idnits 2.17.1 draft-ietf-6man-rfc4941bis-12.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 (November 2, 2020) is 1271 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(See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-15) exists of draft-ietf-mboned-ieee802-mcast-problems-12 -- Obsolete informational reference (is this intentional?): RFC 4941 (Obsoleted by RFC 8981) Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IPv6 Maintenance (6man) Working Group F. Gont 3 Internet-Draft SI6 Networks 4 Obsoletes: 4941 (if approved) S. Krishnan 5 Intended status: Standards Track Kaloom 6 Expires: May 6, 2021 T. Narten 8 R. Draves 9 Microsoft Research 10 November 2, 2020 12 Temporary Address Extensions for Stateless Address Autoconfiguration in 13 IPv6 14 draft-ietf-6man-rfc4941bis-12 16 Abstract 18 This document describes an extension to IPv6 Stateless Address 19 Autoconfiguration that causes hosts to generate global scope 20 addresses with randomized interface identifiers that change over 21 time. Changing global scope addresses over time limits the window of 22 time during which eavesdroppers and other information collectors may 23 trivially perform address-based network activity correlation when the 24 same address is employed for multiple transactions by the same host. 25 Additionally, it reduces the window of exposure of a host as being 26 accessible via an address that becomes revealed as a result of active 27 communication. This document obsoletes RFC4941. 29 Status of This Memo 31 This Internet-Draft is submitted in full conformance with the 32 provisions of BCP 78 and BCP 79. 34 Internet-Drafts are working documents of the Internet Engineering 35 Task Force (IETF). Note that other groups may also distribute 36 working documents as Internet-Drafts. The list of current Internet- 37 Drafts is at https://datatracker.ietf.org/drafts/current/. 39 Internet-Drafts are draft documents valid for a maximum of six months 40 and may be updated, replaced, or obsoleted by other documents at any 41 time. It is inappropriate to use Internet-Drafts as reference 42 material or to cite them other than as "work in progress." 44 This Internet-Draft will expire on May 6, 2021. 46 Copyright Notice 48 Copyright (c) 2020 IETF Trust and the persons identified as the 49 document authors. All rights reserved. 51 This document is subject to BCP 78 and the IETF Trust's Legal 52 Provisions Relating to IETF Documents 53 (https://trustee.ietf.org/license-info) in effect on the date of 54 publication of this document. Please review these documents 55 carefully, as they describe your rights and restrictions with respect 56 to this document. Code Components extracted from this document must 57 include Simplified BSD License text as described in Section 4.e of 58 the Trust Legal Provisions and are provided without warranty as 59 described in the Simplified BSD License. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 64 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 65 1.2. Problem Statement . . . . . . . . . . . . . . . . . . . . 4 66 2. Background . . . . . . . . . . . . . . . . . . . . . . . . . 4 67 2.1. Extended Use of the Same Identifier . . . . . . . . . . . 4 68 2.2. Possible Approaches . . . . . . . . . . . . . . . . . . . 6 69 3. Protocol Description . . . . . . . . . . . . . . . . . . . . 6 70 3.1. Design Guidelines . . . . . . . . . . . . . . . . . . . . 7 71 3.2. Assumptions . . . . . . . . . . . . . . . . . . . . . . . 7 72 3.3. Generation of Randomized Interface Identifiers . . . . . 8 73 3.3.1. Simple Randomized Interface Identifiers . . . . . . . 8 74 3.3.2. Hash-based Generation of Randomized Interface 75 Identifiers . . . . . . . . . . . . . . . . . . . . . 9 76 3.4. Generating Temporary Addresses . . . . . . . . . . . . . 11 77 3.5. Expiration of Temporary Addresses . . . . . . . . . . . . 12 78 3.6. Regeneration of Temporary Addresses . . . . . . . . . . . 13 79 3.7. Implementation Considerations . . . . . . . . . . . . . . 14 80 3.8. Defined Constants and Configuration Variables . . . . . . 14 81 4. Implications of Changing Interface Identifiers . . . . . . . 15 82 5. Significant Changes from RFC4941 . . . . . . . . . . . . . . 17 83 6. Future Work . . . . . . . . . . . . . . . . . . . . . . . . . 18 84 7. Implementation Status . . . . . . . . . . . . . . . . . . . . 19 85 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 86 9. Security Considerations . . . . . . . . . . . . . . . . . . . 19 87 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 20 88 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 21 89 11.1. Normative References . . . . . . . . . . . . . . . . . . 21 90 11.2. Informative References . . . . . . . . . . . . . . . . . 22 91 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 24 93 1. Introduction 95 [RFC4862] specifies "Stateless Address Autoconfiguration (SLAAC) for 96 IPv6", which typically results in hosts configuring one or more 97 "stable" IPv6 addresses composed of a network prefix advertised by a 98 local router and a locally-generated Interface Identifier (IID). The 99 security and privacy implications of such addresses have been 100 discussed in detail in [RFC7721], [RFC7217], and [RFC7707]. This 101 document specifies an extension for SLAAC to generate temporary 102 addresses, that can help mitigate some of the aforementioned issues. 103 This is a revision of RFC4941, and formally obsoletes RFC4941. 104 Section 5 describes the changes from [RFC4941]. 106 The default address selection for IPv6 has been specified in 107 [RFC6724]. The determination as to whether to use stable versus 108 temporary addresses can in some cases only be made by an application. 109 For example, some applications may always want to use temporary 110 addresses, while others may want to use them only in some 111 circumstances or not at all. An Application Programming Interface 112 (API) such as that specified in [RFC5014] can enable individual 113 applications to indicate a preference for the use of temporary 114 addresses. 116 Section 2 provides background information. Section 3 describes a 117 procedure for generating temporary addresses. Section 4 discusses 118 implications of changing interface identifiers (IIDs). Section 5 119 describes the changes from [RFC4941]. 121 1.1. Terminology 123 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 124 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 125 "OPTIONAL" in this document are to be interpreted as described in BCP 126 14 [RFC2119] [RFC8174] when, and only when, they appear in all 127 capitals, as shown here. 129 The terms "public address", "stable address", "temporary address", 130 "constant IID", "stable IID", and "temporary IID" are to be 131 interpreted as specified in [RFC7721]. 