idnits 2.17.1 draft-iab-privsec-confidentiality-mitigations-07.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([RFC7624]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (October 11, 2016) is 2754 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC2119' is defined on line 400, but no explicit reference was found in the text == Unused Reference: 'STRINT' is defined on line 468, but no explicit reference was found in the text -- Obsolete informational reference (is this intentional?): RFC 4306 (Obsoleted by RFC 5996) -- Obsolete informational reference (is this intentional?): RFC 5246 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 5750 (Obsoleted by RFC 8550) -- Obsolete informational reference (is this intentional?): RFC 6962 (Obsoleted by RFC 9162) Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 5 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IAB T. Hardie, Ed. 3 Internet-Draft October 11, 2016 4 Intended status: Informational 5 Expires: April 14, 2017 7 Confidentiality in the Face of Pervasive Surveillance 8 draft-iab-privsec-confidentiality-mitigations-07 10 Abstract 12 The IAB has published [RFC7624] in response to several revelations of 13 pervasive attack on Internet communications. This document surveys 14 the most plausible mitigations to those threats currently available 15 to the designers of Internet protocols. 17 Status of This Memo 19 This Internet-Draft is submitted in full conformance with the 20 provisions of BCP 78 and BCP 79. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF). Note that other groups may also distribute 24 working documents as Internet-Drafts. The list of current Internet- 25 Drafts is at http://datatracker.ietf.org/drafts/current/. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 This Internet-Draft will expire on April 14, 2017. 34 Copyright Notice 36 Copyright (c) 2016 IETF Trust and the persons identified as the 37 document authors. All rights reserved. 39 This document is subject to BCP 78 and the IETF Trust's Legal 40 Provisions Relating to IETF Documents 41 (http://trustee.ietf.org/license-info) in effect on the date of 42 publication of this document. Please review these documents 43 carefully, as they describe your rights and restrictions with respect 44 to this document. Code Components extracted from this document must 45 include Simplified BSD License text as described in Section 4.e of 46 the Trust Legal Provisions and are provided without warranty as 47 described in the Simplified BSD License. 49 Table of Contents 51 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 52 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 2 53 3. Available Mitigations . . . . . . . . . . . . . . . . . . . . 2 54 4. Interplay among Mechanisms . . . . . . . . . . . . . . . . . 8 55 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 8 56 6. Security Considerations . . . . . . . . . . . . . . . . . . . 8 57 7. Contributors {Contributors} . . . . . . . . . . . . . . . . . 9 58 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 9 59 8.1. Normative References . . . . . . . . . . . . . . . . . . 9 60 8.2. Informative References . . . . . . . . . . . . . . . . . 9 61 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 11 63 1. Introduction 65 To ensure that the Internet can be trusted by users, it is necessary 66 for the Internet technical community to address the vulnerabilities 67 exploited in the attacks document in [RFC7258] and the threats 68 described in [RFC7624]. The goal of this document is to describe 69 more precisely the mitigations available for those threats and to lay 70 out the interactions among them should they be deployed in 71 combination. 73 2. Terminology 75 This document makes extensive use of standard security and privacy 76 terminology; see [RFC4949] and [RFC6973]. Terms used from [RFC6973] 77 include Eavesdropper, Observer, Initiator, Intermediary, Recipient, 78 Attack (in a privacy context), Correlation, Fingerprint, Traffic 79 Analysis, and Identifiability (and related terms). In addition, we 80 use a few terms that are specific to the attacks discussed in 81 [RFC7624]. Note especially that "passive" and "active" below do not 82 refer to the effort used to mount the attack; a "passive attack" is 83 any attack that accesses a flow but does not modify it, while an 84 "active attack" is any attack that modifies a flow. Some passive 85 attacks involve active interception and modifications of devices, 86 rather than simple access to the medium. 88 3. Available Mitigations 90 Given the threat model laid out in [RFC7624], how should the Internet 91 technical community respond to pervasive attack? The cost and risk 92 considerations discussed in it provide a guide to responses. Namely, 93 responses to passive attack should close off avenues for those 94 attacks that are safe, scalable, and cheap, forcing the attacker to 95 mount attacks that expose it to higher cost and risk. Protocols and 96 security measures protecting against active attacks must also limit 97 the impact of compromise and malfeasance by avoiding systems which 98 grant universal credentials. 