idnits 2.17.1 draft-ietf-mptcp-threat-02.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** You're using the IETF Trust Provisions' Section 6.b License Notice from 12 Sep 2009 rather than the newer Notice from 28 Dec 2009. (See https://trustee.ietf.org/license-info/) 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 document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (March 29, 2010) is 5114 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- No issues found here. Summary: 2 errors (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Bagnulo 3 Internet-Draft UC3M 4 Intended status: Informational March 29, 2010 5 Expires: September 30, 2010 7 Threat Analysis for Multi-addressed/Multi-path TCP 8 draft-ietf-mptcp-threat-02 10 Abstract 12 Multi-addresses/Multi-path TCP (MPTCP for short) describes the 13 extensions proposed for TCP so that each endpoint of a given TCP 14 connection can use multiple IP addresses to exchange data (instead of 15 a single IP address per endpoint as currently defined). Such 16 extensions enable the exchange of segments using different source- 17 destination address pairs, resulting in the capability of using 18 multiple paths in a significant number of scenarios. In particular, 19 some level of multihoming and mobility support can be achieved 20 through these extensions. However, the support for multiple IP 21 addresses per endpoint may have implications on the security of the 22 resulting MPTCP protocol. This note includes a threat analysis for 23 MPTCP. 25 Status of this Memo 27 This Internet-Draft is submitted to IETF in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF), its areas, and its working groups. Note that 32 other groups may also distribute working documents as Internet- 33 Drafts. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 The list of current Internet-Drafts can be accessed at 41 http://www.ietf.org/ietf/1id-abstracts.txt. 43 The list of Internet-Draft Shadow Directories can be accessed at 44 http://www.ietf.org/shadow.html. 46 This Internet-Draft will expire on September 30, 2010. 48 Copyright Notice 49 Copyright (c) 2010 IETF Trust and the persons identified as the 50 document authors. All rights reserved. 52 This document is subject to BCP 78 and the IETF Trust's Legal 53 Provisions Relating to IETF Documents 54 (http://trustee.ietf.org/license-info) in effect on the date of 55 publication of this document. Please review these documents 56 carefully, as they describe your rights and restrictions with respect 57 to this document. Code Components extracted from this document must 58 include Simplified BSD License text as described in Section 4.e of 59 the Trust Legal Provisions and are provided without warranty as 60 described in the BSD License. 62 Table of Contents 64 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 65 2. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 66 3. Related work . . . . . . . . . . . . . . . . . . . . . . . . . 4 67 4. Basic MPTCP. . . . . . . . . . . . . . . . . . . . . . . . . . 6 68 5. Flooding attacks . . . . . . . . . . . . . . . . . . . . . . . 7 69 6. Hijacking attacks . . . . . . . . . . . . . . . . . . . . . . 9 70 6.1. Hijacking attacks to the Basic MPTCP protocol . . . . . . 9 71 6.2. Time-shifted hijacking attacks . . . . . . . . . . . . . . 12 72 6.3. NAT considerations . . . . . . . . . . . . . . . . . . . . 13 73 7. Reccomendation . . . . . . . . . . . . . . . . . . . . . . . . 14 74 8. Security Considerations . . . . . . . . . . . . . . . . . . . 14 75 9. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 14 76 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 77 11. Informative References . . . . . . . . . . . . . . . . . . . . 15 78 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 15 80 1. Introduction 82 Multi-addresses/Multi-path TCP (MPTCP for short) describes the 83 extensions proposed for TCP so that each endpoint of a given TCP 84 connection can use multiple IP addresses to exchange data (instead of 85 a single IP address per endpoint as currently defined). Such 86 extensions enable the exchange of segments using different source- 87 destination address pairs, resulting in the capability of using 88 multiple paths in a significant number of scenarios. In particular, 89 some level of multihoming and mobility support can be achieved 90 through these extensions. However, the support for multiple IP 91 addresses per endpoint may have implications on the security of the 92 resulting MPTCP protocol. This note includes a threat analysis for 93 MPTCP. 