133 The term "global scope addresses" is used in this document to 134 collectively refer to "Global unicast addresses" as defined in 135 [RFC4291] and "Unique local addresses" as defined in [RFC4193], and 136 not to "globally reachable" addresses, as defined in [RFC8190]. 138 1.2. Problem Statement 140 Addresses generated using stateless address autoconfiguration 141 [RFC4862] contain an embedded interface identifier, which may remain 142 stable over time. Anytime a fixed identifier is used in multiple 143 contexts, it becomes possible to correlate seemingly unrelated 144 activity using this identifier. 146 The correlation can be performed by 148 o An attacker who is in the path between the host in question and 149 the peer(s) to which it is communicating, and who can view the 150 IPv6 addresses present in the datagrams. 152 o An attacker who can access the communication logs of the peers 153 with which the host has communicated. 155 Since the identifier is embedded within the IPv6 address, it cannot 156 be hidden. This document proposes a solution to this issue by 157 generating interface identifiers that vary over time. 159 Note that an attacker, who is on path, may be able to perform 160 significant correlation based on: 162 o The payload contents of unencrypted packets on the wire 164 o The characteristics of the packets such as packet size and timing 166 Use of temporary addresses will not prevent such correlation, nor 167 will it prevent an on-link observer (e.g. the host's default router) 168 from tracking all the host's addresses. 170 2. Background 172 This section discusses the problem in more detail, provides context 173 for evaluating the significance of the concerns in specific 174 environments, and makes comparisons with existing practices. 176 2.1. Extended Use of the Same Identifier 178 The use of a non-changing interface identifier to form addresses is a 179 specific instance of the more general case where a constant 180 identifier is reused over an extended period of time and in multiple 181 independent activities. Any time the same identifier is used in 182 multiple contexts, it becomes possible for that identifier to be used 183 to correlate seemingly unrelated activity. For example, a network 184 sniffer placed strategically on a link across which all traffic to/ 185 from a particular host crosses could keep track of which destinations 186 a host communicated with and at what times. Such information can in 187 some cases be used to infer things, such as what hours an employee 188 was active, when someone is at home, etc. Although it might appear 189 that changing an address regularly in such environments would be 190 desirable to lessen privacy concerns, it should be noted that the 191 network prefix portion of an address also serves as a constant 192 identifier. All hosts at, say, a home, would have the same network 193 prefix, which identifies the topological location of those hosts. 194 This has implications for privacy, though not at the same granularity 195 as the concern that this document addresses. Specifically, all hosts 196 within a home could be grouped together for the purposes of 197 collecting information. If the network contains a very small number 198 of hosts, say, just one, changing just the interface identifier will 199 not enhance privacy, since the prefix serves as a constant 200 identifier. 202 One of the requirements for correlating seemingly unrelated 203 activities is the use (and reuse) of an identifier that is 204 recognizable over time within different contexts. IP addresses 205 provide one obvious example, but there are more. For example, 207 o Many hosts also have DNS names associated with their addresses, in 208 which case the DNS name serves as a similar identifier. Although 209 the DNS name associated with an address is more work to obtain (it 210 may require a DNS query), the information is often readily 211 available. In such cases, changing the address on a host over 212 time would do little to address the concerns raised in this 213 document, unless the DNS name is changed at the same time as well 214 (see Section 4). 216 o Web browsers and servers typically exchange "cookies" with each 217 other [RFC6265]. Cookies allow web servers to correlate a current 218 activity with a previous activity. One common usage is to send 219 back targeted advertising to a user by using the cookie supplied 220 by the browser to identify what earlier queries had been made 221 (e.g., for what type of information). Based on the earlier 222 queries, advertisements can be targeted to match the (assumed) 223 interests of the end-user. 225 The use of a constant identifier within an address is of special 226 concern because addresses are a fundamental requirement of 227 communication and cannot easily be hidden from eavesdroppers and 228 other parties. Even when higher layers encrypt their payloads, 229 addresses in packet headers appear in the clear. Consequently, if a 230 mobile host (e.g., laptop) accessed the network from several 231 different locations, an eavesdropper might be able to track the 232 movement of that mobile host from place to place, even if the upper 233 layer payloads were encrypted. 235 Changing global scope addresses over time limits the time window over 236 which eavesdroppers and other information collectors may trivially 237 correlate network activity when the same address is employed for 238 multiple transactions by the same host. Additionally, it reduces the 239 window of exposure of a host as being accessible via an address that 240 becomes revealed as a result of active communication. 242 The security and privacy implications of IPv6 addresses are discussed 243 in detail in [RFC7721], [RFC7707], and [RFC7217]. 245 2.2. Possible Approaches 247 One approach, compatible with the stateless address autoconfiguration 248 architecture, would be to change the interface identifier portion of 249 an address over time. Changing the interface identifier can make it 250 more difficult to look at the IP addresses in independent 251 transactions and identify which ones actually correspond to the same 252 host, both in the case where the routing prefix portion of an address 253 changes and when it does not. 255 Many hosts function as both clients and servers. In such cases, the 256 host would need a name (e.g. a DNS domain name) for its use as a 257 server. Whether the address stays fixed or changes has little 258 privacy implication since the name remains constant and serves as a 259 constant identifier. When acting as a client (e.g., initiating 260 communication), however, such a host may want to vary the addresses 261 it uses. In such environments, one may need multiple addresses: a 262 stable address associated with the name, that is used to accept 263 incoming connection requests from other hosts, and a temporary 264 address used to shield the identity of the client when it initiates 265 communication. 