idnits 2.17.1 draft-ietf-uta-tls-bcp-03.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 (September 21, 2014) is 3504 days in the past. Is this intentional? Checking references for intended status: Best Current Practice ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 2818 (Obsoleted by RFC 9110) ** Obsolete normative reference: RFC 4492 (Obsoleted by RFC 8422) ** Obsolete normative reference: RFC 5246 (Obsoleted by RFC 8446) ** Obsolete normative reference: RFC 6125 (Obsoleted by RFC 9525) ** Obsolete normative reference: RFC 6347 (Obsoleted by RFC 9147) == Outdated reference: A later version (-01) exists of draft-ietf-tls-prohibiting-rc4-00 == Outdated reference: A later version (-05) exists of draft-ietf-uta-tls-attacks-01 -- Obsolete informational reference (is this intentional?): RFC 2246 (Obsoleted by RFC 4346) -- Obsolete informational reference (is this intentional?): RFC 4346 (Obsoleted by RFC 5246) -- Obsolete informational reference (is this intentional?): RFC 4347 (Obsoleted by RFC 6347) -- Obsolete informational reference (is this intentional?): RFC 5077 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 6961 (Obsoleted by RFC 8446) Summary: 5 errors (**), 0 flaws (~~), 3 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 UTA Y. Sheffer 3 Internet-Draft Porticor 4 Intended status: Best Current Practice R. Holz 5 Expires: March 25, 2015 TUM 6 P. Saint-Andre 7 &yet 8 September 21, 2014 10 Recommendations for Secure Use of TLS and DTLS 11 draft-ietf-uta-tls-bcp-03 13 Abstract 15 Transport Layer Security (TLS) and Datagram Transport Security Layer 16 (DTLS) are widely used to protect data exchanged over application 17 protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the 18 last few years, several serious attacks on TLS have emerged, 19 including attacks on its most commonly used cipher suites and modes 20 of operation. This document provides recommendations for improving 21 the security of both software implementations and deployed services 22 that use TLS and DTLS. 24 Status of This Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at http://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on March 25, 2015. 41 Copyright Notice 43 Copyright (c) 2014 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (http://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 59 2. Intended Audience . . . . . . . . . . . . . . . . . . . . . . 4 60 2.1. Security Services . . . . . . . . . . . . . . . . . . . . 4 61 2.2. Examples . . . . . . . . . . . . . . . . . . . . . . . . 4 62 3. Conventions used in this document . . . . . . . . . . . . . . 4 63 4. General Recommendations . . . . . . . . . . . . . . . . . . . 5 64 4.1. Protocol Versions . . . . . . . . . . . . . . . . . . . . 5 65 4.2. Applicability to DTLS . . . . . . . . . . . . . . . . . . 5 66 4.3. Fallback to SSL . . . . . . . . . . . . . . . . . . . . . 6 67 4.4. Strict TLS . . . . . . . . . . . . . . . . . . . . . . . 6 68 4.5. Compression . . . . . . . . . . . . . . . . . . . . . . . 6 69 4.6. TLS Session Resumption . . . . . . . . . . . . . . . . . 7 70 4.7. TLS Renegotiation . . . . . . . . . . . . . . . . . . . . 7 71 4.8. Server Name Indication . . . . . . . . . . . . . . . . . 7 72 5. Recommendations: Cipher Suites . . . . . . . . . . . . . . . 7 73 5.1. General Guidelines . . . . . . . . . . . . . . . . . . . 8 74 5.2. Recommended Cipher Suites . . . . . . . . . . . . . . . . 9 75 5.3. Cipher Suite Negotiation Details . . . . . . . . . . . . 9 76 5.4. Public Key Length . . . . . . . . . . . . . . . . . . . . 10 77 5.5. Modular vs. Elliptic Curve DH Cipher Suites . . . . . . . 10 78 5.6. Truncated HMAC . . . . . . . . . . . . . . . . . . . . . 11 79 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11 80 7. Security Considerations . . . . . . . . . . . . . . . . . . . 11 81 7.1. Host Name Validation . . . . . . . . . . . . . . . . . . 11 82 7.2. AES-GCM . . . . . . . . . . . . . . . . . . . . . . . . . 12 83 7.3. Forward Secrecy . . . . . . . . . . . . . . . . . . . . . 12 84 7.4. Diffie Hellman Exponent Reuse . . . . . . . . . . . . . . 13 85 7.5. Certificate Revocation . . . . . . . . . . . . . . . . . 13 86 8. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 87 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 14 88 9.1. Normative References . . . . . . . . . . . . . . . . . . 14 89 9.2. Informative References . . . . . . . . . . . . . . . . . 15 90 Appendix A. Change Log . . . . . . . . . . . . . . . . . . . . . 