95 2. Scope 97 There are multiple ways to achieve Multi-path TCP. Essentially what 98 is needed is for different segments of the communication to be 99 forwarded through different paths by enabling the sender to specify 100 some form of path selector. There are multiple options for such path 101 selector, including the usage of different next hops, using tunnels 102 to different egress points and so on. In this note, we will focus on 103 a particular approach, namely MPTCP approaches that rely on the usage 104 of multiple IP address per endpoint and that use different source- 105 destination address pairs as a mean to express different paths. So, 106 in the rest of this note, the MPTCP expression will refer to this 107 Multi-addressed flavour of Multi-path TCP. 109 Scope of the analysis 111 In this note we perform a threat analysis for MPTCP. Introducing the 112 support of multiple addresses per endpoint in a single TCP connection 113 may result in additional vulnerabilities. The scope of this note is 114 to identify and characterize these new vulnerabilities. So, the 115 scope of the analysis is limited to the additional vulnerabilities 116 resulting from the multi-address support compared to the current TCP 117 protocol (where each endpoint only has one address available for use 118 per connection). In other words, a full analysis of the complete set 119 of threats is explicitly out of the scope. The goal of this analysis 120 is to help the MPTCP protocol designers to create a MPTCP that is as 121 secure as the current TCP. It is a non goal of this analysis to help 122 in the design of MPTCP that is more secure than regular TCP. 124 In particular, we will focus on attackers that are not along the 125 path, at least not during the whole duration of the connection. In 126 the current single path TCP, on-path attacker can launch a 127 significant number of attacks, including eavesdropping, connection 128 hijacking Man in the Middle attacks and so on. However, it is not 129 possible for the off-path attackers to launch such attacks. There is 130 a middle ground in case the attacker is located along the path for a 131 short period of time to launch the attack and then moves away, but 132 the attack effects still apply. These are the so-called time-shifted 133 attacks. Since these are not possible in today's TCP, we will also 134 consider them as part of the analysis. So, summarizing, we will 135 consider both attacks launched by off-path attackers and time-shifted 136 attacks. Attacks launched by on-path attackers are out of scope, 137 since they also apply to current single-path TCP. 138 It should be noted, however, that some current on-path attacks may 139 become more difficult with multi-path TCP, since an attacker (on a 140 single path) will not have visibility of the complete data stream. 142 3. Related work 144 There is significant amount of previous work in terms of analysis of 145 protocols that support address agility. In this section we present 146 the most relevant ones and we relate them to the current MPTCP 147 effort. 149 Most of the problems related to address agility have been deeply 150 analyzed and understood in the context of Route Optimization support 151 in Mobile IPv6 (MIPv6 RO). [RFC4225] includes the rational for the 152 design of the security of MIPv6 RO. All the attacks described in the 153 aforementioned analysis apply here and are an excellent basis for our 154 own analysis. The main differences are: 155 In MIPv6 RO, the address binding affects all the communications 156 involving an address, while in the MPTCP case, a single connection 157 is at stake. In other words, if a binding between two address is 158 created at the IP layer, this binding can and will affect all the 159 connections that involve those addresses. However, in MPTCP, if 160 an additional address is added to an ongoing TCP connection, the 161 additional address will/can only affect the connection at hand and 162 not other connections even if the same address is being used for 163 those other connections. The result is that in MPTCP there is 164 much less at stake and the resulting vulnerabilities are less. On 165 the other hand, it is very important to keep the assumption valid 166 that the address bindings for a given connection do not affect 167 other connections. If reusing of binding or security information 168 is to be considered, this assumption could be no longer valid and 169 the full impact of the vulnerabilities must be assessed. 