100 In this section, we discuss a collection of high-level approaches to 101 mitigating pervasive attacks. These approaches are not meant to be 102 exhaustive, but rather to provide general guidance to protocol 103 designers in creating protocols that are resistant to pervasive 104 attack. 106 Many of these are basic tools which already exist. As Edward Snowden 107 put it, "properly implemented strong crypto systems are one of the 108 few things you can rely on". The task for the Internet community is 109 to ensure that applications are able to use the strong crypto and 110 other mitigations already available- and that these are properly 111 implemented and commonly turned on. Some of this work will require 112 architectural changes to applications, e.g., in order to limit the 113 information that is exposed to servers. In many other cases, 114 however, the need is simply to make the best use we can of the 115 cryptographic tools we have. 117 +--------------------------+----------------------------------------+ 118 | Attack Class | High-level mitigations | 119 +--------------------------+----------------------------------------+ 120 | Passive observation | Encryption for confidentiality | 121 | | | 122 | Passive inference | Path differentiation | 123 | | | 124 | Active | Authentication, monitoring | 125 | | | 126 | Metadata Analysis | Data Minimization | 127 | | | 128 | Static key exfiltration | Encryption with per-session state | 129 | | (PFS) | 130 | | | 131 | Dynamic key exfiltration | Transparency, validation of end | 132 | | systems | 133 | | | 134 | Content exfiltration | Object encryption, distributed systems | 135 +--------------------------+----------------------------------------+ 137 Figure 1: Table of Mitigations 139 The traditional mitigation to passive attack is to render content 140 unintelligible to the attacker by applying encryption, for example, 141 by using TLS or IPsec [RFC5246][RFC4301]. Even without 142 authentication, encryption will prevent a passive attacker from being 143 able to read the encrypted content. Exploiting unauthenticated 144 encryption requires an active attack (man in the middle); with 145 authentication, a key exfiltration attack is required. For 146 cryptographic systems providing forward secrecy, even exfiltration of 147 long-term keys will not compromise data captured under session keys 148 used before the exfiltration. 150 The additional capabilities of a pervasive passive attacker, however, 151 require some changes in how protocol designers evaluate what 152 information is encrypted. In addition to directly collecting 153 unencrypted data, a pervasive passive attacker can also make 154 inferences about the content of encrypted messages based on what is 155 observable. For example, if a user typically visits a particular set 156 of web sites, then a pervasive passive attacker observing all of the 157 user's behavior can track the user based on the hosts the user 158 communicates with, even if the user changes IP addresses, and even if 159 all of the connections are encrypted. 161 Thus, in designing protocols to be resistant to pervasive passive 162 attacks, protocol designers should consider what information is left 163 unencrypted in the protocol, and how that information might be 164 correlated with other traffic. Some of the data left unencrypted may 165 be considered "metadata" within the context of a single protocol, as 166 it provides adjunct information used for delivery or display, rather 167 than the data directly created or consumed by protocol users. This 168 does not mean it is not useful to attackers, however, and when this 169 metadata is not protected by encryption it may leak substantial 170 amounts of information. Data minimization strategies should thus be 171 applied to any data left unencrypted, whether it be payload or 172 metadata. Information that cannot be encrypted or omited should be 173 be dissociated from other information. For example, the TOR[TOR] 174 overlay routing network anonymizes IP addresses by using multi-hop 175 onion routing. 177 As with traditional, limited active attacks, a basic mitigation to 178 pervasive active attack is to enable the endpoints of a communication 179 to authenticate each other over the encrypted channel. However, 180 attackers that can mount pervasive active attacks can often subvert 181 the authorities on which authentication systems rely. Thus, in order 182 to make authentication systems more resilient to pervasive attack, it 183 is beneficial to monitor these authorities to detect misbehavior that 184 could enable active attack. For example, DANE and Certificate 185 Transparency both provide mechanisms for detecting when a CA has 186 issued a certificate for a domain name without the authorization of 187 the holder of that domain name [RFC6962][RFC6698]. Other systems may 188 use external notaries to detect certificate authority mismatch (e.g. 189 Convergence [Convergence]). 