17 91 A.1. draft-ietf-uta-tls-bcp-03 . . . . . . . . . . . . . . . . 17 92 A.2. draft-ietf-uta-tls-bcp-02 . . . . . . . . . . . . . . . . 17 93 A.3. draft-ietf-tls-bcp-01 . . . . . . . . . . . . . . . . . . 18 94 A.4. draft-ietf-tls-bcp-00 . . . . . . . . . . . . . . . . . . 18 95 A.5. draft-sheffer-tls-bcp-02 . . . . . . . . . . . . . . . . 18 96 A.6. draft-sheffer-tls-bcp-01 . . . . . . . . . . . . . . . . 18 97 A.7. draft-sheffer-tls-bcp-00 . . . . . . . . . . . . . . . . 19 98 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19 100 1. Introduction 102 Transport Layer Security (TLS) and Datagram Transport Security Layer 103 (DTLS) are widely used to protect data exchanged over application 104 protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the 105 last few years, several serious attacks on TLS have emerged, 106 including attacks on its most commonly used cipher suites and modes 107 of operation. For instance, both AES-CBC and RC4, which together 108 comprise most current usage, have been attacked in the context of 109 TLS. A companion document [I-D.ietf-uta-tls-attacks] provides 110 detailed information about these attacks. 112 Because of these attacks, those who implement and deploy TLS and DTLS 113 need updated guidance on how TLS can be used securely. Note that 114 this document provides guidance for deployed services, as well as 115 software implementations. In fact, this document calls for the 116 deployment of algorithms that are widely implemented but not yet 117 widely deployed. Concerning deployment, this document targets a wide 118 audience, namely all deployers who wish to add authentication, 119 confidentiality and data integrity to their communications. This 120 document does not address the rare deployment scenarios where one of 121 these three properties is not desired. 123 The recommendations herein take into consideration the security of 124 various mechanisms, their technical maturity and interoperability, 125 and their prevalence in implementations at the time of writing. 126 Unless noted otherwise, these recommendations apply to both TLS and 127 DTLS. TLS 1.3, when it is standardized and deployed in the field, 128 should resolve the current vulnerabilities while providing 129 significantly better functionality, and will very likely obsolete 130 this document. 132 These are minimum recommendations for the use of TLS for the 133 specified audience. Individual specifications may have stricter 134 requirements related to one or more aspects of the protocol, based on 135 their particular circumstances. When that is the case, implementers 136 MUST adhere to those stricter requirements. 138 Community knowledge about the strength of various algorithms and 139 feasible attacks can change quickly, and experience shows that a 140 security BCP is a point-in-time statement. Readers are advised to 141 seek out any errata or updates that apply to this document. 143 2. Intended Audience 145 In the following, we specify which audience this document addresses 146 concerning deployment. Most deployers are very likely part of this 147 audience, but very specialized use cases of TLS that are outside of 148 the intended audience can exist. 150 2.1. Security Services 152 This document provides recommendations for an audience that wishes to 153 secure their communication with TLS to achieve the following: 155 o Authentication: this means that an end-point of the TLS 156 communication is authenticated as the intended entity to 157 communicate with. TLS allows to authenticate one or both end- 158 points in the communication. 160 o Confidentiality: all (payload) communication is encrypted with the 161 goal that no party should be able to decrypt it except the 162 intended receiver. 164 o Data integrity: any changes made to the communication are 165 detectable by the receiver. 167 Deployers MUST verify that they do not need one of these three 168 properties if they deviate from the recommendations given in this 169 document. 171 2.2. Examples 173 The intended audience covers those services that are most commonly 174 used on the Internet, among many others: 176 o Operators of WWW servers (HTTPS). 178 o Operators of email servers (SMTPS, IMAPS, POPS). 180 o Operators of instant-messaging services (XMPPS, IRCS). 182 3. Conventions used in this document 184 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 185 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 186 document are to be interpreted as described in [RFC2119]. 