idnits 2.17.1 draft-ietf-dots-requirements-11.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The 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 (January 23, 2018) is 2285 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 793 (Obsoleted by RFC 9293) ** Obsolete normative reference: RFC 5405 (Obsoleted by RFC 8085) == Outdated reference: A later version (-18) exists of draft-ietf-dots-architecture-05 == Outdated reference: A later version (-25) exists of draft-ietf-dots-use-cases-09 Summary: 3 errors (**), 0 flaws (~~), 3 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 DOTS A. Mortensen 3 Internet-Draft Arbor Networks 4 Intended status: Informational R. Moskowitz 5 Expires: July 27, 2018 Huawei 6 T. Reddy 7 McAfee, Inc. 8 January 23, 2018 10 Distributed Denial of Service (DDoS) Open Threat Signaling Requirements 11 draft-ietf-dots-requirements-11 13 Abstract 15 This document defines the requirements for the Distributed Denial of 16 Service (DDoS) Open Threat Signaling (DOTS) protocols enabling 17 coordinated response to DDoS attacks. 19 Status of This Memo 21 This Internet-Draft is submitted in full conformance with the 22 provisions of BCP 78 and BCP 79. 24 Internet-Drafts are working documents of the Internet Engineering 25 Task Force (IETF). Note that other groups may also distribute 26 working documents as Internet-Drafts. The list of current Internet- 27 Drafts is at https://datatracker.ietf.org/drafts/current/. 29 Internet-Drafts are draft documents valid for a maximum of six months 30 and may be updated, replaced, or obsoleted by other documents at any 31 time. It is inappropriate to use Internet-Drafts as reference 32 material or to cite them other than as "work in progress." 34 This Internet-Draft will expire on July 27, 2018. 36 Copyright Notice 38 Copyright (c) 2018 IETF Trust and the persons identified as the 39 document authors. All rights reserved. 41 This document is subject to BCP 78 and the IETF Trust's Legal 42 Provisions Relating to IETF Documents 43 (https://trustee.ietf.org/license-info) in effect on the date of 44 publication of this document. Please review these documents 45 carefully, as they describe your rights and restrictions with respect 46 to this document. Code Components extracted from this document must 47 include Simplified BSD License text as described in Section 4.e of 48 the Trust Legal Provisions and are provided without warranty as 49 described in the Simplified BSD License. 51 Table of Contents 53 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 54 1.1. Context and Motivation . . . . . . . . . . . . . . . . . 2 55 1.2. Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 56 2. Requirements . . . . . . . . . . . . . . . . . . . . . . . . 5 57 2.1. General Requirements . . . . . . . . . . . . . . . . . . 6 58 2.2. Signal Channel Requirements . . . . . . . . . . . . . . . 7 59 2.3. Data Channel Requirements . . . . . . . . . . . . . . . . 12 60 2.4. Security Requirements . . . . . . . . . . . . . . . . . . 13 61 2.5. Data Model Requirements . . . . . . . . . . . . . . . . . 14 62 3. Congestion Control Considerations . . . . . . . . . . . . . . 15 63 3.1. Signal Channel . . . . . . . . . . . . . . . . . . . . . 15 64 3.2. Data Channel . . . . . . . . . . . . . . . . . . . . . . 15 65 4. Security Considerations . . . . . . . . . . . . . . . . . . . 16 66 5. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 17 67 6. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 17 68 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 17 69 7.1. Normative References . . . . . . . . . . . . . . . . . . 17 70 7.2. Informative References . . . . . . . . . . . . . . . . . 19 71 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 20 73 1. Introduction 75 1.1. Context and Motivation 77 Distributed Denial of Service (DDoS) attacks afflict networks of all 78 kinds, plaguing network operators at service providers and 79 enterprises around the world. High-volume attacks saturating inbound 80 links are now common, as attack scale and frequency continue to 81 increase. 83 The prevalence and impact of these DDoS attacks has led to an 84 increased focus on coordinated attack response. However, many 85 enterprises lack the resources or expertise to operate on-premises 86 attack mitigation solutions themselves, or are constrained by local 87 bandwidth limitations. To address such gaps, service providers have 88 begun to offer on-demand traffic scrubbing services, which are 89 designed to separate the DDoS attack traffic from legitimate traffic 90 and forward only the latter. 92 Today, these services offer proprietary interfaces for subscribers to 93 request attack mitigation. Such proprietary interfaces tie a 94 subscriber to a service while also limiting the network elements 95 capable of participating in the attack mitigation. As a result of 96 signaling interface incompatibility, attack responses may be 97 fragmented or otherwise incomplete, leaving operators in the attack 98 path unable to assist in the defense. 100 A standardized method to coordinate a real-time response among 101 involved operators will increase the speed and effectiveness of DDoS 102 attack mitigation, and reduce the impact of these attacks. This 103 document describes the required characteristics of protocols that 104 enable attack coordination and mitigation of DDoS attacks. 