| UTA | Y. Sheffer |
| Internet-Draft | Porticor |
| Intended status: Best Current Practice | R. Holz |
| Expires: May 15, 2015 | TUM |
| P. Saint-Andre | |
| &yet | |
| November 11, 2014 |
Recommendations for Secure Use of TLS and DTLS
draft-ietf-uta-tls-bcp-07
Transport Layer Security (TLS) and Datagram Transport Layer Security (DTLS) are widely used to protect data exchanged over application protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the last few years, several serious attacks on TLS have emerged, including attacks on its most commonly used cipher suites and modes of operation. This document provides recommendations for improving the security of deployed services that use TLS and DTLS. The recommendations are applicable to the majority of use cases.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on May 15, 2015.
Copyright (c) 2014 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Transport Layer Security (TLS) [RFC5246] and Datagram Transport Security Layer (DTLS) [RFC6347] are widely used to protect data exchanged over application protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the last few years, several serious attacks on TLS have emerged, including attacks on its most commonly used cipher suites and modes of operation. For instance, both the AES-CBC [RFC3602] and RC4 [I-D.ietf-tls-prohibiting-rc4] encryption algorithms, which together are the most widely deployed ciphers, have been attacked in the context of TLS. A companion document [I-D.ietf-uta-tls-attacks] provides detailed information about these attacks.
Because of these attacks, those who implement and deploy TLS and DTLS need updated guidance on how TLS can be used securely. This document provides guidance for deployed services as well as for software implementations, assuming the implementer expects his or her code to be deployed in environments defined in the following section. In fact, this document calls for the deployment of algorithms that are widely implemented but not yet widely deployed. Concerning deployment, this document targets a wide audience, namely all deployers who wish to add authentication (be it one-way only or mutual), confidentiality, and data integrity protection to their communications.
The recommendations herein take into consideration the security of various mechanisms, their technical maturity and interoperability, and their prevalence in implementations at the time of writing. Unless it is explicitly called out that a recommendation applies to TLS alone or to DTLS alone, each recommendation applies to both TLS and DTLS.
It is expected that the TLS 1.3 specification will resolve many of the vulnerabilities listed in this document. A system that deploys TLS 1.3 will have fewer vulnerabilities than TLS 1.2 or below. This document is likely to be updated after TLS 1.3 gets noticeable deployment.
These are minimum recommendations for the use of TLS in the vast majority of implementation and deployment scenarios, with the exception of unauthenticated TLS (see Section 5). Other specifications that reference this document can have stricter requirements related to one or more aspects of the protocol, based on their particular circumstances (e.g., for use with a particular application protocol); when that is the case, implementers are advised to adhere to those stricter requirements.
Community knowledge about the strength of various algorithms and feasible attacks can change quickly, and experience shows that a security BCP is a point-in-time statement. Readers are advised to seek out any errata or updates that apply to this document.
A number of security-related terms in this document are used in the sense defined in [RFC4949].
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].
This section provides general recommendations on the secure use of TLS. Recommendations related to cipher suites are discussed in the following section.
It is important both to stop using old, less secure versions of SSL/TLS and to start using modern, more secure versions; therefore, the following are the recommendations concerning TLS/SSL protocol versions:
This BCP applies to TLS 1.2. It is not safe for readers to assume that the recommendations in this BCP apply to any future version of TLS.
DTLS, an adaptation of TLS for UDP datagrams, was introduced when TLS 1.1 was published. The following are the recommendations with respect to DTLS:
Clients that “fallback" to lower versions of the protocol after the server rejects higher versions of the protocol MUST NOT fallback to SSLv3.
Rationale: Some client implementations revert to lower versions of TLS or even to SSLv3 if the server rejected higher versions of the protocol. This fallback can be forced by a man in the middle (MITM) attacker. TLS 1.0 and SSLv3 are significantly less secure than TLS 1.2, the version recommended by this document. While TLS 1.0-only servers are still quite common, IP scans show that SSLv3-only servers amount to only about 3% of the current Web server population.
To prevent SSL Stripping:
Rationale: Combining unprotected and TLS-protected communication opens the way to SSL Stripping and similar attacks, since an initial part of the communication is not integrity protected and therefore can be manipulated by an attacker whose goal is to keep the communication in the clear.
Implementations and deployments SHOULD disable TLS-level compression ([RFC5246], Section 6.2.2).
Rationale: TLS compression has been subject to security attacks, such as the CRIME attack.