267 On the other hand, a host that functions only as a client may want to 268 employ only temporary addresses for public communication. 270 To make it difficult to make educated guesses as to whether two 271 different interface identifiers belong to the same host, the 272 algorithm for generating alternate identifiers must include input 273 that has an unpredictable component from the perspective of the 274 outside entities that are collecting information. 276 3. Protocol Description 278 The following subsections define the procedures for the generation of 279 IPv6 temporary addresses. 281 3.1. Design Guidelines 283 Temporary addresses observe the following properties: 285 1. Temporary addresses are typically employed for initiating 286 outgoing sessions. 288 2. Temporary addresses are used for a short period of time 289 (typically hours to days) and are subsequently deprecated. 290 Deprecated addresses can continue to be used for established 291 connections, but are not used to initiate new connections. 293 3. New temporary addresses are generated over time to replace 294 temporary addresses that expire. 296 4. Temporary addresses must have a limited lifetime (limited "valid 297 lifetime" and "preferred lifetime" from [RFC4862]). The lifetime 298 of an address should be further reduced when privacy-meaningful 299 events (such as a host attaching to a different network, or the 300 regeneration of a new randomized MAC address) takes place. The 301 lifetime of temporary addresses must be statistically different 302 for different addresses, such that it is hard to predict or infer 303 when a new temporary address is generated, or correlate a newly- 304 generated address with an existing one. 306 5. By default, one address is generated for each prefix advertised 307 by stateless address autoconfiguration. The resulting Interface 308 Identifiers must be statistically different when addresses are 309 configured for different prefixes or different network 310 interfaces. This means that, given two addresses, it must be 311 difficult for an outside entity to infer whether the addresses 312 correspond to the same host or network interface. 314 6. It must be difficult for an outside entity to predict the 315 Interface Identifiers that will be employed for temporary 316 addresses, even with knowledge of the algorithm/method employed 317 to generate them and/or knowledge of the Interface Identifiers 318 previously employed for other temporary addresses. These 319 Interface Identifiers must be semantically opaque [RFC7136] and 320 must not follow any specific patterns. 322 3.2. Assumptions 324 The following algorithm assumes that for a given temporary address, 325 an implementation can determine the prefix from which it was 326 generated. When a temporary address is deprecated, a new temporary 327 address is generated. The specific valid and preferred lifetimes for 328 the new address are dependent on the corresponding lifetime values 329 set for the prefix from which it was generated. 331 Finally, this document assumes that when a host initiates outgoing 332 communication, temporary addresses can be given preference over 333 stable addresses (if available), when the device is configured to do 334 so. [RFC6724] mandates implementations to provide a mechanism, which 335 allows an application to configure its preference for temporary 336 addresses over stable addresses. It also allows for an 337 implementation to prefer temporary addresses by default, so that the 338 connections initiated by the host can use temporary addresses without 339 requiring application-specific enablement. This document also 340 assumes that an API will exist that allows individual applications to 341 indicate whether they prefer to use temporary or stable addresses and 342 override the system defaults (see e.g. [RFC5014]). 344 3.3. Generation of Randomized Interface Identifiers 346 The following subsections specify example algorithms for generating 347 temporary interface identifiers that follow the guidelines in 348 Section 3.1 of this document. The algorithm specified in 349 Section 3.3.1 benefits from a Pseudo-Random Number Generator (PRNG) 350 available on the system. The algorithm specified in Section 3.3.2 351 allows for code reuse by hosts that implement [RFC7217]. 353 3.3.1. Simple Randomized Interface Identifiers 355 One approach is to select a pseudorandom number of the appropriate 356 length. A host employing this algorithm should generate IIDs as 357 follows: 359 1. Obtain a random number from a pseudo-random number generator 360 (PRNG) that can produce random numbers of at least as many bits 361 as required for the Interface Identifier (please see the next 362 step). [RFC4086] specifies randomness requirements for security. 364 2. The Interface Identifier is obtained by taking as many bits from 365 the random number obtained in the previous step as necessary. 366 See [RFC7136] for the necessary number of bits, that is, the 367 length of the IID. See also [RFC7421] for a discussion of the 368 privacy implications of the IID length. Note: there are no 369 special bits in an Interface Identifier [RFC7136]. 371 3. The resulting Interface Identifier MUST be compared against the 372 reserved IPv6 Interface Identifiers [RFC5453] [IANA-RESERVED-IID] 373 and against those Interface Identifiers already employed in an 374 address of the same network interface and the same network 375 prefix. In the event that an unacceptable identifier has been 376 generated, a new interface identifier should be generated, by 377 repeating the algorithm from the first step. 379 3.3.2. Hash-based Generation of Randomized Interface Identifiers 381 The algorithm in [RFC7217] can be augmented for the generation of 382 temporary addresses. The benefit of this would be that a host could 383 employ a single algorithm for generating stable and temporary 384 addresses, by employing appropriate parameters. 386 Hosts would employ the following algorithm for generating the 387 temporary IID: 389 1. Compute a random identifier with the expression: 391 RID = F(Prefix, Net_Iface, Network_ID, Time, DAD_Counter, 392 secret_key) 394 Where: 396 RID: 397 Random Identifier 399 F(): 400 A pseudorandom function (PRF) that MUST NOT be computable from 401 the outside (without knowledge of the secret key). F() MUST 402 also be difficult to reverse, such that it resists attempts to 403 obtain the secret_key, even when given samples of the output 404 of F() and knowledge or control of the other input parameters. 