idnits 2.17.1 draft-ietf-intarea-broadcast-consider-02.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 13, 2017) is 2628 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-05) exists of draft-ietf-dnssd-privacy-00 == Outdated reference: A later version (-05) exists of draft-ietf-intarea-hostname-practice-00 -- Obsolete informational reference (is this intentional?): RFC 4941 (Obsoleted by RFC 8981) Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Engineering Task Force R. Winter 3 Internet-Draft M. Faath 4 Intended status: Informational F. Weisshaar 5 Expires: August 17, 2017 University of Applied Sciences Augsburg 6 February 13, 2017 8 Privacy considerations for IP broadcast and multicast protocol designers 9 draft-ietf-intarea-broadcast-consider-02 11 Abstract 13 A number of application-layer protocols make use of IP broadcasts or 14 multicast messages for functions like local service discovery or name 15 resolution. Some of these functions can only be implemented 16 efficiently using such mechanisms. When using broadcasts or 17 multicast messages, a passive observer in the same broadcast/ 18 multicast domain can trivially record these messages and analyze 19 their content. Therefore, broadcast/multicast protocol designers 20 need to take special care when designing their protocols. 22 Status of This Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on August 17, 2017. 39 Copyright Notice 41 Copyright (c) 2017 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 57 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 58 2. Privacy considerations . . . . . . . . . . . . . . . . . . . 4 59 2.1. Message frequency . . . . . . . . . . . . . . . . . . . . 4 60 2.2. Persistent identifiers . . . . . . . . . . . . . . . . . 4 61 2.3. Anticipate user behavior . . . . . . . . . . . . . . . . 5 62 2.4. Consider potential correlation . . . . . . . . . . . . . 6 63 2.5. Configurability . . . . . . . . . . . . . . . . . . . . . 6 64 3. Operational considerations . . . . . . . . . . . . . . . . . 7 65 4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 66 5. Other considerations . . . . . . . . . . . . . . . . . . . . 8 67 6. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 8 68 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 69 8. Security Considerations . . . . . . . . . . . . . . . . . . . 9 70 9. Informative References . . . . . . . . . . . . . . . . . . . 9 71 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 11 73 1. Introduction 75 Broadcast and multicast messages have a large (and to the sender 76 unknown) receiver group by design. Because of that, these two 77 mechanisms are vital for a number of basic network functions such as 78 auto-configuration or link-layer address lookup. Also application 79 developers use broadcast/multicast messages to implement things like 80 local service or peer discovery and it appears that an increasing 81 number of applications make use of it [TRAC2016]. That is not 82 entierly surprising. As RFC 919 [RFC0919] puts it, "The use of 83 broadcasts [...] is a good base for many applications". Broadcast 84 and multicast functionality in a subnetwork are therefore important 85 as a lack thereof renders the protocols underlying these mechanisms 86 inoperable [RFC3819]. 88 Using broadcast/multicast can become problematic if the information 89 that is being distributed can be regarded as sensitive or when the 90 information that is distributed by multiple of these protocols can be 91 correlated in a way that sensitive data can be derived. This is 92 clearly true for any protocol, but broadcast/multicast is special in 93 at least two respects: 95 (a) The aforementioned large receiver group, consisting of receivers 96 unknown to the sender. This makes eavesdropping without special 97 privileges or a special location in the network trivial for 98 anybody in the broadcast/multicast domain. 100 (b) Encryption is more difficult when broadcast/multicast messages, 101 leaving content of these messages in the clear and making it 102 easier to spoof and replay them. 104 Given the above, privacy protection for protocols based on broadcast 105 or multicast communication is significantly more difficult compared 106 to unicast communication and at the same time invading the privacy is 107 much easier. 