188 4. General Recommendations 190 This section provides general recommendations on the secure use of 191 TLS. Recommendations related to cipher suites are discussed in the 192 following section. 194 4.1. Protocol Versions 196 It is important both to stop using old, less secure versions of SSL/ 197 TLS and to start using modern, more secure versions. Therefore: 199 o Implementations MUST NOT negotiate SSL version 2. 201 Rationale: SSLv2 is considered today as insecure [RFC6176]. 203 o Implementations MUST NOT negotiate SSL version 3. 205 Rationale: SSLv3 [RFC6101] was an improvement over SSLv2 and 206 plugged some significant security holes, but did not support 207 strong cipher suites. In addition, SSLv3 does not support TLS 208 extensions, some of which are considered security-critical today. 210 o Implementations SHOULD NOT negotiate TLS version 1.0 [RFC2246]. 212 Rationale: TLS 1.0 (published in 1999) does not support many 213 modern, strong cipher suites. 215 o Implementations MAY negotiate TLS version 1.1 [RFC4346]. 217 Rationale: TLS 1.1 (published in 2006) is a security improvement 218 over TLS 1.0, but still does not support certain stronger cipher 219 suites. 221 o Implementations MUST support, and prefer to negotiate, TLS version 222 1.2 [RFC5246]. 224 Rationale: Several stronger cipher suites are available only with 225 TLS 1.2 (published in 2008). 227 This BCP applies to TLS 1.2. It is not safe for readers to assume 228 that the recommendations in this BCP apply to any future version of 229 TLS. 231 4.2. Applicability to DTLS 233 DTLS [RFC4347] [RFC6347] is an adaptation of TLS for UDP datagrams. 235 With respect to the recommendations in the current document, DTLS 1.0 236 is equivalent to TLS 1.1. The only exception is RC4 which is 237 disallowed in DTLS. DTLS 1.2 is equivalent to TLS 1.2. 239 4.3. Fallback to SSL 241 Some client implementations revert to lower versions of TLS or even 242 to SSLv3 if the server rejected higher versions of the protocol. 243 This fall back can be forced by a man in the middle (MITM) attacker. 244 By default, such clients MUST NOT fall back to SSLv3. 246 Rationale: TLS 1.0 and SSLv3 are significantly less secure than TLS 247 1.2, the version recommended by this document. While TLS 1.0-only 248 servers are still quite common, IP scans show that SSLv3-only servers 249 amount to only about 3% of the current Web server population. 251 4.4. Strict TLS 253 Combining unprotected and TLS-protected communication opens the way 254 to SSL Stripping and similar attacks. Therefore: 256 o In cases where an application protocol allows implementations or 257 deployments a choice between strict TLS configuration and dynamic 258 upgrade from unencrypted to TLS-protected traffic (such as 259 STARTTLS), clients and servers SHOULD prefer strict TLS 260 configuration. 262 o Client and server implementations MUST support the HTTP Strict 263 Transport Security (HSTS) header [RFC6797], in order to allow Web 264 servers to advertise that they are willing to accept TLS-only 265 clients. 267 o When applicable, Web servers SHOULD use HSTS to indicate that they 268 are willing to accept TLS-only clients. 270 4.5. Compression 272 Implementations and deployments SHOULD disable TLS-level compression 273 ([RFC5246], Sec. 6.2.2), because it has been subject to security 274 attacks. 276 Implementers should note that compression at higher protocol levels 277 can allow an active attacker to extract cleartext information from 278 the connection. The BREACH attack is one such case. These issues 279 can only be mitigated outside of TLS and are thus out of scope of the 280 current document. See Sec. 2.5 of [I-D.ietf-uta-tls-attacks] for 281 further details. 283 4.6. TLS Session Resumption 285 If TLS session resumption is used, care ought to be taken to do so 286 safely. In particular, when using session tickets [RFC5077], the 287 resumption information MUST be authenticated and encrypted to prevent 288 modification or eavesdropping by an attacker. Further 289 recommendations apply to session tickets: 291 o A strong cipher suite MUST be used when encrypting the ticket (as 292 least as strong as the main TLS cipher suite). 294 o Ticket keys MUST be changed regularly, e.g. once every week, so as 295 not to negate the benefits of forward secrecy (see Section 7.3 for 296 details on forward secrecy). 298 o Session ticket validity SHOULD be limited to a reasonable duration 299 (e.g. 1 day), for similar reasons. 