information with DNS", Internet-Draft(Expired) 1766 draft-ietf-krb-wg-krb-dns-locate-03.txt, July 2002. 1768 [25] Dobbertin, H., Bosselaers, A. and B. Preneel, "RIPEMD-160: A 1769 strengthened version of RIPEMD in Fast Software Encryption, 1770 LNCS Vol 1039, pp. 71-82", 1996. 1772 [26] Dobbertin, H., "The Status of Md5 After a Recent Attack, RSA 1773 Laboratories CryptoBytes, Volume 2, Number 2", 1996. 1775 [27] Blunk, L. and J. Vollbrecht, "PPP Extensible Authentication 1776 Protocol (EAP)", March 1998. 1778 [28] Rigney, C., Willens, S., Rubens, A. and W. Simpson, "Remote 1779 Authentication Dial In User Service (RADIUS)", June 2000. 1781 [29] ""Microsoft Authorization Data Specification v. 1.0 for 1782 Microsoft Windows 2000 Operating Systems", April 2000. 1784 [30] Cable Television Laboratories, Inc.,, "PacketCable Security 1785 Specification,PKT-SP-SEC-I01-991201", website 1786 http://www.PacketCable.com/ , June 2003. 1788 [31] Myers, M., Ankney, R., Malpani, A., Galperin, S. and C. Adams, 1789 "X.509 Internet Public Key Infrastructure Online Certificate 1790 Status Protocol - OCSP", June 1999. 1792 [32] Malpani, A., Hoffman, P., Housley, R. and T. Freeman, "Simple 1793 Certificate Validation Protocol (SCVP)", Internet-Draft(Work in 1794 progress) draft-ietf-pkix-scvp-11.txt, December 2002. 1796 [33] Housley, R., "Cryptographic Message Syntax", June 1999. 1798 [34] Kaliski, B., "PKCS #7: Cryptographic Message Syntax Version 1799 1.5", March 1998. 1801 [35] "Specifications and standard documents", website 1802 http://www.PacketCable.com/ , March 2002. 1804 [36] Davis, D. and D. Geer, "Kerberos With Clocks Adrift: History, 1805 Protocols and Implementation in "USENIX Computing Systems 1806 Volume 9 no. 1, Winter", 1996. 1808 [37] Tschofenig, H., Buechli, M., Van den Bosch, S. and H. 1809 Schulzrinne, "NSIS Authentication, Authorization and Accounting 1810 Issues", Internet-Draft(Work in progress) 1811 draft-tschofenig-nsis-aaa-issues-01.txt, March 2003. 1813 [38] Tschofenig, H., Buechli, M., Van den Bosch, S., Schulzrinne, H. 1814 and T. Chen, "QoS NSLP Authorization Issues", 1815 Internet-Draft(Work in progress) 1816 draft-tschofenig-nsis-qos-authz-issues-00.txt, June 2003. 1818 [39] Thomas, M., "Analysis of Mobile IP and RSVP Interactions", 1819 Internet-Draft(Work in progress) 1820 draft-thomas-nsis-rsvp-analysis-00.txt, October 2002. 1822 [40] Gai, S., Dutt, D., Elfassy, N. and Y. Bernet, "RSVP Proxy", 1823 Internet-Draft(Expired) draft-ietf-rsvp-proxy-03.txt, March 1824 2002. 1826 [41] Manner, J., Suihko, T., Kojo, M., Liljeberg, M. and K. 1827 Raatikainen, "Localized RSVP", Internet-Draft(Expired) 1828 draft-manner-lrsvp-00.txt, May 2002. 1830 [42] Herzog, S., "Accounting and Access Control in RSVP,", PhD 1831 Dissertation,", Internet-Draft(Expired) 1832 draft-ietf-rsvp-lpm-arch-00.txt, November 1995. 1834 [43] Herzog, S., "Accounting and Access Control for Multicast 1835 Distributions: Models and Mechanisms", June 1996. 1837 [44] Pato, J., "Using Pre-Authentication to Avoid Password Guessing 1838 Attacks ,Open Software Foundation DCE Request for Comments", 1839 December 1992. 1841 [45] Wu, T., "A Real-World Analysis of Kerberos Password Security", 1842 February 1999. 1844 [46] Wu, T., Wu, F. and F. Gong, "Securing QoS: Threats to RSVP 1845 Messages and Their Countermeasures in "IEEE IWQoS, pp. 62-64", 1846 1999. 1848 [47] Talwar, V., Nahrstedt, K. and F. Gong, "Securing RSVP For 1849 Multimedia Applications in "Proceedings of ACM Multimedia 1850 (Multimedia Security Workshop)"", November 2000. 1852 [48] Talwar, V., Nahrstedt, K. and S. Nath, "RSVP-SQoS : A Secure 1853 RSVP Protocol in "International Conference on Multimedia and 1854 Exposition", Tokyo , Japan", August 2001. 1856 [49] Jablon, D., "Strong password-only authenticated key exchange 1857 Computer Communication Review, 26(5), pp. 5-26", 1858 Internet-Draft(Expired) draft-ietf-rap-new-rsvp-ext-00.txt, 1859 October 1996. 1861 [50] Harkins, D. and D. Carrel, "The Internet Key Exchange (IKE)", 1862 November 1998. 