Implementers should note that compression at higher protocol levels can allow an active attacker to extract cleartext information from the connection. The BREACH attack is one such case. These issues can only be mitigated outside of TLS and are thus out of scope of the current document. See Section 2.5 of [I-D.ietf-uta-tls-attacks] for further details.
If TLS session resumption is used, care ought to be taken to do so safely. In particular, when using session tickets [RFC5077], the resumption information MUST be authenticated and encrypted to prevent modification or eavesdropping by an attacker. Further recommendations apply to session tickets:
Rationale: session resumption is another kind of TLS handshake, and therefore must be as secure as the initial handshake. This document (Section 4) recommends the use of cipher suites that provide forward secrecy, i.e. that prevent an attacker who gains momentary access to the TLS endpoint (either client or server) and its secrets from reading either past or future communication. The tickets must be managed so as not to negate this security property.
Where handshake renegotiation is implemented, both clients and servers MUST implement the renegotiation_info extension, as defined in [RFC5746].
To counter the Triple Handshake attack, we adopt the recommendation from [triple-handshake]: TLS clients SHOULD ensure that all certificates received over a connection are valid for the current server endpoint, and abort the handshake if they are not. In some usages, it may be simplest to refuse any change of certificates during renegotiation.
TLS implementations MUST support the Server Name Indication (SNI) extension for those higher level protocols which would benefit from it, including HTTPS. However, unlike implementation, the use of SNI in particular circumstances is a matter of local policy.
Rationale: SNI supports deployment of multiple TLS-protected virtual servers on a single address, and therefore enables fine-grained security for these virtual servers, by allowing each one to have its own certificate.
TLS and its implementations provide considerable flexibility in the selection of cipher suites. Unfortunately, some available cipher suites are insecure, some do not provide the targeted security services, and some no longer provide enough security. Incorrectly configuring a server leads to no or reduced security. This section includes recommendations on the selection and negotiation of cipher suites.
Cryptographic algorithms weaken over time as cryptanalysis improves. In other words, as time progresses, algorithms that were once considered strong but are now weak, need to be phased out over time and replaced with more secure cipher suites to ensure that desired security properties still hold. SSL/TLS has been in existence for almost 20 years at this point and this section provides some much needed recommendations concerning cipher suite selection:
Given the foregoing considerations, implementation and deployment of the following cipher suites is RECOMMENDED:
These cipher suites are supported only in TLS 1.2 since they are authenticated encryption (AEAD) algorithms [RFC5116].
Typically, in order to prefer these suites, the order of suites needs to be explicitly configured in server software.
Clients SHOULD include TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 as the first proposal to any server, unless they have prior knowledge that the server cannot respond to a TLS 1.2 client_hello message.
Servers SHOULD prefer this cipher suite whenever it is proposed, even if it is not the first proposal.
Clients are of course free to offer stronger cipher suites, e.g., using AES-256; when they do, the server SHOULD prefer the stronger cipher suite unless there are compelling reasons (e.g., seriously degraded performance) to choose otherwise.
This document is not an application profile standard, in the sense of Section 9 of [RFC5246]. As a result, clients and servers are still REQUIRED to support the mandatory TLS cipher suite, TLS_RSA_WITH_AES_128_CBC_SHA.
Note that some profiles of TLS 1.2 use different cipher suites. For example, [RFC6460] defines a profile that uses the TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256 and TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384 cipher suites.
[RFC4492] allows clients and servers to negotiate ECDH parameters (curves). Both clients and servers SHOULD include the "Supported Elliptic Curves" extension [RFC4492]. For interoperability, clients and servers SHOULD support the NIST P-256 (secp256r1) curve [RFC4492]. In addition, clients SHOULD send an ec_point_formats extension with a single element, "uncompressed".
When using the cipher suites recommended in this document, two public keys are normally used in the TLS handshake: one for the Diffie-Hellman key agreement and one for server authentication. Where a client certificate is used, a third public key is added.
With a key exchange based on modular Diffie-Hellman ("DHE" cipher suites), DH key lengths of at least 2048 bits are RECOMMENDED.
Rationale: For various reasons, in practice DH keys are typically generated in lengths that are powers of two (e.g., 2^10 = 1024 bits, 2^11 = 2048 bits, 2^12 = 4096 bits). Because a DH key of 1228 bits would be roughly equivalent to only an 80-bit symmetric key [RFC3766], it is better to use keys longer than that for the "DHE" family of cipher suites. A DH key of 1926 bits would be roughly equivalent to a 100-bit symmetric key [RFC3766] and a DH key of 2048 bits might be sufficient for at least the next 10 years. See Section 4.4 for additional information on the use of modular Diffie-Hellman in TLS.