170 In MIPv6 RO, there is the assumption that the original path 171 through which the connection has been established is always 172 available and in case it is not, the communication will be lost. 173 In MPTCP, it is an explicit goal to provide communication 174 resilience when one of the address pairs is no longer usable, so 175 it is not possible to leverage on the original address pair to be 176 always working. 177 MIPv6 RO is of course designed for IPv6 and it is an explicit goal 178 of MPTCP to support both IPv6 and IPv4. Some MIPv6 RO security 179 solutions rely on the usage of some characteristics of IPv6 (such 180 as the usage of CGAs [RFC3972]), which will no be usable in the 181 context of MPTCP. 183 In the Shim6 design, similar issues related to address agility were 184 considered and a threat analysis was also performed [RFC4218]. The 185 analysis performed for Shim6 also largely applies to the MPTCP 186 context, the main difference being: 187 Similarly to the MPTCP case, the Shim6 protocol is a layer 3 188 protocol so all the communications involving the target address 189 are at stake, as opposed to the MPTCP case, where the impact can 190 be limited to a single TCP connection. 191 Similarly to MIPv6 RO, Shim6 only uses IPv6 addresses as 192 identifiers and leverages on some of their properties to provide 193 the security, such as relying on CGAs or HBAs [RFC5535], which is 194 not possible in the MPTCP case where IPv4 addresses must be 195 supported. 197 SCTP is a transport protocol that supports multiple addresses per 198 endpoint and as such, the security implications are very close to the 199 ones of MPTCP. A security analysis, identifying a set of attacks and 200 proposed solutions was performed in [RFC5062]. The results of this 201 analysis apply directly to the case of MPTCP. However, the analysis 202 was performed after the base SCTP protocol was designed and the goal 203 of the document was essentially to improve the security of SCTP. As 204 such, the document is very specific to the actual SCTP specification 205 and relies on the SCTP messages and behaviour to characterize the 206 issues. While some them can be translated to the MPTCP case, some 207 may be caused by specific behaviour of SCTP as defined. In 208 particular, one issue that is different in the MPTCP case compared to 209 the SCTP case is that in MPTCP it is fundamental that multiple paths 210 are used simultaneously, which does have security implications. 212 So, the conclusion is that while we do have a significant amount of 213 previous work that is closely related and we can and will use it as a 214 basis for this analysis, there are a set of characteristics that are 215 specific to MPTCP that grant the need for a specific analysis for 216 MPTCP. The goal of this analysis is to help MPTCP protocol designers 217 to include a set of security mechanisms that prevent the introduction 218 of new vulnerabilities to the Internet due to the adoption of MPTCP. 220 4. Basic MPTCP. 222 As we stated earlier, the goal of this document is to serve as input 223 for MPTCP protocol designers to properly take into account the 224 security issues. As such, the analysis cannot be performed for a 225 specific MPTCP specification, but must be a general analysis that 226 applies to the widest possible set of MPTCP designs. In order to do 227 that, we will characterize what are the fundamental features that any 228 MPTCP protocol must provide and attempt to perform the security 229 implications only assuming those. In some cases, we will have a 230 design choice that will significantly influence the security aspects 231 of the resulting protocol. In that case we will consider both 232 options and try to characterize both designs. 234 We assume that any MPTCP will behave in the case of a single address 235 per endpoint as TCP. This means that a MPTCP connection will be 236 established by using the TCP 3-way handshake and will use a single 237 address pair. 239 The addresses used for the establishment of the connection do have a 240 special role in the sense that this is the address used as identifier 241 by the upper layers. In particular, the address used as destination 242 address in the SYN packet is the address that the application is 243 using to identify the peer and has been obtained either through the 244 DNS (with or without DNSSEC validation) or passed by a referral or 245 manually introduced by the user. As such, the initiator does have a 246 certain amount of trust in the fact that it is establishing a 247 communication with that particular address. If due to MPTCP, packets 248 end up being delivered to an alternative address, the trust that the 249 initiator has placed on that address would be deceived. In any case, 250 the adoption of MPTCP necessitates a slight evolution of the 251 traditional TCP trust model, in that the initiator is additionally 252 trusting the peer to provide additional addresses which it will trust 253 to the same degree as the original pair. An application or 254 implementation that cannot trust the peer in this way should not make 255 use of multiple paths. 