106 DDoS Open Threat Signaling (DOTS) communicates the need for defensive 107 action in anticipation of or in response to an attack, but does not 108 dictate the implementation of these actions. The requirements in 109 this document are derived from [I-D.ietf-dots-use-cases] and 110 [I-D.ietf-dots-architecture]. 112 1.2. Terminology 114 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 115 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 116 document are to be interpreted as described in [RFC2119]. 118 This document adopts the following terms: 120 DDoS: A distributed denial-of-service attack, in which traffic 121 originating from multiple sources is directed at a target on a 122 network. DDoS attacks are intended to cause a negative impact on 123 the availability and/or other functionality of an attack target. 124 Denial-of-service considerations are discussed in detail in 125 [RFC4732]. 127 DDoS attack target: A network connected entity with a finite set of 128 resources, such as network bandwidth, memory or CPU, that is the 129 target of a DDoS attack. Potential targets include (but are not 130 limited to) network elements, network links, servers, and 131 services. 133 DDoS attack telemetry: Collected measurements and behavioral 134 characteristics defining the nature of a DDoS attack. 136 Countermeasure: An action or set of actions focused on recognizing 137 and filtering out specific types of DDoS attack traffic while 138 passing legitimate traffic to the attack target. Distinct 139 countermeasures can be layered to defend against attacks combining 140 multiple DDoS attack types. 142 Mitigation: A set of countermeasures enforced against traffic 143 destined for the target or targets of a detected or reported DDoS 144 attack, where countermeasure enforcement is managed by an entity 145 in the network path between attack sources and the attack target. 146 Mitigation methodology is out of scope for this document. 148 Mitigator: An entity, typically a network element, capable of 149 performing mitigation of a detected or reported DDoS attack. The 150 means by which this entity performs these mitigations and how they 151 are requested of it are out of scope. The mitigator and DOTS 152 server receiving a mitigation request are assumed to belong to the 153 same administrative entity. 155 DOTS client: A DOTS-aware software module responsible for requesting 156 attack response coordination with other DOTS-aware elements. 158 DOTS server: A DOTS-aware software module handling and responding to 159 messages from DOTS clients. The DOTS server enables mitigation on 160 behalf of the DOTS client, if requested, by communicating the DOTS 161 client's request to the mitigator and returning selected mitigator 162 feedback to the requesting DOTS client. 164 DOTS agent: Any DOTS-aware software module capable of participating 165 in a DOTS signal or data channel. It can be a DOTS client, DOTS 166 server, or, as a logical agent, a DOTS gateway. 168 DOTS gateway: A DOTS-aware software module resulting from the 169 logical concatenation of the functionality of a DOTS server and a 170 DOTS client into a single DOTS agent. This functionality is 171 analogous to a Session Initiation Protocol (SIP) [RFC3261] Back- 172 to-Back User Agent (B2BUA) [RFC7092]. A DOTS gateway has a 173 client-facing side, which behaves as a DOTS server for downstream 174 clients, and a server-facing side, which performs the role of DOTS 175 client for upstream DOTS servers. Client-domain DOTS gateways are 176 DOTS gateways that are in the DOTS client's domain, while server- 177 domain DOTS gateways denote DOTS gateways that are in the DOTS 178 server's domain. DOTS gateways are described further in 179 [I-D.ietf-dots-architecture]. 181 Signal channel: A bidirectional, mutually authenticated 182 communication channel between DOTS agents that is resilient even 183 in conditions leading to severe packet loss, such as a volumetric 184 DDoS attack causing network congestion. 186 DOTS signal: A concise authenticated status/control message 187 transmitted over the signal channel between DOTS agents, used to 188 indicate the client's need for mitigation, as well as to convey 189 the status of any requested mitigation. 191 Heartbeat: A message transmitted between DOTS agents over the signal 192 channel, used as a keep-alive and to measure peer health. 194 Data channel: A bidirectional, mutually authentication 195 communincation channel between two DOTS agents used for infrequent 196 but reliable bulk exchange of data not easily or appropriately 197 communicated through the signal channel under attack conditions. 199 Filter: A specification of a matching network traffic flow or set of 200 flows. The filter will typically have a policy associated with 201 it, e.g., rate-limiting or discarding matching traffic [RFC4949]. 203 Blacklist: A list of filters indicating sources from which traffic 204 should be blocked, regardless of traffic content. 206 Whitelist: A list of filters indicating sources from which traffic 207 should always be allowed, regardless of contradictory data gleaned 208 in a detected attack. 210 Multi-homed DOTS client: A DOTS client exchanging messages with 211 multiple DOTS servers, each in a separate administrative domain. 213 2. Requirements 215 This section describes the required features and characteristics of 216 the DOTS protocols. 