As noted in [RFC3766], correcting for the emergence of a TWIRL machine would imply that 1024-bit DH keys yield about 65 bits of equivalent strength and that a 2048-bit DH key would yield about 92 bits of equivalent strength.
Servers SHOULD authenticate using at least 2048-bit certificates. In addition, the use of SHA-256 fingerprints is RECOMMENDED (see [CAB-Baseline] for more details). Clients SHOULD indicate to servers that they request SHA-256, by using the "Signature Algorithms" extension defined in TLS 1.2.
Not all TLS implementations support both modular and EC Diffie-Hellman groups, as required by Section 4.2. Some implementations are severely limited in the length of DH values. When such implementations need to be accommodated, we recommend using (in priority order):
Rationale: Elliptic Curve Cryptography is not universally deployed for several reasons, including its complexity compared to modular arithmetic and longstanding perceptions of IPR concerns (which, for the most part, have now been resolved [RFC6090]). On the other hand, there are two related issues hindering effective use of modular Diffie-Hellman cipher suites in TLS:
We note that with DHE and ECDHE cipher suites, the TLS master key only depends on the Diffie-Hellman parameters and not on the strength of the RSA certificate; moreover, 1024 bit modular DH parameters are generally considered insufficient at this time.
With modular ephemeral DH, deployers SHOULD carefully evaluate interoperability vs. security considerations when configuring their TLS endpoints.
Implementations MUST NOT use the Truncated HMAC extension, defined in Section 7 of [RFC6066].
Rationale: the extension does not apply to the AEAD cipher suites recommended above. However it does apply to most other TLS cipher suites. Its use has been shown to be insecure in [PatersonRS11].
The deployment recommendations of this document address the operators of application layer services that are most commonly used on the Internet, including, but not limited to:
This document provides recommendations for an audience that wishes to secure their communication with TLS to achieve the following:
With regard to authentication, TLS enables authentication of one or both end-points in the communication. Although some TLS usage scenarios do not require authentication, those scenarios are not in scope for this document (a rationale for this decision is provided under Section 5.2).
If deployers deviate from the recommendations given in this document, they MUST verify that they do not need one of the foregoing security services.
This document applies only to environments where confidentiality is required. It recommends algorithms and configuration options that enforce secrecy of the data-in-transit.
This document also assumes that data integrity protection is always one of the goals of a deployment. In cases where integrity is not required, it does not make sense to employ TLS in the first place. There are attacks against confidentiality-only protection that utilize the lack of integrity to also break confidentiality (see for instance [DegabrieleP07] in the context of IPsec).
The intended audience covers those services that are most commonly used on the Internet. Typically, all communication between TLS clients and TLS servers requires all three of the above security services. This is particularly true where TLS clients are user agents like Web browsers or email software.
This document does not address the rarer deployment scenarios where one of the above three properties is not desired, such as the use case described under Section 5.2 below. Another example of an audience not needing confidentiality is the following: a monitored network where the authorities in charge of the respective traffic domain require full access to unencrypted (plaintext) traffic, and where users collaborate and send their traffic in the clear.
Several important applications use TLS to protect data between a TLS client and a TLS server, but do so without the TLS client necessarily verifying the server's certificate. This practice is often called "unauthenticated TLS". The reader is referred to [I-D.ietf-dane-smtp-with-dane] for an example and an explanation of why this less secure practice will likely remain common in the context of SMTP (especially for MTA-to-MTA communications). The practice is also encountered in similar contexts such as server-to-server traffic on the XMPP network (where multi-tenant hosting environments make it difficult for operators to obtain proper certificates for all of the domains they service).
Furthermore, in some scenarios the use of TLS itself is optional, i.e. the client decides dynamically ("opportunistically") whether to use TLS with a particular server or to connect in the clear. This practice, often called "opportunistic encryption", and is described at length in Section 2 of [I-D.farrelll-mpls-opportunistic-encrypt].
It can be argued that the recommendations provided in this document ought to apply equally to unauthenticated TLS as well as authenticated TLS. That would keep TLS implementations and deployments in sync, which is a desirable property given that servers can be used simultaneously for unauthenticated TLS and for authenticated TLS (indeed, a server cannot know whether a client might attempt authenticated or unauthenticated TLS). On the other hand, it has been argued that some of the recommendations in this document might be too strict for unauthenticated scenarios and that any security is better than no security at all (i.e., sending traffic in the clear), even if it means deploying outdated protocol versions and ciphers in unauthenticated scenarios. The sense of the UTA Working Group was to complete work on this document about authenticated TLS and to initiate work on a separate document about unauthenticated TLS.