405 F() SHOULD produce an output of at least as many bits as 406 required for the Interface Identifier. F() could be the 407 result of applying a cryptographic hash over an encoded 408 version of the function parameters. While this document does 409 not recommend a specific mechanism for encoding the function 410 parameters (or a specific cryptographic hash function), a 411 cryptographically robust construction will ensure that the 412 mapping from parameters to the hash function input is an 413 injective map, as might be attained by using fixed-width 414 encodings and/or length-prefixing variable-length parameters. 415 SHA-256 [FIPS-SHS] is one possible option for F(). Note: MD5 416 [RFC1321] is considered unacceptable for F() [RFC6151]. 418 Prefix: 419 The prefix to be used for SLAAC, as learned from an ICMPv6 420 Router Advertisement message. 422 Net_Iface: 424 The MAC address corresponding to the underlying network 425 interface card, in the case the link uses IEEE802 link-layer 426 identifiers. Employing the MAC address for this parameter 427 (over the other suggested options in RFC7217) means that the 428 re-generation of a randomized MAC address will result in a 429 different temporary address. 431 Network_ID: 432 Some network-specific data that identifies the subnet to which 433 this interface is attached -- for example, the IEEE 802.11 434 Service Set Identifier (SSID) corresponding to the network to 435 which this interface is associated. Additionally, "Simple 436 Procedures for Detecting Network Attachment in IPv6" ("Simple 437 DNA") [RFC6059] describes ideas that could be leveraged to 438 generate a Network_ID parameter. This parameter SHOULD be 439 employed if some form of "Network_ID" is available. 441 Time: 442 An implementation-dependent representation of time. One 443 possible example is the representation in UNIX-like systems 444 [OPEN-GROUP], that measure time in terms of the number of 445 seconds elapsed since the Epoch (00:00:00 Coordinated 446 Universal Time (UTC), 1 January 1970). The addition of the 447 "Time" argument results in (statistically) different interface 448 identifiers over time. 450 DAD_Counter: 451 A counter that is employed to resolve Duplicate Address 452 Detection (DAD) conflicts. 454 secret_key: 455 A secret key that is not known by the attacker. The secret 456 key SHOULD be of at least 128 bits. It MUST be initialized to 457 a pseudo-random number (see [RFC4086] for randomness 458 requirements for security) when the operating system is 459 "bootstrapped". The secret_key MUST NOT be employed for any 460 other purpose than the one discussed in this section. For 461 example, implementations MUST NOT employ the same secret_key 462 for the generation of stable addresses [RFC7217] and the 463 generation of temporary addresses via this algorithm. 465 2. The Interface Identifier is finally obtained by taking as many 466 bits from the RID value (computed in the previous step) as 467 necessary, starting from the least significant bit. See 468 [RFC7136] for the necessary number of bits, that is, the length 469 of the IID. See also [RFC7421] for a discussion of the privacy 470 implications of the IID length. Note: there are no special bits 471 in an Interface Identifier [RFC7136]. 473 3. The resulting Interface Identifier MUST be compared against the 474 reserved IPv6 Interface Identifiers [RFC5453] [IANA-RESERVED-IID] 475 and against those Interface Identifiers already employed in an 476 address of the same network interface and the same network 477 prefix. In the event that an unacceptable identifier has been 478 generated, the value DAD_Counter should be incremented by 1, and 479 the algorithm should be restarted from the first step. 481 3.4. Generating Temporary Addresses 483 [RFC4862] describes the steps for generating a link-local address 484 when an interface becomes enabled as well as the steps for generating 485 addresses for other scopes. This document extends [RFC4862] as 486 follows. When processing a Router Advertisement with a Prefix 487 Information option carrying a prefix for the purposes of address 488 autoconfiguration (i.e., the A bit is set), the host MUST perform the 489 following steps: 491 1. Process the Prefix Information Option as defined in [RFC4862], 492 adjusting the lifetimes of existing temporary addresses, with the 493 overall constraint that no temporary addresses should ever remain 494 "valid" or "preferred" for a time longer than 495 (TEMP_VALID_LIFETIME) or (TEMP_PREFERRED_LIFETIME - 496 DESYNC_FACTOR) respectively. The configuration variables 497 TEMP_VALID_LIFETIME and TEMP_PREFERRED_LIFETIME correspond to 498 maximum target lifetimes for temporary addresses. 500 2. One way an implementation can satisfy the above constraints is to 501 associate with each temporary address a creation time (called 502 CREATION_TIME) that indicates the time at which the address was 503 created. When updating the preferred lifetime of an existing 504 temporary address, it would be set to expire at whichever time is 505 earlier: the time indicated by the received lifetime or 506 (CREATION_TIME + TEMP_PREFERRED_LIFETIME - DESYNC_FACTOR). A 507 similar approach can be used with the valid lifetime. 509 3. If the host has not configured any temporary address for the 510 corresponding prefix, the host SHOULD create a new temporary 511 address for such prefix. 513 Note: 514 For example, a host might implement prefix-specific policies 515 such as not configuring temporary addresses for the Unique 516 Local IPv6 Unicast Addresses (ULA) [RFC4193] prefix. 518 4. When creating a temporary address, the DESYNC_FACTOR MUST be 519 computed for this prefix, and the lifetime values MUST be derived 520 from the corresponding prefix as follows: 522 * Its Valid Lifetime is the lower of the Valid Lifetime of the 523 prefix and TEMP_VALID_LIFETIME. 525 * Its Preferred Lifetime is the lower of the Preferred Lifetime 526 of the prefix and TEMP_PREFERRED_LIFETIME - DESYNC_FACTOR. 528 5. A temporary address is created only if this calculated Preferred 529 Lifetime is greater than REGEN_ADVANCE time units. In 530 particular, an implementation MUST NOT create a temporary address 531 with a zero Preferred Lifetime. 533 6. New temporary addresses MUST be created by appending a randomized 534 interface identifier to the prefix that was received. 535 Section 3.3 of this document specifies some sample algorithms for 536 generating the randomized interface identifier. 