301 4.7. TLS Renegotiation 303 Where handshake renegotiation is implemented, both clients and 304 servers MUST implement the renegotiation_info extension, as defined 305 in [RFC5746]. 307 To counter the Triple Handshake attack, we adopt the recommendation 308 from [triple-handshake]: TLS clients SHOULD ensure that all 309 certificates received over a connection are valid for the current 310 server endpoint, and abort the handshake if they are not. In some 311 usages, it may be simplest to refuse any change of certificates 312 during renegotiation. 314 4.8. Server Name Indication 316 TLS implementations MUST support the Server Name Indication (SNI) 317 extension for those higher level protocols which would benefit from 318 it, including HTTPS. However, unlike implementation, the use of SNI 319 in particular circumstances is a matter of local policy. 321 5. Recommendations: Cipher Suites 323 TLS and its implementations provide considerable flexibility in the 324 selection of cipher suites. Unfortunately many available cipher 325 suites are insecure, and so misconfiguration can easily result in 326 reduced security. This section includes recommendations on the 327 selection and negotiation of cipher suites. 329 5.1. General Guidelines 331 It is important both to stop using old, insecure cipher suites and to 332 start using modern, more secure cipher suites. Therefore: 334 o Implementations MUST NOT negotiate the NULL cipher suites. 336 Rationale: The NULL cipher suites offer no encryption whatsoever 337 and thus are completely insecure. 339 o Implementations MUST NOT negotiate RC4 cipher suites 341 Rationale: The RC4 stream cipher has a variety of cryptographic 342 weaknesses, as documented in [I-D.ietf-tls-prohibiting-rc4]. 344 o Implementations MUST NOT negotiate cipher suites offering only so- 345 called "export-level" encryption (including algorithms with 40 346 bits or 56 bits of security). 348 Rationale: These cipher suites are deliberately "dumbed down" and 349 are very easy to break. 351 o Applications MUST NOT negotiate cipher suites of less than 112 352 bits of security. 354 o Implementations SHOULD NOT negotiate cipher suites that use 355 algorithms offering less than 128 bits of security. Note that 356 some legacy cipher suites (e.g. 168-bit 3DES) have an effective 357 key length which is smaller than their nominal key length (112 358 bits in the case of 3DES). Such cipher suites should be evaluated 359 according to their effective key length. 361 Rationale: Although these cipher suites are not actively subject 362 to breakage, their useful lifespan is short enough that stronger 363 cipher suites are desirable. 128-bit ciphers are expected to 364 remain secure for at least several years, and 256-bit ciphers 365 "until the next fundamental technology breakthrough". 367 o Implementations MUST support, and SHOULD prefer to negotiate, 368 cipher suites offering forward secrecy, such as those in the 369 Ephemeral Diffie-Hellman and Elliptic Curve Ephemeral Diffie 370 Hellman ("DHE" and "ECDHE") families. 372 Rationale: Forward secrecy (sometimes called "perfect forward 373 secrecy") prevents the recovery of information that was encrypted 374 with older session keys, thus limiting the amount of time during 375 which attacks can be successful. 377 5.2. Recommended Cipher Suites 379 Given the foregoing considerations, implementation of the following 380 cipher suites is RECOMMENDED: 382 o TLS_DHE_RSA_WITH_AES_128_GCM_SHA256 384 o TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 386 o TLS_DHE_RSA_WITH_AES_256_GCM_SHA384 388 o TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 390 We suggest that TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 be preferred in 391 general. See [RFC5289] for additional implementation details. 393 It is noted that those cipher suites are supported only in TLS 1.2 394 since they are authenticated encryption (AEAD) algorithms [RFC5116]. 396 [RFC4492] allows clients and servers to negotiate ECDH parameters 397 (curves). Both clients and servers SHOULD include the "Supported 398 Elliptic Curves" extension [RFC4492]. For interoperability, clients 399 and servers SHOULD support the NIST P-256 (secp256r1) curve 400 [RFC4492]. In addition, clients SHOULD send an ec_point_formats 401 extension with a single element, "uncompressed". 403 5.3. Cipher Suite Negotiation Details 405 Clients SHOULD include TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 as the 406 first proposal to any server, unless they have prior knowledge that 407 the server cannot respond to a TLS 1.2 client_hello message. 