218 The DOTS protocols enable and manage mitigation on behalf of a 219 network domain or resource which is or may become the focus of a DDoS 220 attack. An active DDoS attack against the entity controlling the 221 DOTS client need not be present before establishing a communication 222 channel between DOTS agents. Indeed, establishing a relationship 223 with peer DOTS agents during normal network conditions provides the 224 foundation for more rapid attack response against future attacks, as 225 all interactions setting up DOTS, including any business or service 226 level agreements, are already complete. Reachability information of 227 peer DOTS agents is provisioned to a DOTS client using a variety of 228 manual or dynamic methods. Once a relationship between DOTS agents 229 is established, regular communication between DOTS clients and 230 servers enables a common understanding of the DOTS agents' health and 231 activity. 233 The DOTS protocol must at a minimum make it possible for a DOTS 234 client to request aid mounting a defense, coordinated by a DOTS 235 server, against a suspected attack, signaling within or between 236 domains as requested by local operators. DOTS clients should 237 similarly be able to withdraw aid requests. DOTS requires no 238 justification from DOTS clients for requests for help, nor do DOTS 239 clients need to justify withdrawing help requests: the decision is 240 local to the DOTS clients' domain. Multi-homed DOTS clients must be 241 able to select the appropriate DOTS server(s) to which a mitigation 242 request is to be sent. The method for selecting the appropriate DOTS 243 server in a multi-homed environment is out of scope. 245 DOTS protocol implementations face competing operational goals when 246 maintaining this bidirectional communication stream. On the one 247 hand, DOTS must include protections ensuring message confidentiality, 248 integrity and authenticity to keep the protocols from becoming 249 additional vectors for the very attadcks it is meant to help fight 250 off. On the other hand, the protocol must be resilient under 251 extremely hostile network conditions, providing continued contact 252 between DOTS agents even as attack traffic saturates the link. Such 253 resiliency may be developed several ways, but characteristics such as 254 small message size, asynchronous, redundant message delivery and 255 minimal connection overhead (when possible given local network 256 policy) will tend to contribute to the robustness demanded by a 257 viable DOTS protocol. Operators of peer DOTS-enabled domains may 258 enable quality- or class-of-service traffic tagging to increase the 259 probability of successful DOTS signal delivery, but DOTS does not 260 require such policies be in place, and should be viable in their 261 absence. 263 The DOTS server and client must also have some standardized method of 264 defining the scope of any mitigation, and negotiating related 265 mitigation communication and actions and communications. 267 Finally, DOTS should be sufficiently extensible to meet future needs 268 in coordinated attack defense, although this consideration is 269 necessarily superseded by the other operational requirements. 271 2.1. General Requirements 273 GEN-001 Extensibility: Protocols and data models developed as part 274 of DOTS MUST be extensible in order to keep DOTS adaptable to 275 operational and proprietary DDoS defenses. Future extensions MUST 276 be backward compatible. DOTS protocols MUST use a version number 277 system to distinguish protocol revisions. Implementations of 278 older protocol versions SHOULD ignore information added to DOTS 279 messages as part of newer protocol versions. 281 GEN-002 Resilience and Robustness: The signaling protocol MUST be 282 designed to maximize the probability of signal delivery even under 283 the severely constrained network conditions caused by particular 284 attack traffic. The protocol MUST be resilient, that is, continue 285 operating despite message loss and out-of-order or redundant 286 message delivery. In support of signaling protocol robustness, 287 DOTS signals SHOULD be conveyed over a transport not susceptible 288 to Head of Line Blocking. 290 GEN-003 Bidirectionality: To support peer health detection, to 291 maintain an active signal channel, and increase the probability of 292 signal delivery during an attack, the signal channel MUST be 293 bidirectional, with client and server transmitting signals to each 294 other at regular intervals, regardless of any client request for 295 mitigation. Unidirectional messages MUST be supported within the 296 bidirectional signal channel to allow for unsolicited message 297 delivery, enabling asynchronous notifications between DOTS agents. 299 GEN-004 Bulk Data Exchange: Infrequent bulk data exchange between 300 DOTS agents can also significantly augment attack response 301 coordination, permitting such tasks as population of black- or 302 white-listed source addresses; address or prefix group aliasing; 303 exchange of incident reports; and other hinting or configuration 304 supplementing attack response. 306 As the resilience requirements for the DOTS signal channel mandate 307 small signal message size, a separate, secure data channel 308 utilizing a reliable transport protocol MUST be used for bulk data 309 exchange. 311 2.2. Signal Channel Requirements 313 SIG-001 Use of Common Transport Protocols: DOTS MUST operate over 314 common widely deployed and standardized transport protocols. 