1864 Authors' Addresses 1866 Hannes Tschofenig 1867 Siemens 1868 Otto-Hahn-Ring 6 1869 Munich, Bavaria 81739 1870 Germany 1872 EMail: Hannes.Tschofenig@siemens.com 1873 Richard Graveman 1874 RFG Security 1875 15 Park Avenue 1876 Morristown, NJ 07960 1877 USA 1879 EMail: rfg@acm.org 1881 Appendix A. Dictionary Attacks and Kerberos 1883 Kerberos might be used with RSVP as described in this document. 1884 Because dictionary attacks are often mentioned in relationship with 1885 Kerberos, a few issues are addressed here. 1887 The initial Kerberos AS_REQ request (without pre-authentication, 1888 without various extensions, and without PKINIT) is unprotected. The 1889 response message AS_REP is encrypted with the client's long-term key. 1890 An adversary can take advantage of this fact by requesting AS_REP 1891 messages to mount an off-line dictionary attack. Pre-authentication 1892 ([44]) can be used to reduce this problem. However, 1893 pre-authentication does not entirely prevent dictionary attacks by an 1894 adversary who can still eavesdrop on Kerberos messages along the path 1895 between a mobile node and a KDC. With mandatory pre-authentication 1896 for the initial request, an adversary cannot request a Ticket 1897 Granting Ticket for an arbitrary user. On-line password guessing 1898 attacks are still possible by choosing a password (e.g., from a 1899 dictionary) and then transmitting an initial request including a 1900 pre-authentication data field. An unsuccessful authentication by the 1901 KDC results in an error message and the gives the adversary a hint to 1902 restart the protocol and try a new password. 1904 There are, however, some proposals that prevent dictionary attacks. 1905 The use of Public Key Cryptography for initial authentication [15] 1906 (PKINIT) is one such solution. Other proposals use 1907 strong-password-based authenticated key agreement protocols to 1908 protect the user's password during the initial Kerberos exchange. 1909 [45] discusses the security of Kerberos and also discusses mechanisms 1910 to prevent dictionary attacks. 1912 Appendix B. Example of User-to-PDP Authentication 1914 The following Section describes an example of user-to-PDP 1915 authentication. Note that the description below is not fully covered 1916 by the RSVP specification and hence it should only be seen as an 1917 example. 1919 Windows 2000, which integrates Kerberos into RSVP, uses a 1920 configuration with the user authentication to the PDP as described in 1921 [29]. The steps for authenticating the user to the PDP in an 1922 intra-realm scenario are the following: 1924 o Windows 2000 requires the user to contact the KDC and to request a 1925 Kerberos service ticket for the PDP account AcsService in the 1926 local realm . 1927 o This ticket is then embedded into the AUTH_DATA element and 1928 included in either the PATH or the RESV message. In case of 1929 Microsoft's implementation, the user identity encoded as a 1930 distinguished name is encrypted with the session key provided with 1931 the Kerberos ticket. The Kerberos ticket is sent without the 1932 Kerberos authdata element that contains authorization information, 1933 as explained in [29]. 1934 o The RSVP message is then intercepted by the PEP, which forwards it 1935 to the PDP. [29] does not state which protocol is used to forward 1936 the RSVP message to the PDP. 1937 o The PDP that finally receives the message decrypts the received 1938 service ticket. The ticket contains the session key used by the 1939 user's host to 1940 * Encrypt the principal name inside the policy locator field of 1941 the AUTH_DATA object and to 1942 * Create the integrity-protected Keyed Message Digest field in 1943 the INTEGRITY object of the POLICY_DATA element. The 1944 protection described here is between the user's host and the 1945 PDP. The RSVP INTEGRITY object on the other hand is used to 1946 protect the path between the user's host and the first-hop 1947 router, because the two message parts terminate at different 1948 nodes and different security associations must be used. The 1949 interface between the message-intercepting, first-hop router 1950 and the PDP must be protected as well. 1951 * The PDP does not maintain a user database, and [29] describes 1952 how the PDP may query the Active Directory (a LDAP based 1953 directory service) for user policy information. 