409 Servers SHOULD prefer this cipher suite whenever it is proposed, even 410 if it is not the first proposal. 412 Clients are of course free to offer stronger cipher suites, e.g. 413 using AES-256; when they do, the server SHOULD prefer the stronger 414 cipher suite unless there are compelling reasons (e.g., seriously 415 degraded performance) to choose otherwise. 417 Note that other profiles of TLS 1.2 exist that use different cipher 418 suites. For example, [RFC6460] defines a profile that uses the 419 TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256 and 420 TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384 cipher suites. 422 This document is not an application profile standard, in the sense of 423 Sec. 9 of [RFC5246]. As a result, clients and servers are still 424 REQUIRED to support the mandatory TLS cipher suite, 425 TLS_RSA_WITH_AES_128_CBC_SHA. 427 5.4. Public Key Length 429 With a key exchange based on modular Diffie-Hellman ("DHE" cipher 430 suites), key lengths of at least 2048 bits are RECOMMENDED. 432 Rationale: because Diffie-Hellman keys of 1024 bits are estimated to 433 be roughly equivalent to 80-bit symmetric keys, it is better to use 434 longer keys for the "DHE" family of cipher suites. Unfortunately, 435 some existing software cannot handle (or cannot easily handle) key 436 lengths greater than 1024 bits. The most common workaround for these 437 systems is to prefer the "ECDHE" family of cipher suites instead of 438 the "DHE" family. For modular groups, key lengths of at least 2048 439 bits are estimated to be roughly equivalent to 112-bit symmetric keys 440 and might be sufficient for at least the next 10 years. 442 Servers SHOULD authenticate using 2048-bit certificates. In 443 addition, the use of SHA-256 fingerprints is RECOMMENDED (see 444 [CAB-Baseline] for more details). Clients SHOULD indicate to servers 445 that they request SHA-256, by using the "Signature Algorithms" 446 extension defined in TLS 1.2. 448 5.5. Modular vs. Elliptic Curve DH Cipher Suites 450 Not all TLS implementations support both modular and EC Diffie- 451 Hellman groups, as required by Section 5.2. Some implementations are 452 severely limited in the length of DH values. When such 453 implementations need to be accommodated, we recommend using (in 454 priority order): 456 1. Elliptic Curve DHE with negotiated parameters [RFC5289] 458 2. TLS_DHE_RSA_WITH_AES_128_GCM_SHA256 [RFC5288], with 2048-bit 459 Diffie-Hellman parameters 461 3. The same cipher suite, with 1024-bit parameters. 463 Rationale: Elliptic Curve Cryptography is not universally deployed 464 for several reasons, including its complexity compared to modular 465 arithmetic and longstanding IPR concerns. On the other hand, there 466 are two related issues hindering effective use of modular Diffie- 467 Hellman cipher suites in TLS: 469 o There are no protocol mechanisms to negotiate the DH groups or 470 parameter lengths supported by client and server. 472 o There are widely deployed client implementations that reject 473 received DH parameters if they are longer than 1024 bits. 475 We note that with DHE and ECDHE cipher suites, the TLS master key 476 only depends on the Diffie Hellman parameters and not on the strength 477 of the RSA certificate; moreover, 1024 bit modular DH parameters are 478 generally considered insufficient at this time. 480 With modular ephemeral DH, deployers SHOULD carefully evaluate 481 interoperability vs. security considerations when configuring their 482 TLS endpoints. 484 5.6. Truncated HMAC 486 The truncated HMAC extension, defined in Sec. 7 of [RFC6066] does not 487 apply to the AEAD cipher suites recommended above. However it does 488 apply to most other TLS cipher suites. Its use has been shown to be 489 insecure in [PatersonRS11], and implementations MUST NOT use it. 491 6. IANA Considerations 493 This document requests no actions of IANA. [Note to RFC Editor: 494 please remove this whole section before publication.] 496 7. Security Considerations 498 This entire document discusses the security practices directly 499 affecting applications using the TLS protocol. This section contains 500 broader security considerations related to technologies used in 501 conjunction with or by TLS. 503 7.1. Host Name Validation 505 Application authors should take note that TLS implementations 506 frequently do not validate host names, and must therefore determine 507 if the TLS implementation they are using does, and if not write their 508 own validation code or consider changing the TLS implementation. 