538 7. The host MUST perform duplicate address detection (DAD) on the 539 generated temporary address. If DAD indicates the address is 540 already in use, the host MUST generate a new randomized interface 541 identifier, and repeat the previous steps as appropriate up to 542 TEMP_IDGEN_RETRIES times. If after TEMP_IDGEN_RETRIES 543 consecutive attempts no non-unique address was generated, the 544 host MUST log a system error and SHOULD NOT attempt to generate a 545 temporary address for the given prefix for the duration of the 546 host's attachment to the network via this interface. This allows 547 hosts to recover from occasional DAD failures, or otherwise log 548 the recurrent address collisions. 550 3.5. Expiration of Temporary Addresses 552 When a temporary address becomes deprecated, a new one MUST be 553 generated. This is done by repeating the actions described in 554 Section 3.4, starting at step 4). Note that, except for the 555 transient period when a temporary address is being regenerated, in 556 normal operation at most one temporary address per prefix should be 557 in a non-deprecated state at any given time on a given interface. 558 Note that if a temporary address becomes deprecated as result of 559 processing a Prefix Information Option with a zero Preferred 560 Lifetime, then a new temporary address MUST NOT be generated. To 561 ensure that a preferred temporary address is always available, a new 562 temporary address SHOULD be regenerated slightly before its 563 predecessor is deprecated. This is to allow sufficient time to avoid 564 race conditions in the case where generating a new temporary address 565 is not instantaneous, such as when duplicate address detection must 566 be run. The host SHOULD start the address regeneration process 567 REGEN_ADVANCE time units before a temporary address would actually be 568 deprecated. 570 As an optional optimization, an implementation MAY remove a 571 deprecated temporary address that is not in use by applications or 572 upper layers as detailed in Section 6. 574 3.6. Regeneration of Temporary Addresses 576 The frequency at which temporary addresses change depends on how a 577 device is being used (e.g., how frequently it initiates new 578 communication) and the concerns of the end user. The most egregious 579 privacy concerns appear to involve addresses used for long periods of 580 time (weeks to months to years). The more frequently an address 581 changes, the less feasible collecting or coordinating information 582 keyed on interface identifiers becomes. Moreover, the cost of 583 collecting information and attempting to correlate it based on 584 interface identifiers will only be justified if enough addresses 585 contain non-changing identifiers to make it worthwhile. Thus, having 586 large numbers of clients change their address on a daily or weekly 587 basis is likely to be sufficient to alleviate most privacy concerns. 589 There are also client costs associated with having a large number of 590 addresses associated with a host (e.g., in doing address lookups, the 591 need to join many multicast groups, etc.). Thus, changing addresses 592 frequently (e.g., every few minutes) may have performance 593 implications. 595 Hosts following this specification SHOULD generate new temporary 596 addresses on a periodic basis. This can be achieved by generating a 597 new temporary address at least once every (TEMP_PREFERRED_LIFETIME - 598 REGEN_ADVANCE - DESYNC_FACTOR) time units. As described above, 599 generating a new temporary address REGEN_ADVANCE time units before a 600 temporary address becomes deprecated produces addresses with a 601 preferred lifetime no larger than TEMP_PREFERRED_LIFETIME. The value 602 DESYNC_FACTOR is a random value computed for a prefix when a 603 temporary address is generated, that ensures that clients do not 604 generate new addresses with a fixed frequency, and that clients do 605 not synchronize with each other and generate new addresses at exactly 606 the same time. When the preferred lifetime expires, a new temporary 607 address MUST be generated using the algorithm specified in 608 Section 3.4. 610 Because the precise frequency at which it is appropriate to generate 611 new addresses varies from one environment to another, implementations 612 SHOULD provide end users with the ability to change the frequency at 613 which addresses are regenerated. The default value is given in 614 TEMP_PREFERRED_LIFETIME and is one day. In addition, the exact time 615 at which to invalidate a temporary address depends on how 616 applications are used by end users. Thus, the suggested default 617 value of two days (TEMP_VALID_LIFETIME) may not be appropriate in all 618 environments. Implementations SHOULD provide end users with the 619 ability to override both of these default values. 621 Finally, when an interface connects to a new (different) link, 622 existing temporary addresses for the corresponding interface MUST be 623 eliminated, and new temporary addresses MUST be generated immediately 624 for use on the new link. If a device moves from one link to another, 625 generating new temporary addresses ensures that the device uses 626 different randomized interface identifiers for the temporary 627 addresses associated with the two links, making it more difficult to 628 correlate addresses from the two different links as being from the 629 same hosts. The host MAY follow any process available to it, to 630 determine that the link change has occurred. One such process is 631 described by Simple DNA [RFC6059]. Detecting link changes would 632 prevent link down/up events from causing temporary addresses to be 633 (unnecessarily) regenerated. 635 3.7. Implementation Considerations 637 Devices implementing this specification MUST provide a way for the 638 end user to explicitly enable or disable the use of temporary 639 addresses. In addition, a site might wish to disable the use of 640 temporary addresses in order to simplify network debugging and 641 operations. Consequently, implementations SHOULD provide a way for 642 trusted system administrators to enable or disable the use of 643 temporary addresses. 645 Additionally, sites might wish to selectively enable or disable the 646 use of temporary addresses for some prefixes. For example, a site 647 might wish to disable temporary address generation for "Unique local" 648 [RFC4193] prefixes while still generating temporary addresses for all 649 other global prefixes. Another site might wish to enable temporary 650 address generation only for the prefixes 2001:db8:1::/48 and 651 2001:db8:2::/48 while disabling it for all other prefixes. To 652 support this behavior, implementations SHOULD provide a way to enable 653 and disable generation of temporary addresses for specific prefix 654 subranges. This per-prefix setting SHOULD override the global 655 settings on the host with respect to the specified prefix subranges. 