1955 Appendix C. Literature on RSVP Security 1957 Few documents address the security of RSVP signaling. This section 1958 briefly describes some important documents. 1960 Improvements to RSVP are proposed in [46] to deal with insider 1961 attacks. Insider attacks are caused by malicious RSVP routers that 1962 modify RSVP signaling messages in such a way that they cause harm to 1963 the nodes participating in the signaling message exchange. 1965 As a solution, non-mutable RSVP objects are digitally signed by the 1966 sender. This digital signature is added to the RSVP PATH message. 1967 Additionally, the receiver attaches an object to the RSVP RESV 1968 message containing a "signed" history. This value allows 1969 intermediate RSVP routers (by examining the previously signed value) 1970 to detect a malicious RSVP node. 1972 A few issues are, however, left open in the document. Replay attacks 1973 are not covered, and it is therefore assumed that timestamp-based 1974 replay protection is used. To detect a malicious node, it is 1975 necessary that all routers along the path are able to verify the 1976 digital signature. This may require a global public key 1977 infrastructure and also client-side certificates. Furthermore the 1978 bandwidth and computational requirements to compute, transmit, and 1979 verify digital signatures for each signaling message might place a 1980 burden on a real-world deployment. 1982 Authorization is not considered in the document, which might have an 1983 influence on the implications of signaling message modification. 1984 Hence, the chain-of-trust relationship (or this step in a different 1985 direction) should be considered in relationship with authorization. 1987 In [47], the above-described idea of detecting malicious RSVP nodes 1988 is improved by addressing performance aspects. The proposed solution 1989 is somewhere between hop-by-hop security and the approach in [46], 1990 insofar as it separates the end-to-end path into individual networks. 1991 Furthermore, some additional RSVP messages (e.g., feedback messages) 1992 are introduced to implement a mechanism called "delayed integrity 1993 checking." In [48], the approach presented in [47] is enhanced. 1995 Intellectual Property Statement 1997 The IETF takes no position regarding the validity or scope of any 1998 Intellectual Property Rights or other rights that might be claimed to 1999 pertain to the implementation or use of the technology described in 2000 this document or the extent to which any license under such rights 2001 might or might not be available; nor does it represent that it has 2002 made any independent effort to identify any such rights. Information 2003 on the procedures with respect to rights in RFC documents can be 2004 found in BCP 78 and BCP 79. 2006 Copies of IPR disclosures made to the IETF Secretariat and any 2007 assurances of licenses to be made available, or the result of an 2008 attempt made to obtain a general license or permission for the use of 2009 such proprietary rights by implementers or users of this 2010 specification can be obtained from the IETF on-line IPR repository at 2011 http://www.ietf.org/ipr. 2013 The IETF invites any interested party to bring to its attention any 2014 copyrights, patents or patent applications, or other proprietary 2015 rights that may cover technology that may be required to implement 2016 this standard. Please address the information to the IETF at 2017 ietf-ipr@ietf.org. 2019 Disclaimer of Validity 2021 This document and the information contained herein are provided on an 2022 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 2023 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET 2024 ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, 2025 INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE 2026 INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 2027 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 2029 Copyright Statement 2031 Copyright (C) The Internet Society (2005). This document is subject 2032 to the rights, licenses and restrictions contained in BCP 78, and 2033 except as set forth therein, the authors retain all their rights. 2035 Acknowledgment 2037 Funding for the RFC Editor function is currently provided by the 2038 Internet Society.