510 It is noted that the requirements regarding host name validation (and 511 in general, binding between the TLS layer and the protocol that runs 512 above it) vary between different protocols. For HTTPS, these 513 requirements are defined by Sec. 3 of [RFC2818]. 515 Readers are referred to [RFC6125] for further details regarding 516 generic host name validation in the TLS context. In addition, the 517 RFC contains a long list of example protocols, some of which 518 implement a policy very different from HTTPS. 520 With some protocols, the host name is obtained indirectly and in an 521 insecure manner, e.g. by an insecure DNS query for an MX record. In 522 these cases, the host name SHOULD NOT be used as a trusted identity 523 even when it matches the presented certificate. 525 7.2. AES-GCM 527 Sec. Section 5.2 above recommends the use of the AES-GCM 528 authenticated encryption algorithm. Please refer to [RFC5246], Sec. 529 11 for general security considerations when using TLS 1.2, and to 530 [RFC5288], Sec. 6 for security considerations that apply specifically 531 to AES-GCM when used with TLS. 533 7.3. Forward Secrecy 535 Forward secrecy (also often called Perfect Forward Secrecy or "PFS", 536 and defined in [RFC4949]) is a defense against an attacker who 537 records encrypted conversations where the session keys are only 538 encrypted with the communicating parties' long-term keys. Should the 539 attacker be able to obtain these long-term keys at some point later 540 in time, he will be able to decrypt the session keys and thus the 541 entire conversation. In the context of TLS and DTLS, such compromise 542 of long-term keys is not entirely implausible. It can happen, for 543 example, due to: 545 o A client or server being attacked by some other attack vector, and 546 the private key retrieved. 548 o A long-term key retrieved from a device that has been sold or 549 otherwise decommissioned without prior wiping. 551 o A long-term key used on a device as a default key [Heninger2012]. 553 o A key generated by a Trusted Third Party like a CA, and later 554 retrieved from it either by extortion or compromise 555 [Soghoian2011]. 557 o A cryptographic break-through, or the use of asymmetric keys with 558 insufficient length [Kleinjung2010]. 560 PFS ensures in such cases that the session keys cannot be determined 561 even by an attacker who obtains the long-term keys some time after 562 the conversation. It also protects against an attacker who is in 563 possession of the long-term keys, but remains passive during the 564 conversation. 566 PFS is generally achieved by using the Diffie-Hellman scheme to 567 derive session keys. The Diffie-Hellman scheme has both parties 568 maintain private secrets and send parameters over the network as 569 modular powers over certain cyclic groups. The properties of the so- 570 called Discrete Logarithm Problem (DLP) allow to derive the session 571 keys without an eavesdropper being able to do so. There is currently 572 no known attack against DLP if sufficiently large parameters are 573 chosen. A variant of the Diffie-Hellman scheme uses Elliptic Curves 574 instead of the originally proposed modular arithmetics. 576 Unfortunately, many TLS/DTLS cipher suites were defined that do not 577 feature PFS, e.g. TLS_RSA_WITH_AES_256_CBC_SHA256. We thus advocate 578 strict use of PFS-only ciphers. 580 7.4. Diffie Hellman Exponent Reuse 582 For performance reasons, many TLS implementations reuse Diffie- 583 Hellman and Elliptic Curve Diffie-Hellman exponents across multiple 584 connections. Such reuse can result in major security issues: 586 o If exponents are reused for a long time (e.g., more than a few 587 hours), an attacker who gains access to the host can decrypt 588 previous connections. In other words, exponent reuse negates the 589 effects of forward secrecy. 591 o TLS implementations that reuse exponents should test the DH public 592 key they receive, in order to avoid some known attacks. These 593 tests are not standardized in TLS at the time of writing. See 594 [RFC6989] for recipient tests required of IKEv2 implementations 595 that reuse DH exponents. 