109 Privacy considerations of IETF-specified protocols have received some 110 attention in the recent past (e.g. RFC 7721 [RFC7721] or RFC 7919 111 [RFC7819]). There is also general guidance available for document 112 authors on when and how to include a privacy considerations section 113 in their documents and on how to evaluate the privacy implications of 114 Internet protocols [RFC6973]. RFC6973 also describes potential 115 threats to privacy in great detail and lists terminology that is also 116 used in this document. 118 In contrast to RFC6973, this document contains a number of privacy 119 considerations especially for broadcast/multicast protocol designers 120 that are intended to reduce the likelihood that a broadcast/multicast 121 protocol can be misused to collect sensitive data about devices, 122 users and groups of users on a broadcast/multicast domain. These 123 considerations particularly apply to protocols designed outside the 124 IETF for two reasons. For one, non-standard protocols will likely 125 not receive operational attention and support in making them more 126 secure such as e.g. DHCP snooping does for DHCP because they 127 typically are not documented. The other reason is that these 128 protocols have been designed in isolation, where a set of 129 considerations to follow is useful in the absence of a larger 130 community providing feedback. In particular, carelessly designed 131 broadcast/multicast protocols can break privacy efforts at different 132 layers of the protocol stack such as MAC address or IP address 133 randomization [RFC4941]. 135 1.1. Requirements Language 137 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 138 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 139 document are to be interpreted as described in RFC 2119 [RFC2119]. 141 2. Privacy considerations 143 There are a few obvious and a few not necessarily obvious things 144 designers of broadcast/multicast protocols should consider in respect 145 to the privacy implications of their protocol. Most of these items 146 are based on protocol behavior observed as part of experiments on 147 operational networks [TRAC2016]. 149 2.1. Message frequency 151 Frequent broadcast/multicast traffic caused by an application can 152 give user behavior and online times away. This allows a passive 153 observer to potentially deduce a user's current activity (e.g. a 154 game) and it allows to create an online profile (i.e. times the user 155 is on the network). The higher the frequency of these messages, the 156 more accurate this profile will be. Given that broadcasts/multicasts 157 are only visible in the same broadcast/multicast domain, these 158 messages also give the rough location of the user away (e.g. a campus 159 or building). 161 This behavior has e.g. been observed by a synchronization mechanism 162 of a popular application, where multiple messages have been sent per 163 minute via broadcast. Given this behavior, it is possible to record 164 a device's time on the network with a sub-minute accuracy given only 165 the traffic of this single application installed on the device. But 166 also services used for local name resolution in modern operating 167 systems utilize broadcast/multicast protocols (e.g. mDNS, LLMNR or 168 NetBIOS) to announce for example their shares regularly and allow a 169 tracking of the online time of a device. 171 If a protocol relies on frequent or periodic broadcast/multicast 172 messages, the frequency SHOULD be chosen conservatively, in 173 particular if the messages contain persistent identifiers (see next 174 subsection). Also, intelligent message suppression mechanisms such 175 as the ones employed in mDNS [RFC6762] SHOULD be implemented. The 176 lower the frequency of broadcast messages, the harder traffic 177 analysis and surveillance becomes. 179 2.2. Persistent identifiers 181 A few broadcast/multicast protocols observed in the wild make use of 182 persistent identifiers. This includes the use of host names or more 183 abstract persistent identifiers such as a UUID or similar. These 184 IDs, which e.g. identify the installation of a certain application 185 might not change across updates of the software and are therefore 186 extremely long lived. This allows a passive observer to track a user 187 precisely if broadcast/multicast messages are frequent. This is even 188 true in case the IP and/or MAC address changes. Such identifiers 189 also allow two different interfaces (e.g. WiFi and Ethernet) to be 190 correlated to the same device. If the application makes use of 191 persistent identifiers for multiple installations of the same 192 application for the same user, this even allows to infer that 193 different devices belong to the same user. 195 The aforementioned broadcast messages from a synchronization 196 mechanism of a popular application also included a persistent 197 identifier in every broadcast. This identifier did never change 198 after the application was installaed and allowed to track a device 199 even when it changed its network interface or when it connected to a 200 different network. 