257 During the 3-way handshake, the sequence number will be synchronized 258 for both ends, as in regular TCP. We assume that a MPTCP connection 259 will use a sequence number for the data, even if the data is 260 exchanged through different paths. 262 Once the connection is established, the MPTCP extensions can be used 263 to add addresses for each of the endpoints. In order to do that each 264 end will need to send a control message containing the additional 265 address(es). In order to associate the additional address to an 266 ongoing connection, the connection needs to be identified. We assume 267 that the connection can be identified by the 4-tuple of source 268 address, source port, destination address, destination port used for 269 the establishment of the connection. So, at least, the control 270 message that will convey the additional address information can also 271 contain the 4-tuple in order to inform about what connection the 272 address belong to (if no other connection identifier is defined). 273 There are two different ways to convey address information: 274 o Explicit mode: the control message contain a list of addresses. 275 o Implicit mode: the address added is the one included in the source 276 address field of the IP header 278 These two modes have significantly different security properties. 279 The explicit mode seems to be the more vulnerable to abuse. In 280 particular, the implicit mode may benefit from forms of ingress 281 filtering security, which would reduce the possibility of an attacker 282 to add any arbitrary address to an ongoing connection. 284 In addition, we will assume that MPTCP will use all the address pairs 285 that it has available for sending packets and that it will distribute 286 the load based on congestion among the different paths. 288 5. Flooding attacks 290 The first type of attacks that are introduced by address agility are 291 the so called flooding (or bombing) attacks. The setup for this 292 attack is depicted in the following figure: 294 +--------+ (step 1) +------+ 295 |Attacker| ------------------------- |Source| 296 | A |IPA IPS| S | 297 +--------+ /+------+ 298 / 299 (step 2) / 300 / 301 v IPT 302 +------+ 303 |Target| 304 | T | 305 +------+ 307 The scenario consists of an attacker A who has an IP address IPA. A 308 server that can generate a significant amount of traffic (such as a 309 streaming server), called source S and that has IP address IPS. In 310 addition, we have the target of the flooding attack, target T which 311 has an IP address IPT. 313 In the first step of this attack (depicted as step 1 in the figure), 314 the attacker A establishes a MPTCP connection with the source of the 315 traffic server S and starts downloading a significant amount of 316 traffic. The initial connection only involves one IP address per 317 endpoint, namely IPA and IPS. Once that the download is on course, 318 the second step of the attack (depicted as step 2 in the figure) is 319 that the attacker A adds IPT as one of the available addresses for 320 the communication. How the additional address is added depends on 321 the MPTCP address management mode. In explicit address management, 322 the attacker A only needs to send a signaling packet conveying 323 address IPT. In implicit mode, the attacker A would need to send a 324 packet with IPT as the source address. Depending on whether ingress 325 filtering is deployed and the location of the attacker, it may be 326 possible or not for the attacker to send such packet. At this stage, 327 the MPTCP connection still has a single address for the Source S i.e. 328 IPS but has two addresses for the Attacker A, namely IPA and IPT. 329 The attacker now attempts to get the Source S to send the traffic of 330 the ongoing download to the Target T IP address i.e. IPT. The 331 attacker can do that by pretending that the path between IPA and IPT 332 is congested but that the path between IPS and IPT is not. In order 333 to do that, it needs to send ACKs for the data that flows through the 334 path between IPS and IPT and