597 7.5. Certificate Revocation 599 Unfortunately there is currently no effective, Internet-scale 600 mechanism to affect certificate revocation: 602 o Certificate Revocation Lists (CRLs) are non-scalable and therefore 603 rarely used. 605 o The On-Line Certification Status Protocol (OCSP) presents both 606 scaling and privacy issues when used for heavy traffic Web 607 servers. In addition, clients typically "soft-fail", meaning they 608 do not abort the TLS connection if the OCSP server does not 609 respond. 611 o OCSP stapling (Sec. 8 of [RFC6066]) resolves the operational 612 issues with OCSP, but is still ineffective in the presence of a 613 MITM attacker because they can simply ignore the client's request 614 for a stapled OCSP response. 616 o OCSP stapling as defined in [RFC6066] does not extend to 617 intermediate certificates used in a certificate chain. [RFC6961] 618 addresses this shortcoming, but is a recent addition without much 619 deployment. 621 o Proprietary mechanisms that embed revocation lists in the Web 622 browser's configuration database cannot scale beyond a small 623 number of the most heavily used Web servers. 625 The current consensus appears to be that OCSP stapling, combined with 626 a "must staple" mechanism similar to HSTS, would finally resolve this 627 problem; in particular when used together with the extension defined 628 in [RFC6961]. But such a mechanism has not been standardized yet. 630 8. Acknowledgments 632 We would like to thank Viktor Dukhovni, Stephen Farrell, Simon 633 Josefsson, Watson Ladd, Orit Levin, Johannes Merkle, Bodo Moeller, 634 Yoav Nir, Kenny Paterson, Patrick Pelletier, Tom Ritter, Rich Salz, 635 Aaron Zauner for their review and improvements. Thanks to Brian 636 Smith whose "browser cipher suites" page is a great resource. 637 Finally, thanks to all others who commented on the TLS, UTA and other 638 lists and are not mentioned here by name. 640 9. References 642 9.1. Normative References 644 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 645 Requirement Levels", BCP 14, RFC 2119, March 1997. 647 [RFC2818] Rescorla, E., "HTTP Over TLS", RFC 2818, May 2000. 649 [RFC4492] Blake-Wilson, S., Bolyard, N., Gupta, V., Hawk, C., and B. 650 Moeller, "Elliptic Curve Cryptography (ECC) Cipher Suites 651 for Transport Layer Security (TLS)", RFC 4492, May 2006. 653 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 654 (TLS) Protocol Version 1.2", RFC 5246, August 2008. 656 [RFC5288] Salowey, J., Choudhury, A., and D. McGrew, "AES Galois 657 Counter Mode (GCM) Cipher Suites for TLS", RFC 5288, 658 August 2008. 660 [RFC5289] Rescorla, E., "TLS Elliptic Curve Cipher Suites with 661 SHA-256/384 and AES Galois Counter Mode (GCM)", RFC 5289, 662 August 2008. 664 [RFC5746] Rescorla, E., Ray, M., Dispensa, S., and N. Oskov, 665 "Transport Layer Security (TLS) Renegotiation Indication 666 Extension", RFC 5746, February 2010. 668 [RFC6125] Saint-Andre, P. and J. Hodges, "Representation and 669 Verification of Domain-Based Application Service Identity 670 within Internet Public Key Infrastructure Using X.509 671 (PKIX) Certificates in the Context of Transport Layer 672 Security (TLS)", RFC 6125, March 2011. 674 [RFC6176] Turner, S. and T. Polk, "Prohibiting Secure Sockets Layer 675 (SSL) Version 2.0", RFC 6176, March 2011. 677 [RFC6347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 678 Security Version 1.2", RFC 6347, January 2012. 680 9.2. Informative References 682 [CAB-Baseline] 683 CA/Browser Forum, , "Baseline Requirements for the 684 Issuance and Management of Publicly-Trusted Certificates 685 Version 1.1.6", 2013, . 688 [Heninger2012] 689 Heninger, N., Durumeric, Z., Wustrow, E., and J. 690 Halderman, "Mining Your Ps and Qs: Detection of Widespread 691 Weak Keys in Network Devices", Usenix Security Symposium 692 2012, 2012. 694 [I-D.ietf-tls-prohibiting-rc4] 695 Popov, A., "Prohibiting RC4 Cipher Suites", draft-ietf- 696 tls-prohibiting-rc4-00 (work in progress), July 2014. 698 [I-D.ietf-uta-tls-attacks] 699 Sheffer, Y., Holz, R., and P. Saint-Andre, "Summarizing 700 Current Attacks on TLS and DTLS", draft-ietf-uta-tls- 701 attacks-01 (work in progress), June 2014. 703 [Kleinjung2010] 704 Kleinjung, T., "Factorization of a 768-Bit RSA Modulus", 705 CRYPTO 10, 2010, . 707 [PatersonRS11] 708 Paterson, K., Ristenpart, T., and T. Shrimpton, "Tag size 709 does matter: attacks and proofs for the TLS record 710 protocol", 2011, 711 . 713 [RFC2246] Dierks, T. and C. Allen, "The TLS Protocol Version 1.0", 714 RFC 2246, January 1999. 716 [RFC4346] Dierks, T. and E. Rescorla, "The Transport Layer Security 717 (TLS) Protocol Version 1.1", RFC 4346, April 2006. 