202 If a broadcast/multicast protocol relies on IDs to be transmitted, it 203 SHOULD be considered if frequent ID rotations are possible in order 204 to make user tracking more difficult. Persistent IDs are considered 205 bad practice in general for broadcast and multicast communication as 206 persistent application layer IDs will make efforts on lower layers to 207 randomize identifiers (e.g. [I-D.huitema-6man-random-addresses]) 208 useless or at least much more difficult. 210 2.3. Anticipate user behavior 212 A large number of users name their device after themselves, either 213 using their first name, last name or both. Often a host name 214 includes the type, model or maker of a device, its function or 215 includes language specific information. Based on gathered data, this 216 appears currently to be prevalent user behavior [TRAC2016]. For 217 protocols using the host name as part of the messages, this clearly 218 will reveal personally identifiable information to everyone on the 219 local network. This information can also be used to mount more 220 sophisticated attacks, when e.g. the owner of a device is identified 221 (as an interesting target) or properties of the device are known 222 (e.g. known vulnerabilities). 224 A popular operating system vendor includes the name the user chooses 225 for the user account during the installation process as part of the 226 host name of the device. The name of the operating system is also 227 included, revealing therefore two pieces of information, which can be 228 regarded as private information if the host name is used in 229 broadcast/multicast messages. 231 Where possible, the use of host names and other user provided 232 information in broadcast/multicast protocols SHOULD be avoided. If 233 only a persistent ID is needed, this can be generated. An 234 application might want to display the information it will broadcast 235 on the LAN at install/config time, so the user is at least aware of 236 the application's behavior. More host name considerations can be 237 found in [I-D.ietf-intarea-hostname-practice]. More information on 238 user participation can be found in RFC 6973 [RFC6973]. 240 2.4. Consider potential correlation 242 A large number of services and applications make use of the 243 broadcast/multicast mechanism. That means there are various sources 244 of information that are easily accessible by a passive observer. In 245 isolation, the information these protocols reveal might seem 246 harmless, but given multiple such protocols, it might be possible to 247 correlate this information. E.g. a protocol that uses frequent 248 messages including a UUID to identify the particular installation 249 does not give the identity of the user away. But a single message 250 including the user's host name might just do that and it can be 251 correlated using e.g. the MAC address of the device's interface. 253 In the experiments described in [TRAC2016], it was possible to 254 correlate frequently sent broadcast messages that included a unique 255 identifier with other broadcast/multicast messages containing 256 usernames (e.g. mDNS, LLMNR or NetBIOS), but also relationships to 257 other users. This allowed to reveal the real identity of the users 258 of many devices but it also gave some information about their social 259 environment away. 261 A broadcast protocol designer should be aware of the fact that even 262 if - in isolation - the information a protocol leaks seems harmless, 263 there might be ways to correlate that information with other 264 broadcast protocol information to reveal sensitive information about 265 a user. 267 2.5. Configurability 269 A lot of applications and services using broadcast/multicast 270 protocols do not include the means to declare "safe" environments 271 (e.g. based on the SSID of a WiFi network and the MAC addresses of 272 the access points). E.g. a device connected to a public WiFi will 273 likely broadcast the same information as when connected to the home 274 network. It would be beneficial if certain behavior could be 275 restricted to "safe" environments. 