do not send ACKs for the data that is 335 sent to IPA. The actual details of this will depend on how the data 336 sent through the different paths is ACKed. One possibility is that 337 ACKs for the data sent using a given a given address pair should come 338 in packets containing the same address pair. If so, the attacker 339 would need to send ACKs using packets containing IPT as the source 340 address to keep the attack flowing. This may be possible or not 341 depending on the deployment of ingress filtering and the location of 342 the attacker. The attacker would also need to guess the sequence 343 number of the data being sent to the Target. Once the attacker 344 manages to perform these actions the attack is on place and the 345 download will hit the target. It should be noted that in this type 346 of attacks, the Source S still thinks it is sending packets to the 347 Attacker A while in reality it is sending the packet to Target T. 349 Once that the traffic from the Source S start hitting the Target T, 350 the target will react. In particular, since the packets are likely 351 to belong to a non existent TCP connection, the Target T will issue 352 RST packets. It is relevant then to understand how MPTCP reacts to 353 incoming RST packets. It seems that the at least the MPTCP that 354 receives a RST packet should terminate the packet exchange 355 corresponding to the particular address pair (maybe not the complete 356 MPTCP connection, but at least it should not send more packets with 357 the address pair involved in the RST packet). However, if the 358 attacker, before redirecting the traffic has managed to increase the 359 window size considerably, the flight size could be enough to impose a 360 significant amount of traffic to the Target node. There is a subtle 361 operation that the attacker needs to achieve in order to launch a 362 significant attack. On the one hand it needs to grow the window 363 enough so that the flight size is big enough to cause enough effect 364 and on the other hand the attacker needs to be able to simulate 365 congestion on the IPA-IPS path so that traffic is actually redirected 366 to the alternative path without significantly reducing the window. 367 This will heavily depend on how the coupling of the windows between 368 the different paths works, in particular how the windows are 369 increased. Some designs of the congestion control window coupling 370 could render this attack ineffective. 372 Previous protocols that have to deal with this type of attacks have 373 done so by adding a reachability check before actually sending data 374 to a new address. In other words, the solution used in other 375 protocols such as MIPv6 RO, would include the Source S to explicitly 376 asking the host sitting in the new address (in this case the Target T 377 sitting in IPT) whether it is willing to accept packets from the 378 MPTCP connection identified by the 4-tuple IPA, port A, IPS, port S. 379 Since this is not part of the established connection that Target T 380 has, T would not accept the request and Source S would not use IPT to 381 send packets for this MPTCP connection. Usually, the request also 382 includes a nonce that cannot be guessed by the attacker A so that it 383 cannot fake the reply to the request easily. 385 One possible approach to do this reachability test would be to 386 perform a 3-way handshake for each new address pair that is going to 387 be used in a MPTCP connection. While there are other reasons for 388 doing this (such as NAT traversal), such approach would also act as a 389 reachability test and would prevent the flooding attacks described in 390 this section. 392 6. Hijacking attacks 394 6.1. Hijacking attacks to the Basic MPTCP protocol 396 The hijacking attacks essentially use the MPTCP address agility to 397 allow an attacker to hijack a connection. This means that the victim 398 of a connection thinks that it is talking to a peer, while it is 399 actually exchanging packets with the attacker. In some sense it is 400 the dual of the flooding attacks (where the victim thinks it is 401 exchanging packets with the attacker but in reality is sending the 402 packets to the target). 404 The scenario for a hijacking attack is described in the next figure. 