191 While encryption and authentication protect the security of 192 individual sessions, these sessions may still leak information, such 193 as IP addresses or server names, that a pervasive attacker can use to 194 correlate sessions and derive additional information about the 195 target. Thus, pervasive attack highlights the need for anonymization 196 technologies, which make correlation more difficult. Typical 197 approaches to anonymization against traffic analysis include: 199 o Aggregation: Routing sessions for many endpoints through a common 200 mid-point (e.g, an HTTP proxy). The midpoint appears as the 201 origin of the communication when traffic analysis is conducted 202 from points after it, so individual sources cannot be 203 distinguished. If traffic analysis is being conducted prior to 204 the mid-point, all flows appear to be destined to the same point, 205 which leaks very little information. Even when traffic analysis 206 is being performed both before and after the mid-point, 207 simultaneous connections may make it difficult to correlate the 208 traffic going into and out of the mid-point. For this to be 209 effective as a mitigation, traffic to the mid-point must be 210 encrypted and traffic from the mid-point should be. 212 o Onion routing: Routing a session through several mid-points, 213 rather than directly end-to-end, with encryption that guarantees 214 that each node can only see the previous and next hops. This 215 ensures that the source and destination of a communication are 216 never revealed simultaneously. Note, however, that onion routing 217 anonymity guarantees depend on an attacker being unable to control 218 many of the routing nodes [TorPaper]. 220 o Multi-path: Routing different sessions via different paths (even 221 if they originate from the same endpoint). This reduces the 222 probability that the same attacker will be able to collect many 223 sessions or associate them with the same individual. If, for 224 example, a device has both a cellular and 802.11 interface, 225 routing some traffic across the cellular network and other traffic 226 over the 802.11 interface means that traffic analysis conducted 227 only with one network will be incomplete. Even if conducted in 228 both, it may be more difficult for the attacker to associate the 229 traffic in each network with the other. For this to be effective 230 as a mitigation, signalling protocols which gather and transmit 231 data about multiple interfaces (such as SIP) must be encrypted to 232 avoid the information being used in cross-correlation. 234 An encrypted, authenticated session is safe from content-monitoring 235 attacks in which neither end collaborates with the attacker, but can 236 still be subverted by the endpoints. Ciphersuites used for HTTPS 237 based on RSA encryption, for example, allow an attacker with the 238 private key to derive the session keys from passive observation of a 239 session. These ciphersuites are thus vulnerable to a static key 240 exfiltration attack - if the attacker obtains the server's private 241 key once, then they can decrypt all past and future sessions for that 242 server. 244 Static key exfiltration attacks are prevented by including ephemeral, 245 per-session secret information in the keys used for a session. Most 246 IETF security protocols include modes of operation that have this 247 property. These modes are known in the literature under the heading 248 "perfect forward secrecy" (PFS) because even if an adversary has all 249 of the secrets for one session, the next session will use new, 250 different secrets and the attacker will not be able to decrypt it. 251 The Internet Key Exchange (IKE) protocol used by IPsec supports PFS 252 by default [RFC4306], and TLS supports PFS via the use of specific 253 ciphersuites [RFC5246]. 255 Dynamic key exfiltration cannot be prevented by protocol means. By 256 definition, any secrets that are used in the protocol will be 257 transmitted to the attacker and used to decrypt what the protocol 258 encrypts. Likewise, no technical means will stop a willing 259 collaborator from sharing keys with an attacker. However, this 260 attack model also covers "unwitting collaborators", whose technical 261 resources are collaborating with the attacker without their owners' 262 knowledge. This could happen, for example, if flaws are built into 263 products or if malware is injected later on. 265 Standards can also define protocols that provide greater or lesser 266 opportunity for dynamic key exfiltration. Collaborators engaging in 267 key exfiltration through a standard protocol will need to use covert 268 channels in the protocol to leak information that can be used by the 269 attacker to recover the key. Such use of covert channels has been 270 demonstrated for SSL, TLS, and SSH. Any protocol bits that can be 271 freely set by the collaborator can be used as a covert channel, 272 including, for example, TCP options or unencrypted traffic sent 273 before a STARTTLS message in SMTP or XMPP. Protocol designers should 274 consider what covert channels their protocols expose, and how those 275 channels can be exploited to exfiltrate key information. 