315 While connectionless transport such as the User Datagram Protocol 316 (UDP) [RFC0768] SHOULD be used for the signal channel, the 317 Transmission Control Protocol (TCP) [RFC0793] MAY be used if 318 necessary due to network policy or middlebox capabilities or 319 configurations. 321 SIG-002 Sub-MTU Message Size: To avoid message fragmentation and the 322 consequently decreased probability of message delivery over a 323 congested link, signaling protocol message size MUST be kept under 324 signaling Path Maximum Transmission Unit (PMTU), including the 325 byte overhead of any encapsulation, transport headers, and 326 transport- or message-level security. 328 DOTS agents SHOULD attempt to learn the PMTU through mechanisms 329 such as Path MTU Discovery [RFC1191] or Packetization Layer Path 330 MTU Discovery [RFC4821]. If the PMTU cannot be discovered, DOTS 331 agents SHOULD assume a PMTU of 1280 bytes. If IPv4 support on 332 legacy or otherwise unusual networks is a consideration and PMTU 333 is unknown, DOTS implementations MAY rely on a PMTU of 576 bytes, 334 as discussed in [RFC0791] and [RFC1122]. 336 SIG-003 Channel Health Monitoring: DOTS agents MUST support exchange 337 of heartbeat messages over the signal channel to monitor channel 338 health. Peer DOTS agents SHOULD regularly send heartbeats to each 339 other while a mitigation request is active. The heartbeat 340 interval during active mitigation could be negotiable, but SHOULD 341 be frequent enough to maintain any on-path NAT or Firewall 342 bindings during mitigation. 344 To support scenarios in which loss of heartbeat is used to trigger 345 mitigation, and to keep the channel active, DOTS clients MAY 346 solicit heartbeat exchanges after successful mutual 347 authentication. When DOTS agents are exchanging heartbeats and no 348 mitigation request is active, either agent MAY request changes to 349 the heartbeat rate. For example, a DOTS server might want to 350 reduce heartbeat frequency or cease heartbeat exchanges when an 351 active DOTS client has not requested mitigation, in order to 352 control load. 354 Following mutual authentication, a signal channel MUST be 355 considered active until a DOTS agent explicitly ends the session, 356 or either DOTS agent fails to receive heartbeats from the other 357 after a mutually agreed upon retransmission procedure has been 358 exhausted. Because heartbeat loss is much more likely during 359 volumetric attack, DOTS agents SHOULD avoid signal channel 360 termination when mitigation is active and heartbeats are not 361 received by either DOTS agent for an extended period. In such 362 circumstances, DOTS clients MAY attempt to reestablish the signal 363 channel, but SHOULD continue to send heartbeats so that the DOTS 364 server knows the session is still alive. DOTS servers are assumed 365 to have the ability to monitor the attack, using feedback from the 366 mitigator and other available sources, and MAY use the absence of 367 attack traffic and lack of client heartbeats as an indication the 368 signal channel is defunct. 370 SIG-004 Channel Redirection: In order to increase DOTS operational 371 flexibility and scalability, DOTS servers SHOULD be able to 372 redirect DOTS clients to another DOTS server at any time. DOTS 373 clients MUST NOT assume the redirection target DOTS server shares 374 security state with the redirecting DOTS server. DOTS clients are 375 free to attempt abbreviated security negotiation methods supported 376 by the protocol, such as DTLS session resumption, but MUST be 377 prepared to negotiate new security state with the redirection 378 target DOTS server. 380 Due to the increased likelihood of packet loss caused by link 381 congestion during an attack, DOTS servers SHOULD NOT redirect 382 while mitigation is enabled during an active attack against a 383 target in the DOTS client's domain. 385 SIG-005 Mitigation Requests and Status: Authorized DOTS clients MUST 386 be able to request scoped mitigation from DOTS servers. DOTS 387 servers MUST send status to the DOTS clients about mitigation 388 requests. If a DOTS server rejects an authorized request for 389 mitigation, the DOTS server MUST include a reason for the 390 rejection in the status message sent to the client. 392 Due to the higher likelihood of packet loss during a DDoS attack, 393 DOTS servers SHOULD regularly send mitigation status to authorized 394 DOTS clients which have requested and been granted mitigation, 395 regardless of client requests for mitigation status. 397 When DOTS client-requested mitigation is active, DOTS server 398 status messages SHOULD include the following mitigation metrics: 400 * Total number of packets blocked by the mitigation 402 * Current number of packets per second blocked 404 * Total number of bytes blocked 406 * Current number of bytes per second blocked 408 DOTS clients MAY take these metrics into account when determining 409 whether to ask the DOTS server to cease mitigation. 411 A DOTS client MAY withdraw a mitigation request at any time, 412 regardless of whether mitigation is currently active. The DOTS 413 server MUST immediately acknowledge a DOTS client's request to 414 stop mitigation. 416 To protect against route or DNS flapping caused by a client 417 rapidly toggling mitigation, and to dampen the effect of 418 oscillating attacks, DOTS servers MAY allow mitigation to continue 419 for a limited period after acknowledging a DOTS client's 420 withdrawal of a mitigation request. During this period, DOTS 421 server status messages SHOULD indicate that mitigation is active 422 but terminating. 