656 Note that the per-prefix setting can be applied at any granularity, 657 and not necessarily on a per subnet basis. 659 3.8. Defined Constants and Configuration Variables 661 Constants and configuration variables defined in this document 662 include: 664 TEMP_VALID_LIFETIME -- Default value: 2 days. Users should be able 665 to override the default value. 667 TEMP_PREFERRED_LIFETIME -- Default value: 1 day. Users should be 668 able to override the default value. 670 Note: 671 The TEMP_PREFERRED_LIFETIME value MUST be less than the 672 TEMP_VALID_LIFETIME value, to avoid the pathological case where an 673 address is employed for new communications, but becomes invalid in 674 less than 1 second, disrupting those communications 676 REGEN_ADVANCE -- 2 + (TEMP_IDGEN_RETRIES * DupAddrDetectTransmits * 677 RetransTimer / 1000) 679 Notes: 680 This parameter is specified as a function of other protocol 681 parameters, to account for the time possibly spent in Duplicate 682 Address Detection (DAD) in the worst-case scenario of 683 TEMP_IDGEN_RETRIES. This prevents the pathological case where the 684 generation of a new temporary address is not started with enough 685 anticipation such that a new preferred address is generated before 686 the currently-preferred temporary address becomes deprecated. 688 RetransTimer is specified in [RFC4861], while 689 DupAddrDetectTransmits is specified in [RFC4862]. Since 690 RetransTimer is specified in units of milliseconds, this 691 expression employs the constant "1000" such that REGEN_ADVANCE is 692 expressed in seconds. 694 MAX_DESYNC_FACTOR -- 0.4 * TEMP_PREFERRED_LIFETIME. Upper bound on 695 DESYNC_FACTOR. 697 Note: 698 Setting MAX_DESYNC_FACTOR to 0.4 TEMP_PREFERRED_LIFETIME results 699 in addresses that have statistically different lifetimes, and a 700 maximum of 3 concurrent temporary addresses when the default 701 parameters specified in this section are employed. 703 DESYNC_FACTOR -- A random value within the range 0 - 704 MAX_DESYNC_FACTOR. It is computed for a prefix each time a temporary 705 address is generated, and must be smaller than 706 (TEMP_PREFERRED_LIFETIME - REGEN_ADVANCE). 708 TEMP_IDGEN_RETRIES -- Default value: 3 710 4. Implications of Changing Interface Identifiers 712 The desire to protect individual privacy can conflict with the desire 713 % to effectively maintain and debug a network. Having clients use 714 addresses that change over time will make it more difficult to track 715 down and isolate operational problems. For example, when looking at 716 packet traces, it could become more difficult to determine whether 717 one is seeing behavior caused by a single errant host, or by a number 718 of them. 720 Network deployments are currently recommended to provide multiple 721 IPv6 addresses from each prefix to general-purpose hosts [RFC7934]. 722 However, in some scenarios, use of a large number of IPv6 addresses 723 may have negative implications on network devices that need to 724 maintain entries for each IPv6 address in some data structures (e.g., 725 [RFC7039]). For example, concurrent active use of multiple IPv6 726 addresses will increase neighbor discovery traffic if Neighbor Caches 727 in network devices are not large enough to store all addresses on the 728 link. This can impact performance and energy efficiency on networks 729 on which multicast is expensive (e.g. 730 [I-D.ietf-mboned-ieee802-mcast-problems]). Additionally, some 731 network security devices might incorrectly infer IPv6 address forging 732 if temporary addresses are regenerated at a high rate. 734 The use of temporary addresses may cause unexpected difficulties with 735 some applications. For example, some servers refuse to accept 736 communications from clients for which they cannot map the IP address 737 into a DNS name. That is, they perform a DNS PTR query to determine 738 the DNS name, and may then also perform an AAAA query on the returned 739 name to verify that the returned DNS name maps back into the address 740 being used. Consequently, clients not properly registered in the DNS 741 may be unable to access some services. However, a host's DNS name 742 (if non-changing) would serve as a constant identifier. The wide 743 deployment of the extension described in this document could 744 challenge the practice of inverse-DNS-based "validation", which has 745 little validity, though it is widely implemented. In order to meet 746 server challenges, hosts could register temporary addresses in the 747 DNS using random names (for example, a string version of the random 748 address itself), albeit at the expense of increased complexity. 750 In addition, some applications may not behave robustly if temporary 751 addresses are used and an address expires before the application has 752 terminated, or if it opens multiple sessions, but expects them to all 753 use the same addresses. 755 [RFC4941] employed a randomized temporary Interface Identifier for 756 generating a set of temporary addresses, such that temporary 757 addresses configured at a given time for multiple SLAAC prefixes 758 would employ the same Interface Identifier. Sharing the same IID 759 among multiple address allowed host to join only one solicited-node 760 multicast group per temporary address set. 762 This document requires that the Interface Identifiers of all 763 temporary addresses on a host are statistically different from each 764 other. This means that when a network employs multiple prefixes, 765 each temporary address of a set will result in a different solicited- 766 node multicast address, and thus the number of multicast groups that 767 a host must join becomes a function of the number of SLAAC prefixes 768 employed for generating temporary addresses. 770 Thus, a network that employs multiple prefixes may require hosts to 771 join more multicast groups than for an RFC4941 implementation. If 772 the number of multicast groups were large enough, a node might need 773 to resort to setting the network interface card to promiscuous mode. 774 This could cause the node to process more packets than strictly 775 necessary, and might have a negative impact on battery-life, and on 776 system performance in general. 778 We note that since this document reduces the default 779 TEMP_VALID_LIFETIME from 7 days (in [RFC4941]) to 2 days, the number 780 of concurrent temporary addresses per SLAAC prefix will be smaller 781 than for RFC4941 implementations, and thus the number of multicast 782 groups for a network that employs, say, between 1 and three prefixes 783 will be similar than of RFC4941 implementations. 