277 Content exfiltration has some similarity to the dynamic exfiltration 278 case, in that nothing can prevent a collaborator from revealing what 279 they know, and the mitigations against becoming an unwitting 280 collaborator apply. In this case, however, applications can limit 281 what the collaborator is able to reveal. For example, the S/MIME and 282 PGP systems for secure email both deny intermediate servers access to 283 certain parts of the message [RFC5750][RFC2015]. Even if a server 284 were to provide an attacker with full access, the attacker would 285 still not be able to read the protected parts of the message. 287 Mechanisms like S/MIME and PGP are often referred to as "end-to-end" 288 security mechanisms, as opposed to "hop-by-hop" or "end-to-middle" 289 mechanisms like the use of SMTP over TLS. These two different 290 mechanisms address different types of attackers: Hop-by-hop 291 mechanisms protect from attackers on the wire (passive or active), 292 while end-to-end mechanisms protect against attackers within 293 intermediate nodes as well as those on the wire. Even end-to-end 294 mechanisms are not complete protection in themselves, as intermediate 295 nodes can still access some metadata. For example: 297 o Two users messaging via Facebook over HTTPS are protected against 298 passive and active attackers in the network between the users and 299 Facebook. However, if Facebook is a collaborator in an 300 exfiltration attack, their communications can still be monitored. 301 They would need to encrypt their messages end-to-end in order to 302 protect themselves against this risk. 304 o Two users exchanging PGP-protected email have protected the 305 content of their exchange from network attackers and intermediate 306 servers, but the header information (e.g., To and From addresses) 307 is unnecessarily exposed to passive and active attackers that can 308 see communications among the mail agents handling the email 309 messages. These mail agents need to use hop-by-hop encryption and 310 traffic analysis mitigation to address this risk. 312 Mechanisms such as S/MIME and PGP are also known as "object-based" 313 security mechanisms (as opposed to "communications security" 314 mechanisms), since they operate at the level of objects, rather than 315 communications sessions. Such secure object can be safely handled by 316 intermediaries in order to realize, for example, store and forward 317 messaging. In the examples above, the encrypted instant messages or 318 email messages would be the secure objects. Hop-to-hop security 319 mechanisms are generally difficult to retrofit onto a deployed 320 system, and they make it difficult to determine the security posture 321 of remote hops (e.g., STARTTLS-secured SMTP has been downgraded by 322 intermediate network nodes [WaPo-STARTTLS]). End-to-end mechanisms 323 are advised. 325 The mitigations to the content exfiltration case regard participants 326 in the protocol as potential passive attackers themselves, and apply 327 the mitigations discussed above with regard to passive attack. 328 Information that is not necessary for these participants to fulfill 329 their role in the protocol can be encrypted, and other information 330 can be anonymized. 332 The tools that we currently have have not generally been designed 333 with mitigation in mind, so they may need elaboration or adjustment 334 to be completely suitable. The next section examines one common 335 reason for such adjustment: managing the integration of one 336 mitigation with the environment in which it is deployed. 338 4. Interplay among Mechanisms 340 One of the key considerations in selecting mitigations is how to 341 manage the interplay among different mechanisms. Care must be taken 342 to avoid situations where a mitigation is rendered fruitless because 343 of mechanisms which working at a different time scale or with a 344 different aim. 346 As an example, there is work in progress in IEEE 802 to standardize a 347 method for the randomization of MAC Addresses. This work aims to 348 enable the MAC address to vary as the device connects to different 349 networks, or connects at different times. In theory, the 350 randomization will mitigate tracking by MAC address. However, the 351 randomization will be defeated if the adversary can link the 352 randomized MAC address to other identifiers such as the interface 353 identifier used in IPv6 addresses, the unique identifiers used in 354 DHCP or DHCPv6, or unique identifiers used in various link-local 355 discovery protocols. 