424 The initial active-but-terminating period is implementation- and 425 deployment- specific, but SHOULD be sufficiently long to absorb 426 latency incurred by route propagation. If the client requests 427 mitigation again before the initial active-but-terminating period 428 elapses, the DOTS server MAY exponentially increase the active- 429 but-terminating period up to a maximum of 300 seconds (5 minutes). 430 After the active-but-terminating period elapses, the DOTS server 431 MUST treat the mitigation as terminated, as the DOTS client is no 432 longer responsible for the mitigation. 434 SIG-006 Mitigation Lifetime: DOTS servers MUST support mitigations 435 for a negotiated time interval or lifetime, and MUST terminate a 436 mitigation when the lifetime elapses. DOTS servers also MUST 437 support renewal of mitigation lifetimes in mitigation requests 438 from DOTS clients, allowing clients to extend mitigation as 439 necessary for the duration of an attack. 441 DOTS servers MUST treat a mitigation terminated due to lifetime 442 expiration exactly as if the DOTS client originating the 443 mitigation had asked to end the mitigation, including the active- 444 but-terminating period, as described above in SIG-005. 446 DOTS clients MUST include a mitigation lifetime in all mitigation 447 requests. 449 DOTS servers SHOULD support indefinite mitigation lifetimes, 450 enabling architectures in which the mitigator is always in the 451 traffic path to the resources for which the DOTS client is 452 requesting protection. DOTS clients MUST be prepared to not be 453 granted mitigations with indefinite lifetimes. DOTS servers MAY 454 refuse mitigations with indefinite lifetimes, for policy reasons. 455 The reasons themselves are out of scope. If the DOTS server does 456 not grant a mitigation request with an indefinite mitigation 457 lifetime, it MUST set the lifetime to a value that is configured 458 locally. That value MUST be returned in a reply to the requesting 459 DOTS client. 461 SIG-007 Mitigation Scope: DOTS clients MUST indicate desired 462 mitigation scope. The scope type will vary depending on the 463 resources requiring mitigation. All DOTS agent implementations 464 MUST support the following required scope types: 466 * IPv4 prefixes in CIDR notation [RFC4632] 468 * IPv6 prefixes [RFC4291][RFC5952] 470 * Domain names [RFC1035] 472 The following mitigation scope types are OPTIONAL: 474 * Uniform Resource Identifiers [RFC3986] 476 DOTS servers MUST be able to resolve domain names and (when 477 supported) URIs. How name resolution is managed on the DOTS 478 server is implementation-specific. 480 DOTS agents MUST support mitigation scope aliases, allowing DOTS 481 clients and servers to refer to collections of protected resources 482 by an opaque identifier created through the data channel, direct 483 configuration, or other means. Domain name and URI mitigation 484 scopes may be thought of as a form of scope alias, in which the 485 addresses to which the domain name or URI resolve represent the 486 full scope of the mitigation. 488 If there is additional information available narrowing the scope 489 of any requested attack response, such as targeted port range, 490 protocol, or service, DOTS clients SHOULD include that information 491 in client mitigation requests. DOTS clients MAY also include 492 additional attack details. DOTS servers MAY ignore such 493 supplemental information when enabling countermeasures on the 494 mitigator. 496 As an active attack evolves, DOTS clients MUST be able to adjust 497 as necessary the scope of requested mitigation by refining the 498 scope of resources requiring mitigation. 500 A DOTS client may obtain the mitigation scope through direct 501 provisioning or through implementation-specific methods of 502 discovery. DOTS clients MUST support at least one mechanism to 503 obtain mitigation scope. 505 SIG-008 Mitigation Efficacy: When a mitigation request is active, 506 DOTS clients SHOULD transmit a metric of perceived mitigation 507 efficacy to the DOTS server. DOTS servers MAY use the efficacy 508 metric to adjust countermeasures activated on a mitigator on 509 behalf of a DOTS client. 511 SIG-009 Conflict Detection and Notification: Multiple DOTS clients 512 controlled by a single administrative entity may send conflicting 513 mitigation requests as a result of misconfiguration, operator 514 error, or compromised DOTS clients. DOTS servers in the same 515 administrative domain attempting to honor conflicting requests may 516 flap network route or DNS information, degrading the networks 517 attempting to participate in attack response with the DOTS 518 clients. DOTS servers in a single administrative domain SHALL 519 detect such conflicting requests, and SHALL notify the DOTS 520 clients in conflict. The notification SHOULD indicate the nature 521 and scope of the conflict, for example, the overlapping prefix 522 range in a conflicting mitigation request. 524 SIG-010: Network Address Translator Traversal: DOTS clients may be 525 deployed behind a Network Address Translator (NAT), and need to 526 communicate with DOTS servers through the NAT. DOTS protocols 527 MUST therefore be capable of traversing NATs. 529 If UDP is used as the transport for the DOTS signal channel, all 530 considerations in "Middlebox Traversal Guidelines" in [RFC8085] 531 apply to DOTS. Regardless of transport, DOTS protocols MUST 532 follow established best common practices established in BCP 127 533 for NAT traversal [RFC4787][RFC6888][RFC7857]. 