785 Implementations concerned with the maximum number of multicast groups 786 that would be required to join as a result of configured addresses, 787 or the overall number of configured addresses, should consider 788 enforcing implementation-specific limits on e.g. the maximum number 789 of configured addresses, the maximum number of SLAAC prefixes that 790 are employed for auto-configuration, and/or the maximum ratio for 791 TEMP_VALID_LIFETIME/TEMP_PREFERRED_LIFETIME (that ultimately controls 792 the approximate number of concurrent temporary addresses per SLAAC 793 prefix). Many of these configuration limits are readily available in 794 SLAAC and RFC4941 implementations. We note that these configurable 795 limits are meant to prevent pathological behaviors (as opposed to 796 simply limiting the usage of IPv6 addresses), since IPv6 797 implementations are expected to leverage the usage of multiple 798 addresses [RFC7934]. 800 5. Significant Changes from RFC4941 802 This section summarizes the substantive changes in this document 803 relative to RFC 4941. 805 Broadly speaking, this document introduces the following changes: 807 o Addresses a number of flaws in the algorithm for generating 808 temporary addresses: The aforementioned flaws include the use of 809 MD5 for computing the temporary IIDs, and reusing the same IID for 810 multiple prefixes (see [RAID2015] and [RFC7721] for further 811 details). 813 o Allows hosts to employ only temporary addresses: 814 [RFC4941] assumed that temporary addresses were configured in 815 addition to stable addresses. This document does not imply or 816 require the configuration of stable addresses, and thus 817 implementations can now configure both stable and temporary 818 addresses, or temporary addresses only. 820 o Removes the recommendation that temporary addresses be disabled by 821 default: 822 This is in line with BCP188 ([RFC7258]), and also with BCP204 823 ([RFC7934]). 825 o Reduces the default maximum Valid Lifetime for temporary 826 addresses: The default Valid Lifetime for temporary addresses has 827 been reduced from 1 week to 2 days, decreasing the typical number 828 of concurrent temporary addresses from 7 to 3. This reduces the 829 possible stress on network elements (see Section 4 for further 830 details). 832 o DESYNC_FACTOR is computed on a per-prefix basis each time a 833 temporary address is generated, such that each temporary address 834 has a statistically different preferred lifetime, and that 835 temporary addresses are not generated at a constant frequency. 837 o Changes the requirement to not try to regenerate temporary 838 addresses upon DAD failures from "MUST NOT" to "SHOULD NOT". 840 o The discussion about the security and privacy implications of 841 different address generation techniques has been replaced with 842 references to recent work in this area ([RFC7707], [RFC7721], and 843 [RFC7217]). 845 o Addresses all errata submitted for [RFC4941]. 847 6. Future Work 849 An implementation might want to keep track of which addresses are 850 being used by upper layers so as to be able to remove a deprecated 851 temporary address from internal data structures once no upper layer 852 protocols are using it (but not before). This is in contrast to 853 current approaches where addresses are removed from an interface when 854 they become invalid [RFC4862], independent of whether or not upper 855 layer protocols are still using them. For TCP connections, such 856 information is available in control blocks. For UDP-based 857 applications, it may be the case that only the applications have 858 knowledge about what addresses are actually in use. Consequently, an 859 implementation generally will need to use heuristics in deciding when 860 an address is no longer in use. 862 7. Implementation Status 864 [The RFC-Editor should remove this section before publishing this 865 document as an RFC] 867 The following are known implementations of this document: 869 o FreeBSD kernel: There is a FreeBSD kernel implementation of this 870 document, albeit not yet committed. The implementation has been 871 done in April 2020 by Fernando Gont . The 872 corresponding patch can be found at: 873 875 o Linux kernel: A Linux kernel implementation of this document has 876 been committed to the net-next tree. The implementation has been 877 produced in April 2020 by Fernando Gont . 878 The corresponding patch can be found at: 879 882 o slaacd(8): slaacd(8) has traditionally used different randomized 883 interface identifiers for each prefix, and it has recently reduced 884 the Valid Lifetime of temporary addresses as specified in 885 Section 3.8, thus fully implementing this document. The 886 implementation has been done by Florian Obser 887 , with the update to the temporary address 888 Valid Lifetime applied in March 2020. The implementation can be 889 found at: 891 8. IANA Considerations 893 There are no IANA registries within this document. The RFC-Editor 894 can remove this section before publication of this document as an 895 RFC. 897 9. Security Considerations 899 If a very small number of hosts (say, only one) use a given prefix 900 for extended periods of time, just changing the interface identifier 901 part of the address may not be sufficient to mitigate address-based 902 network activity correlation, since the prefix acts as a constant 903 identifier. The procedures described in this document are most 904 effective when the prefix is reasonably non static or is used by a 905 fairly large number of hosts. Additionally, if a temporary address 906 is used in a session where the user authenticates, any notion of 907 "privacy" for that address is compromised for the part(ies) that 908 receive the authentication information. 910 While this document discusses ways to limit the lifetime of Interface 911 Identifiers to reduce the ability of attackers to perform address- 912 based network activity correlation, the method described is believed 913 to be ineffective against sophisticated forms of traffic analysis. 914 To increase effectiveness, one may need to consider the use of more 915 advanced techniques, such as Onion Routing [ONION]. 917 Ingress filtering has been and is being deployed as a means of 918 preventing the use of spoofed source addresses in Distributed Denial 919 of Service (DDoS) attacks. In a network with a large number of 920 hosts, new temporary addresses are created at a fairly high rate. 921 This might make it difficult for ingress filtering mechanisms to 922 distinguish between legitimately changing temporary addresses and 923 spoofed source addresses, which are "in-prefix" (using a 924 topologically correct prefix and non-existent interface ID). This 925 can be addressed by using access control mechanisms on a per-address 926 basis on the network egress point, though as noted in Section 4 there 927 are corresponding costs for doing so. 929 10. Acknowledgments 931 The authors would like to thank (in alphabetical order) Fred Baker, 932 Brian Carpenter, Tim Chown, Lorenzo Colitti, Roman Danyliw, David 933 Farmer, Tom Herbert, Bob Hinden, Christian Huitema, Benjamin Kaduk, 934 Erik Kline, Gyan Mishra, Dave Plonka, Alvaro Retana, Michael 935 Richardson, Mark Smith, Dave Thaler, Pascal Thubert, Ole Troan, 936 Johanna Ullrich, Eric Vyncke, and Timothy Winters, for providing 937 valuable comments on earlier versions of this document. 