277 A popular operating system e.g. allows the user to specify the trust 278 level of the network the device connects to, which for example 279 restricts specific system services (using broadcast/multicast 280 messages for their normal operation) to be used in untrusted 281 networks. Such functionality could implemented as part of an 282 application. 284 An application developer making use of broadcasts/multicasts as part 285 of the application SHOULD make the broadcast feature, if possible, 286 configurable, so that potentially sensitive information does not leak 287 on public networks, where the thread to privacy is much larger. 289 3. Operational considerations 291 Besides changing end-user behavior, choosing sensible defaults as an 292 operating system vendor (e.g. for suggesting host names) and the 293 considerations for protocol designers mentioned in this document, 294 there are things that the network administrators/operators can do to 295 limit the above mentioned problems. 297 A feature not uncommonly found on access points e.g. is to filter 298 broadcast and multicast traffic. This will potentially break certain 299 applications or some of their functionality but will also protect the 300 users from potentially leaking sensitive information. 302 4. Summary 304 Increasingly, applications rely on broadcast and multicast messages. 305 For some, broadcasts/multicasts are the basis of their application 306 logic, others use broadcasts/multicasts to improve certain aspects of 307 the application but are fully functional in case broadcasts/ 308 multicasts fail. Irrespective of the role of broadcast and multicast 309 messages for the application, the designers of protocols that make 310 use of them should be very careful in their protocol design because 311 of the special nature of broad- and multicast. 313 It is not always possible to implement certain functionality via 314 unicast, but in case a protocol designer chooses to rely on 315 broadcast/multicast, the following should be carefully considered: 317 o IETF-specified protocols, such as mDNS [RFC6762], should be used 318 if possible as operational support might exist to protect against 319 the leakage of private information. Also, for some protocols 320 privacy extensions are being specified, which can be used if 321 implemented. E.g. for DNS-SD privacy extensions are documented in 322 [I-D.ietf-dnssd-privacy] 324 o Avoid using user-specified information inside broadcast/multicast 325 messages as users will often use personal information or other 326 information aiding attackers, in particular if the user is unaware 327 about how that information is being used 329 o Avoid persistent IDs in messages as this allows user tracking, 330 correlation and potentially has a devastating effect on other 331 privacy protection mechanisms 333 o If you really must use a broadcast/multicast protocol and cannot 334 use an IETF-specified protocol, then: 336 * Be very conservative in how frequently you send messages as an 337 effort in data minimization 339 * Seek advice from IETF-specifies protocols such as message 340 suppression in mDNS 342 * Try to design the protocol in a way that the information cannot 343 be correlated with other information in broadcast/multicast 344 messages 346 * Let the user configure safe environments if possible (e.g. 347 based on the SSID) 349 [Note: discussions on this document should be take place on the 350 Intarea mailing list of the IETF. Subscription: 351 https://www.ietf.org/mailman/listinfo/int-area, Mailing list archive: 352 https://www.ietf.org/mail-archive/web/int-area/current/maillist.html] 354 5. Other considerations 356 Besides the privacy implications of frequent broadcasting, it also 357 represents a performance problem. In particular in certain wireless 358 technologies such as 802.11, broadcast and multicast are transmitted 359 at a much lower rate (the lowest common denominator rate) compared to 360 unicast and therefore have a much bigger impact on the overall 361 available airtime. Further, it will limit the ability for devices to 362 go to sleep if frequent broadcasts are being sent. A similar problem 363 in respect to Router Advertisements is addressed in 364 [I-D.ietf-v6ops-reducing-ra-energy-consumption]. In that respect 365 broadcasts can be used for another class of attacks that not related 366 to privacy. The potential impact on network performance should 367 nevertheless be considered by broadcast protocol designers. 