535 2.3. Data Channel Requirements 537 The data channel is intended to be used for bulk data exchanges 538 between DOTS agents. Unlike the signal channel, the data channel is 539 not expected to be constructed to deal with attack conditions. As 540 the primary function of the data channel is data exchange, a reliable 541 transport is required in order for DOTS agents to detect data 542 delivery success or failure. 544 The data channel provides a protocol for DOTS configuration, 545 management. For example, a DOTS client may submit to a DOTS server a 546 collection of prefixes it wants to refer to by alias when requesting 547 mitigation, to which the server would respond with a success status 548 and the new prefix group alias, or an error status and message in the 549 event the DOTS client's data channel request failed. 551 DATA-001 Reliable transport: Messages sent over the data channel 552 MUST be delivered reliably, in order sent. 554 DATA-002 Data privacy and integrity: Transmissions over the data 555 channel are likely to contain operationally or privacy-sensitive 556 information or instructions from the remote DOTS agent. Theft or 557 modification of data channel transmissions could lead to 558 information leaks or malicious transactions on behalf of the 559 sending agent (see Section 4 below). Consequently data sent over 560 the data channel MUST be encrypted and authenticated using current 561 IETF best practices. DOTS servers MUST enable means to prevent 562 leaking operationally or privacy-sensitive data. Although 563 administrative entities participating in DOTS may detail what data 564 may be revealed to third-party DOTS agents, such considerations 565 are not in scope for this document. 567 DATA-003 Resource Configuration: To help meet the general and signal 568 channel requirements in Section 2.1 and Section 2.2, DOTS server 569 implementations MUST provide an interface to configure resource 570 identifiers, as described in SIG-007. DOTS server implementations 571 MAY expose additional configurability. Additional configurability 572 is implementation-specific. 574 DATA-004 Black- and whitelist management: DOTS servers MUST provide 575 methods for DOTS clients to manage black- and white-lists of 576 traffic destined for resources belonging to a client. 578 For example, a DOTS client should be able to create a black- or 579 whitelist entry, retrieve a list of current entries from either 580 list, update the content of either list, and delete entries as 581 necessary. 583 How a DOTS server authorizes DOTS client management of black- and 584 white-list entries is implementation-specific. 586 2.4. Security Requirements 588 DOTS must operate within a particularly strict security context, as 589 an insufficiently protected signal or data channel may be subject to 590 abuse, enabling or supplementing the very attacks DOTS purports to 591 mitigate. 593 SEC-001 Peer Mutual Authentication: DOTS agents MUST authenticate 594 each other before a DOTS signal or data channel is considered 595 valid. The method of authentication is not specified, but should 596 follow current industry best practices with respect to any 597 cryptographic mechanisms to authenticate the remote peer. 599 SEC-002 Message Confidentiality, Integrity and Authenticity: DOTS 600 protocols MUST take steps to protect the confidentiality, 601 integrity and authenticity of messages sent between client and 602 server. While specific transport- and message-level security 603 options are not specified, the protocols MUST follow current 604 industry best practices for encryption and message authentication. 606 In order for DOTS protocols to remain secure despite advancements 607 in cryptanalysis and traffic analysis, DOTS agents MUST be able to 608 negotiate the terms and mechanisms of protocol security, subject 609 to the interoperability and signal message size requirements in 610 Section 2.2. 612 While the interfaces between downstream DOTS server and upstream 613 DOTS client within a DOTS gateway are implementation-specific, 614 those interfaces nevertheless MUST provide security equivalent to 615 that of the signal channels bridged by gateways in the signaling 616 path. For example, when a DOTS gateway consisting of a DOTS 617 server and DOTS client is running on the same logical device, the 618 two DOTS agents could be implemented within the same process 619 security boundary. 621 SEC-003 Message Replay Protection: To prevent a passive attacker 622 from capturing and replaying old messages, and thereby potentially 623 disrupting or influencing the network policy of the receiving DOTS 624 agent's domain, DOTS protocols MUST provide a method for replay 625 detection and prevention. 627 Within the signal channel, messages MUST be uniquely identified 628 such that replayed or duplicated messages can be detected and 629 discarded. Unique mitigation requests MUST be processed at most 630 once. 632 SEC-004 Authorization: DOTS servers MUST authorize all messages from 633 DOTS clients which pertain to mitigation, configuration, 634 filtering, or status. 636 DOTS servers MUST reject mitigation requests with scopes which the 637 DOTS client is not authorized to manage. 