In summary: this document does not apply to unauthenticated TLS use cases.
This document requests no actions of IANA. [Note to RFC Editor: please remove this whole section before publication.]
This entire document discusses the security practices directly affecting applications using the TLS protocol. This section contains broader security considerations related to technologies used in conjunction with or by TLS.
Application authors should take note that TLS implementations frequently do not validate host names and must therefore determine if the TLS implementation they are using does and, if not, write their own validation code or consider changing the TLS implementation.
It is noted that the requirements regarding host name validation (and in general, binding between the TLS layer and the protocol that runs above it) vary between different protocols. For HTTPS, these requirements are defined by Section 3 of [RFC2818].
Readers are referred to [RFC6125] for further details regarding generic host name validation in the TLS context. In addition, the RFC contains a long list of example protocols, some of which implement a policy very different from HTTPS.
If the host name is discovered indirectly and in an insecure manner (e.g., by an insecure DNS query for an MX or SRV record), it SHOULD NOT be used as a reference identifier [RFC6125] even when it matches the presented certificate. This proviso does not apply if the host name is discovered securely (for further discussion, see for example [I-D.ietf-dane-srv] and [I-D.ietf-dane-smtp-with-dane]).
Host name validation typically applies only to the leaf "end entity" certificate. Naturally, in order to ensure proper authentication in the context of the PKI, application clients need to verify the entire certification path in accordance with [RFC5280] (see also [RFC6125]).
Section 4.2 above recommends the use of the AES-GCM authenticated encryption algorithm. Please refer to [RFC5246], Section 11 for general security considerations when using TLS 1.2, and to [RFC5288], Section 6 for security considerations that apply specifically to AES-GCM when used with TLS.
Forward secrecy (also often called Perfect Forward Secrecy or "PFS" and defined in [RFC4949]) is a defense against an attacker who records encrypted conversations where the session keys are only encrypted with the communicating parties' long-term keys. Should the attacker be able to obtain these long-term keys at some point later in time, he will be able to decrypt the session keys and thus the entire conversation. In the context of TLS and DTLS, such compromise of long-term keys is not entirely implausible. It can happen, for example, due to:
Forward secrecy ensures in such cases that the session keys cannot be determined even by an attacker who obtains the long-term keys some time after the conversation. It also protects against an attacker who is in possession of the long-term keys, but remains passive during the conversation.
Forward secrecy is generally achieved by using the Diffie-Hellman scheme to derive session keys. The Diffie-Hellman scheme has both parties maintain private secrets and send parameters over the network as modular powers over certain cyclic groups. The properties of the so-called Discrete Logarithm Problem (DLP) allow to derive the session keys without an eavesdropper being able to do so. There is currently no known attack against DLP if sufficiently large parameters are chosen. A variant of the Diffie-Hellman scheme uses Elliptic Curves instead of the originally proposed modular arithmetics.
Unfortunately, many TLS/DTLS cipher suites were defined that do not feature forward secrecy, e.g., TLS_RSA_WITH_AES_256_CBC_SHA256. We thus advocate strict use of forward-secrecy-only ciphers.
For performance reasons, many TLS implementations reuse Diffie-Hellman and Elliptic Curve Diffie-Hellman exponents across multiple connections. Such reuse can result in major security issues:
Unfortunately, no mechanism exists at this time that we can recommend as a complete and efficient solution for the problem of checking the revocation status of common public key certificates (a.k.a. PKIX certificates, [RFC5280]). The current state of the art is as follows:
With regard to PKIX certificates, servers SHOULD support OCSP and OCSP stapling, including the OCSP stapling extension defined in [RFC6961], as a best practice given the current state of the art and as a foundation for a possible future solution.
The foregoing considerations do not apply to scenarios where the DANE-TLSA resource record [RFC6698] is used to signal to a client which certificate a server considers valid and good to use for TLS connections.
We would like to thank Uri Blumenthal, Viktor Dukhovni, Stephen Farrell, Paul Hoffman, Simon Josefsson, Watson Ladd, Orit Levin, Ilari Liusvaara, Johannes Merkle, Bodo Moeller, Yoav Nir, Kenny Paterson, Patrick Pelletier, Tom Ritter, Rich Salz, Sean Turner, and Aaron Zauner for their feedback and suggested improvements. Thanks to Brian Smith, whose "browser cipher suites" page is a great resource. Finally, thanks to all others who commented on the TLS, UTA, and other discussion lists but who are not mentioned here by name.
Note to RFC Editor: please remove this section before publication.