357 For mitigations which rely on aggregation to separate the origin of 358 traffic from its destination, care must be taken that the protocol 359 mechanics do not expose origin IP through secondary means. 360 [I-D.ietf-dnsop-edns-client-subnet] for example, documents a method 361 to carry the IP address or subnet of a querying party through a 362 recursive resolver to an authoritative resolver. Even with a 363 truncated IP address, this mechanism increases the likelihood that a 364 pervasive monitor would be able to associate query traffic and 365 responses. If a client wished to ensure that its traffic did not 366 expose this data, it would need to require that its stub resolver 367 emit any privacy-sensitive queries with a source NETMASK set to 0, as 368 detailed in Section 5.1 of [I-D.ietf-dnsop-edns-client-subnet]. 369 Given that setting this only occasionally might also be used a signal 370 to observers, any client wishing to have any privacy sensitive 371 traffic would, in essence have to emit this for every query. While 372 this would succeed at providing the required privacy, given the 373 mechanism proposed, it would also mean no split-DNS adjustments in 374 response would be possible for the privacy sensitive client. 376 5. IANA Considerations 378 This memo makes no request of IANA. 380 6. Security Considerations 382 This memorandum describes a series of mitigations to the attacks 383 described in [RFC7258]. No such list could possibly be 384 comprehensive, nor is the attack therein described the only possible 385 attack. 387 7. Contributors {Contributors} 389 This document is derived in part from the work initially done on the 390 Perpass mailing list and at the STRINT workshop. Work from Brian 391 Trammell, Bruce Schneier, Christian Huitema, Cullen Jennings, Daniel 392 Borkmann, and Richard Barnes is incorporated here, as are ideas and 393 commentary from Jeff Hodges, Phillip Hallam-Baker, and Stephen 394 Farrell. 396 8. References 398 8.1. Normative References 400 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 401 Requirement Levels", BCP 14, RFC 2119, 402 DOI 10.17487/RFC2119, March 1997, 403 . 405 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2", 406 FYI 36, RFC 4949, DOI 10.17487/RFC4949, August 2007, 407 . 409 [RFC6973] Cooper, A., Tschofenig, H., Aboba, B., Peterson, J., 410 Morris, J., Hansen, M., and R. Smith, "Privacy 411 Considerations for Internet Protocols", RFC 6973, 412 DOI 10.17487/RFC6973, July 2013, 413 . 415 [RFC7258] Farrell, S. and H. Tschofenig, "Pervasive Monitoring Is an 416 Attack", BCP 188, RFC 7258, DOI 10.17487/RFC7258, May 417 2014, . 419 [RFC7624] Barnes, R., Schneier, B., Jennings, C., Hardie, T., 420 Trammell, B., Huitema, C., and D. Borkmann, 421 "Confidentiality in the Face of Pervasive Surveillance: A 422 Threat Model and Problem Statement", RFC 7624, 423 DOI 10.17487/RFC7624, August 2015, 424 . 426 8.2. Informative References 428 [Convergence] 429 M Marlinspike, ., "Convergence Project", August 2011, 430 . 432 [I-D.ietf-dnsop-edns-client-subnet] 433 Contavalli, C., Gaast, W., tale, t., and W. Kumari, 434 "Client Subnet in DNS Queries", draft-ietf-dnsop-edns- 435 client-subnet-08 (work in progress), April 2016. 437 [RFC2015] Elkins, M., "MIME Security with Pretty Good Privacy 438 (PGP)", RFC 2015, DOI 10.17487/RFC2015, October 1996, 439 . 441 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 442 Internet Protocol", RFC 4301, DOI 10.17487/RFC4301, 443 December 2005, . 445 [RFC4306] Kaufman, C., Ed., "Internet Key Exchange (IKEv2) 446 Protocol", RFC 4306, DOI 10.17487/RFC4306, December 2005, 447 . 449 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 450 (TLS) Protocol Version 1.2", RFC 5246, 451 DOI 10.17487/RFC5246, August 2008, 452 . 454 [RFC5750] Ramsdell, B. and S. Turner, "Secure/Multipurpose Internet 455 Mail Extensions (S/MIME) Version 3.2 Certificate 456 Handling", RFC 5750, DOI 10.17487/RFC5750, January 2010, 457 . 459 [RFC6698] Hoffman, P. and J. Schlyter, "The DNS-Based Authentication 460 of Named Entities (DANE) Transport Layer Security (TLS) 461 Protocol: TLSA", RFC 6698, DOI 10.17487/RFC6698, August 462 2012, . 464 [RFC6962] Laurie, B., Langley, A., and E. Kasper, "Certificate 465 Transparency", RFC 6962, DOI 10.17487/RFC6962, June 2013, 466 . 468 [STRINT] S Farrell, ., "Strint Workshop Report", April 2014, 469 . 472 [TOR] The Tor Project, "Tor", 2013, 473 . 475 [TorPaper] 476 Dingledine, R., Mathewson, N., and P. Syverson, "Tor: The 477 Second-Generation Onion Router", 2004, 478 . 481 [WaPo-STARTTLS] 482 Scola, N. and A. Soltani, "Mobile ISP Cricket was 483 thwarting encrypted emails, researchers find", 2014, 484 . 488 Author's Address 490 Ted Hardie (editor) 492 Email: ted.ietf@gmail.com