639 Likewise, DOTS servers MUST refuse to allow creation, modification 640 or deletion of scope aliases and black-/white-lists when the DOTS 641 client is unauthorized. 643 The modes of authorization are implementation-specific. 645 2.5. Data Model Requirements 647 A well-structured DOTS data model is critical to the development of 648 successful DOTS protocols. 650 DM-001: Structure: The data model structure for the DOTS protocol 651 MAY be described by a single module, or be divided into related 652 collections of hierarchical modules and sub-modules. If the data 653 model structure is split across modules, those distinct modules 654 MUST allow references to describe the overall data model's 655 structural dependencies. 657 DM-002: Versioning: To ensure interoperability between DOTS protocol 658 implementations, data models MUST be versioned. How the protocols 659 represent data model versions is not defined in this document. 661 DM-003: Mitigation Status Representation: The data model MUST 662 provide the ability to represent a request for mitigation and the 663 withdrawal of such a request. The data model MUST also support a 664 representation of currently requested mitigation status, including 665 failures and their causes. 667 DM-004: Mitigation Scope Representation: The data model MUST support 668 representation of a requested mitigation's scope. As mitigation 669 scope may be represented in several different ways, per SIG-007 670 above, the data model MUST be capable of flexible representation 671 of mitigation scope. 673 DM-005: Mitigation Lifetime Representation: The data model MUST 674 support representation of a mitigation request's lifetime, 675 including mitigations with no specified end time. 677 DM-006: Mitigation Efficacy Representation: The data model MUST 678 support representation of a DOTS client's understanding of the 679 efficacy of a mitigation enabled through a mitigation request. 681 DM-007: Acceptable Signal Loss Representation: The data model MUST 682 be able to represent the DOTS agent's preference for acceptable 683 signal loss when establishing a signal channel, as described in 684 GEN-002. 686 DM-008: Heartbeat Interval Representation: The data model MUST be 687 able to represent the DOTS agent's preferred heartbeat interval, 688 which the client may include when establishing the signal channel, 689 as described in SIG-003. 691 DM-009: Relationship to Transport: The DOTS data model MUST NOT 692 depend on the specifics of any transport to represent fields in 693 the model. 695 3. Congestion Control Considerations 697 3.1. Signal Channel 699 As part of a protocol expected to operate over links affected by DDoS 700 attack traffic, the DOTS signal channel MUST NOT contribute 701 significantly to link congestion. To meet the signal channel 702 requirements above, DOTS signal channel implementations SHOULD 703 support connectionless transports. However, some connectionless 704 transports when deployed naively can be a source of network 705 congestion, as discussed in [RFC5405]. Signal channel 706 implementations using such connectionless transports, such as UDP, 707 therefore MUST include a congestion control mechanism. 709 Signal channel implementations using TCP may rely on built-in TCP 710 congestion control support. 712 3.2. Data Channel 714 As specified in DATA-001, the data channel requires reliable, in- 715 order message delivery. Data channel implementations using TCP may 716 rely on the TCP implementation's built-in congestion control 717 mechanisms. 719 4. Security Considerations 721 This document informs future protocols under development, and so does 722 not have its security considerations of its own. However, naive DOTS 723 deployment potentially exposes networks to new attack vectors. The 724 three primary attack vectors are DOTS agent impersonation, traffic 725 injection, and signal blocking. 727 Impersonation of either DOTS server or DOTS client could have 728 catastrophic impact on operations in either domain. Should an 729 attacker develop the ability to impersonate a DOTS client, that 730 attacker can affect policy on the network path to the DOTS client's 731 domain, up to and including instantiation of blacklists blocking all 732 inbound traffic to networks for which the DOTS client is authorized 733 to request mitigation. Similarly, an impersonated DOTS server may be 734 able to act as a sort of malicious DOTS gateway, intercepting 735 requests from the downstream DOTS client, modifying them to inflict 736 the desired impact on traffic to or from the DOTS client's domain. 737 Among other things, this malicious DOTS gateway might receive 738 mitigation requests from the DOTS client, and simply discard them, 739 ensuring no mitigation is ever applied. 741 Traffic injection into a naive DOTS deployment could allow an 742 attacker to affect DOTS operations selectively. Rather than 743 impersonating a DOTS agent directly, the attacker crafts DOTS signal 744 or data channel messages in such a way that the targeted DOTS agent 745 treats them as if they originated with a legitimate DOTS agent, for 746 example, by spoofing the sender's IP address. As with agent 747 impersonation, the attacker capable of injecting traffic can affect 748 the network path to addresses for which the DOTS client is authorized 749 to request mitigation. 