939 This document incorporates errata submitted for [RFC4941] by Jiri 940 Bohac and Alfred Hoenes. 942 This document is based on [RFC4941] (a revision of RFC3041). Suresh 943 Krishnan was the sole author of RFC4941. He would like to 944 acknowledge the contributions of the IPv6 working group and, in 945 particular, Jari Arkko, Pekka Nikander, Pekka Savola, Francis Dupont, 946 Brian Haberman, Tatuya Jinmei, and Margaret Wasserman for their 947 detailed comments. 949 Rich Draves and Thomas Narten were the authors of RFC 3041. They 950 would like to acknowledge the contributions of the IPv6 working group 951 and, in particular, Ran Atkinson, Matt Crawford, Steve Deering, 952 Allison Mankin, and Peter Bieringer. 954 11. References 956 11.1. Normative References 958 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 959 Requirement Levels", BCP 14, RFC 2119, 960 DOI 10.17487/RFC2119, March 1997, 961 . 963 [RFC4086] Eastlake 3rd, D., Schiller, J., and S. Crocker, 964 "Randomness Requirements for Security", BCP 106, RFC 4086, 965 DOI 10.17487/RFC4086, June 2005, 966 . 968 [RFC4193] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 969 Addresses", RFC 4193, DOI 10.17487/RFC4193, October 2005, 970 . 972 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 973 Architecture", RFC 4291, DOI 10.17487/RFC4291, February 974 2006, . 976 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 977 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 978 DOI 10.17487/RFC4861, September 2007, 979 . 981 [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless 982 Address Autoconfiguration", RFC 4862, 983 DOI 10.17487/RFC4862, September 2007, 984 . 986 [RFC5453] Krishnan, S., "Reserved IPv6 Interface Identifiers", 987 RFC 5453, DOI 10.17487/RFC5453, February 2009, 988 . 990 [RFC6724] Thaler, D., Ed., Draves, R., Matsumoto, A., and T. Chown, 991 "Default Address Selection for Internet Protocol Version 6 992 (IPv6)", RFC 6724, DOI 10.17487/RFC6724, September 2012, 993 . 995 [RFC7136] Carpenter, B. and S. Jiang, "Significance of IPv6 996 Interface Identifiers", RFC 7136, DOI 10.17487/RFC7136, 997 February 2014, . 999 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1000 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1001 May 2017, . 1003 11.2. Informative References 1005 [FIPS-SHS] 1006 NIST, "Secure Hash Standard (SHS)", FIPS 1007 Publication 180-4, August 2015, 1008 . 1011 [I-D.ietf-mboned-ieee802-mcast-problems] 1012 Perkins, C., McBride, M., Stanley, D., Kumari, W., and J. 1013 Zuniga, "Multicast Considerations over IEEE 802 Wireless 1014 Media", draft-ietf-mboned-ieee802-mcast-problems-12 (work 1015 in progress), October 2020. 1017 [IANA-RESERVED-IID] 1018 IANA, "Reserved IPv6 Interface Identifiers", 1019 . 1021 [ONION] Reed, MGR., Syverson, PFS., and DMG. Goldschlag, "Proxies 1022 for Anonymous Routing", Proceedings of the 12th Annual 1023 Computer Security Applications Conference, San Diego, CA, 1024 December 1996. 1026 [OPEN-GROUP] 1027 The Open Group, "The Open Group Base Specifications Issue 1028 7 / IEEE Std 1003.1-2008, 2016 Edition", 1029 Section 4.16 Seconds Since the Epoch, 2016, 1030 . 1033 [RAID2015] 1034 Ullrich, J. and E. Weippl, "Privacy is Not an Option: 1035 Attacking the IPv6 Privacy Extension", International 1036 Symposium on Recent Advances in Intrusion Detection 1037 (RAID), 2015, . 1040 [RFC1321] Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321, 1041 DOI 10.17487/RFC1321, April 1992, 1042 . 1044 [RFC4941] Narten, T., Draves, R., and S. Krishnan, "Privacy 1045 Extensions for Stateless Address Autoconfiguration in 1046 IPv6", RFC 4941, DOI 10.17487/RFC4941, September 2007, 1047 . 1049 [RFC5014] Nordmark, E., Chakrabarti, S., and J. Laganier, "IPv6 1050 Socket API for Source Address Selection", RFC 5014, 1051 DOI 10.17487/RFC5014, September 2007, 1052 . 1054 [RFC6059] Krishnan, S. and G. Daley, "Simple Procedures for 1055 Detecting Network Attachment in IPv6", RFC 6059, 1056 DOI 10.17487/RFC6059, November 2010, 1057 . 1059 [RFC6151] Turner, S. and L. Chen, "Updated Security Considerations 1060 for the MD5 Message-Digest and the HMAC-MD5 Algorithms", 1061 RFC 6151, DOI 10.17487/RFC6151, March 2011, 1062 . 1064 [RFC6265] Barth, A., "HTTP State Management Mechanism", RFC 6265, 1065 DOI 10.17487/RFC6265, April 2011, 1066 . 1068 [RFC7039] Wu, J., Bi, J., Bagnulo, M., Baker, F., and C. Vogt, Ed., 1069 "Source Address Validation Improvement (SAVI) Framework", 1070 RFC 7039, DOI 10.17487/RFC7039, October 2013, 1071 . 1073 [RFC7217] Gont, F., "A Method for Generating Semantically Opaque 1074 Interface Identifiers with IPv6 Stateless Address 1075 Autoconfiguration (SLAAC)", RFC 7217, 1076 DOI 10.17487/RFC7217, April 2014, 1077 . 1079 [RFC7258] Farrell, S. and H. Tschofenig, "Pervasive Monitoring Is an 1080 Attack", BCP 188, RFC 7258, DOI 10.17487/RFC7258, May 1081 2014, . 1083 [RFC7421] Carpenter, B., Ed., Chown, T., Gont, F., Jiang, S., 1084 Petrescu, A., and A. Yourtchenko, "Analysis of the 64-bit 1085 Boundary in IPv6 Addressing", RFC 7421, 1086 DOI 10.17487/RFC7421, January 2015, 1087 . 1089 [RFC7707] Gont, F. and T. Chown, "Network Reconnaissance in IPv6 1090 Networks", RFC 7707, DOI 10.17487/RFC7707, March 2016, 1091 . 1093 [RFC7721] Cooper, A., Gont, F., and D. Thaler, "Security and Privacy 1094 Considerations for IPv6 Address Generation Mechanisms", 1095 RFC 7721, DOI 10.17487/RFC7721, March 2016, 1096 . 1098 [RFC7934] Colitti, L., Cerf, V., Cheshire, S., and D. Schinazi, 1099 "Host Address Availability Recommendations", BCP 204, 1100 RFC 7934, DOI 10.17487/RFC7934, July 2016, 1101 . 1103 [RFC8190] Bonica, R., Cotton, M., Haberman, B., and L. Vegoda, 1104 "Updates to the Special-Purpose IP Address Registries", 1105 BCP 153, RFC 8190, DOI 10.17487/RFC8190, June 2017, 1106 . 1108 Authors' Addresses 1110 Fernando Gont 1111 SI6 Networks 1112 Evaristo Carriego 2644 1113 Haedo, Provincia de Buenos Aires 1706 1114 Argentina 1116 Phone: +54 11 4650 8472 1117 Email: fgont@si6networks.com 1118 URI: https://www.si6networks.com 1120 Suresh Krishnan 1121 Kaloom 1123 Email: suresh@kaloom.com 1125 Thomas Narten 1127 Email: narten@cs.duke.edu 1129 Richard Draves 1130 Microsoft Research 1131 One Microsoft Way 1132 Redmond, WA 1133 USA 1135 Email: richdr@microsoft.com