406 +------+ +------+ 407 | Node | ------------------------- | Node | 408 | 1 |IP1 IP2| 2 | 409 +------+ /+------+ 410 / 411 / 412 / 413 v IPA 414 +--------+ 415 |Attacker| 416 | A | 417 +--------+ 419 In this case, we have a MPTCP connection established between Node 1 420 and Node 2. The connection is using only one address per endpoint, 421 namely IP1 and IP2. The attacker then launches the hijacking attack 422 by adding IPA as an additional address for Node 1. In this case, 423 there is not much difference between explicit or implicit address 424 management, since in both cases the Attacker A could easily send a 425 control packet adding the address IPA, either as control data or as 426 the source address of the control packet. In order to be able to 427 hijack the connection, the attacker needs to know the 4-tuple that 428 identifies the connection, including the pair of addresses and the 429 pair of ports. It seems reasonable to assume that knowing the source 430 and destination IP addresses and the port of the server side is 431 fairly easy for the attacker. Learning the port of the client (i.e. 432 of the initiator of the connection) may prove to be more challenging. 433 The attacker would need to guess what the port is or to learn it by 434 intercepting the packets. Assuming that the attacker can gather the 435 4-tuple and issue the message adding IPA to the addresses available 436 for the MPTCP connection, then the attacker A has been able to 437 participate in the communication. In particular: 438 o Segments flowing from the Node 2:Depending how the usage of 439 addresses is defined, Node 2 will start using IPA to send data to. 440 In general, since the main goal is to achieve multi-path 441 capabilities, we can assume that unless there are already many IP 442 address pairs in use in the MPTCP connection, Node 2 will start 443 sending data to IPA. This means that part of the data of the 444 communication will reach the Attacker but probably not all of it. 445 This per se, already has negative effects, since Node 1 will not 446 receive all the data from Node 2. However, it is not enough to 447 achieve full hijacking of the connection, since part of data will 448 be still delivered to IP1, so it would reach Node 1 and not the 449 Attacker. In order for the attacker to receive all the data of 450 the MPTCP connection, the Attacker must somehow remove IP1 of the 451 set of available addresses for the connection. in the case of 452 implicit address management, this operation is likely to imply 453 sending a termination packet with IP1 as source address, which may 454 or not be possible for the attacker depending on whether ingress 455 filtering is in place and the location of the attacker. If 456 explicit address management is used, then the attacker will send a 457 remove address control packet containing IP1. The result is that 458 once IP1 is removed, all the data sent by Node 2 will reach the 459 Attacker and the incoming traffic has been hijacked. 460 o Segments flowing to the Node 2: As soon as IPA is accepted by Node 461 2 as part of the address set for the MPTCP connection, the 462 Attacker can send packets using IPA and those packets will be 463 considered by Node 2 as part of MPTCP connection. This means that 464 the attacker will be able to inject data into the MPTCP 465 connection, so from this perspective, the attacker has hijacked 466 part of the outgoing traffic. However, Node 1 would still be able 467 to send traffic that will be received by Node 2 as part of the 468 MPTCP connection. This means that there will be two source of 469 data i.e. Node 1 and the attacker, potentially preventing the 470 full hijacking of the outgoing traffic by the attacker. In order 471 to achieve a full hijacking, the attacker would need to remove IP1 472 from the set of available addresses. This can be done using the 473 same techniques described in the previous paragraph. 475 A related attack that can be achieved using similar techniques would 476 be a Man in the Middle (MitM) attack. The scenario for the attack is 477 depicted in the figure below. 479 +------+ +------+ 480 | Node | --------------- | Node | 481 | 1 |IP1 IP2| 2 | 482 +------+ \ /+------+ 483 \ / 484 \ / 485 \ / 486 v IPA v 487 +--------+ 488 |Attacker| 489 | A | 490 +--------+ 492 In this case, there is an established connection between Node 1 and 493 Node 2. The Attacker A will use the MPTCP address agility 494 capabilities to place itself as a MitM. In order to do so, it will 495 add IP address IPA as an additional address for the MPTCP connection 496 on both Node 1 and Node 2. This is essentially the same technique 497 described earlier in this section, only that it is used against both 498 nodes involved in the communication. The main difference is that in 499 this case, the attacker can simply sniff the content of the 500 communication that is forwarded through it and in turn forward the 501 data to the peer of the communication. The result is that the 502 attacker can place himself in the middle of the communication and 503 sniff part of the traffic unnoticed. Similar considerations about 504 how the attacker can manage to get to see all the traffic by removing 505 the genuine address of the peer apply. 