751 Blocking communication between DOTS agents-signal blocking-has the 752 potential to disrupt the core function of DOTS, which is to request 753 mitigation of active or expected DDoS attacks. The DOTS signal 754 channel is expected to operate over congested inbound links, and, as 755 described in Section 2.2, the signal channel protocol must be 756 designed for minimal data transfer to reduce the incidence of signal 757 blocking. 759 As detailed in Section 2.4, DOTS implementations require mutual 760 authentication of DOTS agents in order to make agent impersonation 761 and traffic injection more difficult. However, impersonation or 762 traffic injection may still be possible as a result of credential 763 theft, implementation flaws, or compromise of DOTS agents. Operators 764 should take steps to reduce attack surfaces through current secure 765 network communications best practices. 767 5. Contributors 769 Mohamed Boucadair 770 Orange 772 mohamed.boucadair@orange.com 774 Flemming Andreasen 775 Cisco Systems, Inc. 777 fandreas@cisco.com 779 Dave Dolson 780 Sandvine 782 ddolson@sandvine.com 784 6. Acknowledgments 786 Thanks to Roman Danyliw and Matt Richardson for careful reading and 787 feedback. 789 7. References 791 7.1. Normative References 793 [RFC0768] Postel, J., "User Datagram Protocol", STD 6, RFC 768, 794 DOI 10.17487/RFC0768, August 1980, 795 . 797 [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791, 798 DOI 10.17487/RFC0791, September 1981, 799 . 801 [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, 802 RFC 793, DOI 10.17487/RFC0793, September 1981, 803 . 805 [RFC1035] Mockapetris, P., "Domain names - implementation and 806 specification", STD 13, RFC 1035, DOI 10.17487/RFC1035, 807 November 1987, . 809 [RFC1122] Braden, R., Ed., "Requirements for Internet Hosts - 810 Communication Layers", STD 3, RFC 1122, 811 DOI 10.17487/RFC1122, October 1989, 812 . 814 [RFC1191] Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191, 815 DOI 10.17487/RFC1191, November 1990, 816 . 818 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 819 Requirement Levels", BCP 14, RFC 2119, 820 DOI 10.17487/RFC2119, March 1997, 821 . 823 [RFC3986] Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform 824 Resource Identifier (URI): Generic Syntax", STD 66, 825 RFC 3986, DOI 10.17487/RFC3986, January 2005, 826 . 828 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 829 Architecture", RFC 4291, DOI 10.17487/RFC4291, February 830 2006, . 832 [RFC4632] Fuller, V. and T. Li, "Classless Inter-domain Routing 833 (CIDR): The Internet Address Assignment and Aggregation 834 Plan", BCP 122, RFC 4632, DOI 10.17487/RFC4632, August 835 2006, . 837 [RFC4787] Audet, F., Ed. and C. Jennings, "Network Address 838 Translation (NAT) Behavioral Requirements for Unicast 839 UDP", BCP 127, RFC 4787, DOI 10.17487/RFC4787, January 840 2007, . 842 [RFC4821] Mathis, M. and J. Heffner, "Packetization Layer Path MTU 843 Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007, 844 . 846 [RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines 847 for Application Designers", RFC 5405, 848 DOI 10.17487/RFC5405, November 2008, 849 . 851 [RFC6888] Perreault, S., Ed., Yamagata, I., Miyakawa, S., Nakagawa, 852 A., and H. Ashida, "Common Requirements for Carrier-Grade 853 NATs (CGNs)", BCP 127, RFC 6888, DOI 10.17487/RFC6888, 854 April 2013, . 856 [RFC7857] Penno, R., Perreault, S., Boucadair, M., Ed., Sivakumar, 857 S., and K. Naito, "Updates to Network Address Translation 858 (NAT) Behavioral Requirements", BCP 127, RFC 7857, 859 DOI 10.17487/RFC7857, April 2016, 860 . 862 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 863 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 864 March 2017, . 866 [RFC5952] Kawamura, S. and M. Kawashima, "A Recommendation for IPv6 867 Address Text Representation", RFC 5952, 868 DOI 10.17487/RFC5952, August 2010, 869 . 871 7.2. Informative References 873 [I-D.ietf-dots-architecture] 874 Mortensen, A., Andreasen, F., Reddy, T., 875 christopher_gray3@cable.comcast.com, c., Compton, R., and 876 N. Teague, "Distributed-Denial-of-Service Open Threat 877 Signaling (DOTS) Architecture", draft-ietf-dots- 878 architecture-05 (work in progress), October 2017. 880 [I-D.ietf-dots-use-cases] 881 Dobbins, R., Migault, D., Fouant, S., Moskowitz, R., 882 Teague, N., Xia, L., and K. Nishizuka, "Use cases for DDoS 883 Open Threat Signaling", draft-ietf-dots-use-cases-09 (work 884 in progress), November 2017. 886 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 887 A., Peterson, J., Sparks, R., Handley, M., and E. 888 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 889 DOI 10.17487/RFC3261, June 2002, 890 . 892 [RFC7092] Kaplan, H. and V. Pascual, "A Taxonomy of Session 893 Initiation Protocol (SIP) Back-to-Back User Agents", 894 RFC 7092, DOI 10.17487/RFC7092, December 2013, 895 . 897 [RFC4732] Handley, M., Ed., Rescorla, E., Ed., and IAB, "Internet 898 Denial-of-Service Considerations", RFC 4732, 899 DOI 10.17487/RFC4732, December 2006, 900 . 902 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2", 903 FYI 36, RFC 4949, DOI 10.17487/RFC4949, August 2007, 904 . 906 Authors' Addresses 908 Andrew Mortensen 909 Arbor Networks 910 2727 S. State St 911 Ann Arbor, MI 48104 912 United States 914 Email: amortensen@arbor.net 916 Robert Moskowitz 917 Huawei 918 Oak Park, MI 42837 919 United States 921 Email: rgm@htt-consult.com 923 Tirumaleswar Reddy 924 McAfee, Inc. 925 Embassy Golf Link Business Park 926 Bangalore, Karnataka 560071 927 India 929 Email: TirumaleswarReddy_Konda@McAfee.com