719 [RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 720 Security", RFC 4347, April 2006. 722 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2", RFC 723 4949, August 2007. 725 [RFC5077] Salowey, J., Zhou, H., Eronen, P., and H. Tschofenig, 726 "Transport Layer Security (TLS) Session Resumption without 727 Server-Side State", RFC 5077, January 2008. 729 [RFC5116] McGrew, D., "An Interface and Algorithms for Authenticated 730 Encryption", RFC 5116, January 2008. 732 [RFC6066] Eastlake, D., "Transport Layer Security (TLS) Extensions: 733 Extension Definitions", RFC 6066, January 2011. 735 [RFC6101] Freier, A., Karlton, P., and P. Kocher, "The Secure 736 Sockets Layer (SSL) Protocol Version 3.0", RFC 6101, 737 August 2011. 739 [RFC6460] Salter, M. and R. Housley, "Suite B Profile for Transport 740 Layer Security (TLS)", RFC 6460, January 2012. 742 [RFC6797] Hodges, J., Jackson, C., and A. Barth, "HTTP Strict 743 Transport Security (HSTS)", RFC 6797, November 2012. 745 [RFC6961] Pettersen, Y., "The Transport Layer Security (TLS) 746 Multiple Certificate Status Request Extension", RFC 6961, 747 June 2013. 749 [RFC6989] Sheffer, Y. and S. Fluhrer, "Additional Diffie-Hellman 750 Tests for the Internet Key Exchange Protocol Version 2 751 (IKEv2)", RFC 6989, July 2013. 753 [Soghoian2011] 754 Soghoian, C. and S. Stamm, "Certified lies: Detecting and 755 defeating government interception attacks against SSL.", 756 Proc. 15th Int. Conf. Financial Cryptography and Data 757 Security , 2011. 759 [triple-handshake] 760 Delignat-Lavaud, A., Bhargavan, K., and A. Pironti, 761 "Triple Handshakes Considered Harmful: Breaking and Fixing 762 Authentication over TLS", 2014, . 765 Appendix A. Change Log 767 Note to RFC Editor: please remove this section before publication. 769 A.1. draft-ietf-uta-tls-bcp-03 771 o Disallow truncated HMAC. 773 o Applicability to DTLS. 775 o Some more text restructuring. 777 o Host name validation is sometimes irrelevant. 779 o HSTS: MUST implement, SHOULD deploy. 781 o Session identities are not protected, only tickets are. 783 o Clarified the target audience. 785 A.2. draft-ietf-uta-tls-bcp-02 787 o Rearranged some sections for clarity and re-styled the text so 788 that normative text is followed by rationale where possible. 790 o Removed the recommendation to use Brainpool curves. 792 o Triple Handshake mitigation. 794 o MUST NOT negotiate algorithms lower than 112 bits of security. 796 o MUST implement SNI, but use per local policy. 798 o Changed SHOULD NOT negotiate or fall back to SSLv3 to MUST NOT. 800 o Added hostname validation. 802 o Non-normative discussion of DH exponent reuse. 804 A.3. draft-ietf-tls-bcp-01 806 o Clarified that specific TLS-using protocols may have stricter 807 requirements. 809 o Changed TLS 1.0 from MAY to SHOULD NOT. 811 o Added discussion of "optional TLS" and HSTS. 813 o Recommended use of the Signature Algorithm and Renegotiation Info 814 extensions. 816 o Use of a strong cipher for a resumption ticket: changed SHOULD to 817 MUST. 819 o Added an informational discussion of certificate revocation, but 820 no recommendations. 822 A.4. draft-ietf-tls-bcp-00 824 o Initial WG version, with only updated references. 826 A.5. draft-sheffer-tls-bcp-02 828 o Reorganized the content to focus on recommendations. 830 o Moved description of attacks to a separate document (draft- 831 sheffer-uta-tls-attacks). 833 o Strengthened recommendations regarding session resumption. 835 A.6. draft-sheffer-tls-bcp-01 837 o Clarified our motivation in the introduction. 839 o Added a section justifying the need for PFS. 841 o Added recommendations for RSA and DH parameter lengths. Moved 842 from DHE to ECDHE, with a discussion on whether/when DHE is 843 appropriate. 845 o Recommendation to avoid fallback to SSLv3. 847 o Initial information about browser support - more still needed! 849 o More clarity on compression. 851 o Client can offer stronger cipher suites. 853 o Discussion of the regular TLS mandatory cipher suite. 855 A.7. draft-sheffer-tls-bcp-00 857 o Initial version. 859 Authors' Addresses 861 Yaron Sheffer 862 Porticor 863 29 HaHarash St. 864 Hod HaSharon 4501303 865 Israel 867 Email: yaronf.ietf@gmail.com 869 Ralph Holz 870 Technische Universitaet Muenchen 871 Boltzmannstr. 3 872 Garching 85748 873 Germany 875 Email: holz@net.in.tum.de 877 Peter Saint-Andre 878 &yet 879 P.O. Box 787 880 Parker, CO 80134 881 USA 883 Email: peter@andyet.com