507 6.2. Time-shifted hijacking attacks 509 A simple way to prevent off-path attackers to launch hijacking 510 attacks is to provide security of the control messages that add and 511 remove addresses by the usage of a cookie. In this type of 512 approaches, the peers involved in the MPTCP connection agree on a 513 cookie, that is exchanged in plain text during the establishment of 514 the connection and that needs to be presented in every control packet 515 that adds or removes an address for any of the peers. The result is 516 that the attacker needs to know the cookie in order to launch any of 517 the hijacking attacks described earlier. This implies that off path 518 attackers can no longer perform the hijacking attacks and that only 519 on-path attackers can do so, so one may consider that a cookie based 520 approach to secure MPTCP connection results in similar security than 521 current TCP. While it is close, it is not entirely true. 523 The main difference between the security of a MPTCP protocol secured 524 through cookies and the current TCP protocol are the time shifted 525 attacks. As we described earlier, a time shifted attack is one where 526 the attacker is along the path during a period of time, and then 527 moves away but the effects of the attack still remains, after the 528 attacker is long gone. In the case of a MPTCP protocol secured 529 through the usage of cookies, the attacker needs to be along the path 530 until the cookie is exchanged. After the attacker has learnt the 531 cookie, it can move away from the path and can still launch the 532 hijacking attacks described in the previous section. 534 There are several type of approaches that provide some protection 535 against hijacking attacks and that are vulnerable to some forms of 536 time-shifted attacks. We will next present some general taxonomy of 537 solutions and we describe the residual threats: 538 o Cookie-based solution: As we described earlier, one possible 539 approach is to use a cookie, that is sent in clear text in every 540 MPTCP control message that adds a new address to the existing 541 connection. The residual threat in this type of solution is that 542 any attacker that can sniff any of these control messages will 543 learn the cookie and will be able to add new addresses at any 544 given point in the lifetime of the connection. Moreover, the 545 endpoints will not detect the attack since the original cookie is 546 being used by the attacker. Summarizing, the vulnerability window 547 of this type of attacks includes all the flow establishment 548 exchanges and it is undetectable by the endpoints. 549 o Shared secret exchanged in plain text: An alternative option that 550 is somehow more secure than the cookie based approach is to 551 exchange a key in clear text during the establishment of the first 552 subflow and then validate the following subflows by using an keyed 553 HMAC signature using the shared key. This solution would be 554 vulnerable to attackers sniffing the message exchange for the 555 establishment of the first subflow, but after that, the shared key 556 is not transmitted any more, so the attacker cannot learn it 557 through sniffing any other message. Unfortunately, in order to be 558 compatible with NATs (see analysis below) even though this 559 approach includes a keyed HMAC signature, this signature cannot 560 cover the IP address that is being added. This basically means 561 that this type of approaches are also vulnerable to integrity 562 attacks of the exchanged messages. This means that even though 563 the attacker cannot learn the shared key by sniffing the 564 subsequent sublfow establishment, the attacker can modify the 565 subflow establishment message and change the address that is being 566 added. So, the vulnerability window for confidentially to the 567 shared key is limited to the establishment of the first subflow, 568 but the vulnerability window for integrity attacks still includes 569 all the subflow establishment exchanges. These attacks are still 570 undetectable by the endpoints. It should be noted that the SCTP 571 security falls in this category. 