369 6. Acknowledgments 371 We would like to thank Eliot Lear and Stephane Bortzmeyer for their 372 input. 374 This work was partly supported by the European Commission under grant 375 agreement FP7-318627 mPlane. Support does not imply endorsement. 377 7. IANA Considerations 379 This memo includes no request to IANA. 381 8. Security Considerations 383 This document deals with privacy-related considerations of broadcast- 384 and multicast-based protocols. It contains advice for designers of 385 such protocols to minimize the leakage of privacy-sensitive 386 information. The intent of the advice is to make sure that 387 identities will remain anonymous and user tracking will be made 388 difficult. 390 It should be noted that certain applications could make use of 391 existing mechanisms to protect multicast traffic such as the ones 392 defined in [RFC5374]. Examples of such applications can be found in 393 Appendix A. of [RFC5374]. Given the required infrastructure and 394 assumptions about these applications and the security infrastructure, 395 many applications will not be able to make use of such mechanisms. 397 9. Informative References 399 [I-D.huitema-6man-random-addresses] 400 Huitema, C., "Implications of Randomized Link Layers 401 Addresses for IPv6 Address Assignment", draft-huitema- 402 6man-random-addresses-03 (work in progress), March 2016. 404 [I-D.ietf-dnssd-privacy] 405 Huitema, C. and D. Kaiser, "Privacy Extensions for DNS- 406 SD", draft-ietf-dnssd-privacy-00 (work in progress), 407 October 2016. 409 [I-D.ietf-intarea-hostname-practice] 410 Huitema, C. and D. Thaler, "Current Hostname Practice 411 Considered Harmful", draft-ietf-intarea-hostname- 412 practice-00 (work in progress), October 2015. 414 [I-D.ietf-v6ops-reducing-ra-energy-consumption] 415 Yourtchenko, A. and L. Colitti, "Reducing energy 416 consumption of Router Advertisements", draft-ietf-v6ops- 417 reducing-ra-energy-consumption-03 (work in progress), 418 November 2015. 420 [RFC0919] Mogul, J., "Broadcasting Internet Datagrams", STD 5, RFC 421 919, DOI 10.17487/RFC0919, October 1984, 422 . 424 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 425 Requirement Levels", BCP 14, RFC 2119, March 1997. 427 [RFC3819] Karn, P., Ed., Bormann, C., Fairhurst, G., Grossman, D., 428 Ludwig, R., Mahdavi, J., Montenegro, G., Touch, J., and L. 429 Wood, "Advice for Internet Subnetwork Designers", BCP 89, 430 RFC 3819, DOI 10.17487/RFC3819, July 2004, 431 . 433 [RFC4941] Narten, T., Draves, R., and S. Krishnan, "Privacy 434 Extensions for Stateless Address Autoconfiguration in 435 IPv6", RFC 4941, DOI 10.17487/RFC4941, September 2007, 436 . 438 [RFC5374] Weis, B., Gross, G., and D. Ignjatic, "Multicast 439 Extensions to the Security Architecture for the Internet 440 Protocol", RFC 5374, DOI 10.17487/RFC5374, November 2008, 441 . 443 [RFC6762] Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762, 444 DOI 10.17487/RFC6762, February 2013, 445 . 447 [RFC6973] Cooper, A., Tschofenig, H., Aboba, B., Peterson, J., 448 Morris, J., Hansen, M., and R. Smith, "Privacy 449 Considerations for Internet Protocols", RFC 6973, DOI 450 10.17487/RFC6973, July 2013, 451 . 453 [RFC7721] Cooper, A., Gont, F., and D. Thaler, "Security and Privacy 454 Considerations for IPv6 Address Generation Mechanisms", 455 RFC 7721, DOI 10.17487/RFC7721, March 2016, 456 . 458 [RFC7819] Jiang, S., Krishnan, S., and T. Mrugalski, "Privacy 459 Considerations for DHCP", RFC 7819, DOI 10.17487/RFC7819, 460 April 2016, . 462 [TRAC2016] 463 Faath, M., Weisshaar, F., and R. Winter, "How Broadcast 464 Data Reveals Your Identity and Social Graph", 7th 465 International Workshop on TRaffic Analysis and 466 Characterization IEEE TRAC 2016, September 2016. 468 Authors' Addresses 470 Rolf Winter 471 University of Applied Sciences Augsburg 472 Augsburg 473 DE 475 Email: rolf.winter@hs-augsburg.de 477 Michael Faath 478 University of Applied Sciences Augsburg 479 Augsburg 480 DE 482 Email: michael.faath@hs-augsburg.de 484 Fabian Weisshaar 485 University of Applied Sciences Augsburg 486 Augsburg 487 DE 489 Email: fabian.weisshaar@hs-augsburg.de