572 o Strong crypto anchor exchange. another approach that could be used 573 would be to exchange some strong crypto anchor while the 574 establishment of the first subflow, such as a public key or a hash 575 chain anchor. In this case, subsequent sublfows could be 576 protected by using the crypto material associated to that anchor. 577 An attacker in this case would need to change the crypto material 578 exchanged in the connection establishment phase. As a result the 579 vulnerability window for forging the crypto anchor is limited to 580 the initial connection establishment exchange. Similarly to the 581 previous case, due to NAT traversal considerations, the 582 vulnerability window for integrity attacks include all the subflow 583 establishment exchanges. As opposed to the previous one, because 584 the attacker needs to change the crypto anchor, this approach are 585 detectable by the endpoints, if they communicate directly. 587 6.3. NAT considerations 589 In order to be widely adopted MPTCP must work through NATs. NATs are 590 an interesting device from a security perspective. In terms of MPTCP 591 they essentially behave as a Man-in-the-middle attacker. As we have 592 described earlier, MPTCP security goal is to prevent from any 593 attacker to insert their addresses as valid addresses for a given 594 MPTCP connection. But that is exactly what a NAT does, they modify 595 the addresses. So, if MPTCP is to work through NATs, MPTCP must 596 accept address rewritten by NATs as valid addresses for a given 597 session. The most direct corollary is that the MPTCP messages that 598 add addresses in the implicit mode (i.e. the SYN of new subflows) 599 cannot be protected against integrity attacks, since they must allow 600 for NATs to change their addresses. This basically rules out any 601 solution that would rely on providing integrity protection to prevent 602 an attacker from changing the address used in a subflow establishment 603 exchange. This implies that alternative creative mechanisms are 604 needed to protect from integrity attacks to the MPTCP signaling that 605 adds new addresses to a connection. It is far from obvious how one 606 such creative approach could look like at this point. 608 7. Reccomendation 610 The presented analysis shows that there is a tradeoff between the 611 complexity of the security solution and the residual threats. In 612 order to define a proper security solution, we need to assess the 613 tradeoff and make a recommendation. After evaluating the different 614 aspects in the MPTCP WG, our conclusion is that the default security 615 mechanisms for MPTCP should be to exchange a key in the establishment 616 of the first subflow and then secure following address additions by 617 using a keyed HMAC using the exchanged key (i.e. similar to the one 618 used by SCTP). 620 In addition, our recommendation is that the MPTCP protocol should be 621 extensible and it should able to accommodate multiple security 622 solutions, in order to enable the usage of more secure mechanisms if 623 needed. 625 8. Security Considerations 627 This note contains a security analysis for MPTCP, so no further 628 security considerations need to be described in this section 630 9. Contributors 632 Alan Ford - Roke Manor Research Ltd. 634 10. Acknowledgments 636 Rolf Winter reviewed an earlier version of this document and provided 637 comments to improve it. 639 Mark Handley pointed out the problem with NATs and integrity 640 protection of MPTCP signaling. 642 Marcelo Bagnulo is partly funded by Trilogy, a research project 643 supported by the European Commission under its Seventh Framework 644 Program. 646 11. Informative References 648 [RFC4225] Nikander, P., Arkko, J., Aura, T., Montenegro, G., and E. 649 Nordmark, "Mobile IP Version 6 Route Optimization Security 650 Design Background", RFC 4225, December 2005. 652 [RFC4218] Nordmark, E. and T. Li, "Threats Relating to IPv6 653 Multihoming Solutions", RFC 4218, October 2005. 655 [RFC3972] Aura, T., "Cryptographically Generated Addresses (CGA)", 656 RFC 3972, March 2005. 658 [RFC5062] Stewart, R., Tuexen, M., and G. Camarillo, "Security 659 Attacks Found Against the Stream Control Transmission 660 Protocol (SCTP) and Current Countermeasures", RFC 5062, 661 September 2007. 663 [RFC5535] Bagnulo, M., "Hash-Based Addresses (HBA)", RFC 5535, 664 June 2009. 666 Author's Address 668 Marcelo Bagnulo 669 Universidad Carlos III de Madrid 670 Av. Universidad 30 671 Leganes, Madrid 28911 672 SPAIN 674 Phone: 34 91 6248814 675 Email: marcelo@it.uc3m.es 676 URI: http://www.it.uc3m.es