RFC 6421

Crypto-Agility Requirements for Remote Authentication Dial-In User Service (RADIUS)

📅 published November 2011, from ietf

RFC 6421: Crypto-Agility Requirements for Remote Authentication Dial-In User Service (RADIUS) 💖

okay so bestie, this one is an Informational RFC from November 2011 🎀 It was written by D. Nelson, Ed. from Elbrys Networks, Inc., it's published by the IETF stream, and it has ISSN 2070-1721. The title block doesn't list anything it obsoletes or updates, so it just stands on its own ✨

Abstract 🌸

This memo lays out the requirements for a crypto-agility solution for Remote Authentication Dial-In User Service (RADIUS). main character requirements energy 💅

Status of This Memo 🦋

This doc is not an Internet Standards Track spec. It's published for informational purposes, so no pressure 🙈

This doc is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community, and it got public review and was approved for publication by the Internet Engineering Steering Group (IESG). Not every doc the IESG approves is a candidate for any level of Internet Standard, so see Section 2 of RFC 5741 for the details 🧚

For info on the current status of this doc, any errata, and how to give feedback, check http://www.rfc-editor.org/info/rfc6421 ✨

Copyright (c) 2011 IETF Trust and the persons identified as the document authors. All rights reserved.

This doc is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) as they stand on the date this doc was published. Please read these carefully, because they describe your rights and restrictions for this doc. Code Components pulled from this doc must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions, and they're provided without warranty as described in the Simplified BSD License 🫶

This doc may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The people who control the copyright in some of that material may not have given the IETF Trust the right to allow modifications outside the IETF Standards Process. So unless you get an adequate license from those copyright holders, this doc may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process. The exception is formatting it for publication as an RFC or translating it into languages other than English 🌈

Table of Contents 🧾

   1. Introduction ....................................................2
      1.1. General ....................................................2
      1.2. Requirements Language ......................................3
      1.3. Publication Process ........................................3
   2. A Working Definition of Crypto-Agility ..........................4
   3. The Current State of RADIUS Security ............................5
   4. The Requirements ................................................5
      4.1. Overall Solution Approach ..................................5
      4.2. Security Services ..........................................6
      4.3. Backwards Compatibility ....................................7
      4.4. Interoperability and Change Control ........................9
      4.5. Scope of Work ..............................................9
      4.6. Applicability of Automated Key Management Requirements .....9
   5. Security Considerations ........................................10
   6. Acknowledgments ................................................10
   7. References .....................................................10
      7.1. Normative References ......................................10
      7.2. Informative References ....................................11

1. Introduction 🎀

1.1. General

okay so at IETF 66, the RADIUS Extensions (RADEXT) Working Group (WG) got asked by people in the Security Area Directorate to write up a formal description of a crypto-agility work item and the matching charter milestones 💅 After chatting with one of the Security Area Directors (Russ Housley), text was first proposed on the RADEXT WG mailing list on October 26, 2006. Here's the gist of that proposal:

The RADEXT WG will review the security requirements for crypto- agility in IETF protocols, and identify the deficiencies of the existing RADIUS protocol specifications against these requirements. Specific attention will be paid to RFC 4962 [RFC4962]. The RADEXT WG will propose one or more specifications to remediate any identified deficiencies in the crypto-agility properties of the RADIUS protocol. The known deficiencies include the issue of negotiation of substitute algorithms for the message digest functions, the key-wrap functions, and the password-hiding function. Additionally, at least one mandatory to implement cryptographic algorithm will be defined in each of these areas, as required.

This doc describes the features, properties, and limitations of RADIUS crypto-agility solutions, defines the term "crypto-agility" as it's used in this context, and explains why this work matters 🦋

The requirements in this memo were built from email messages posted to the RADEXT WG mailing list, which you can find in that list's archives. The whole point of framing the requirements here is to formalize them, archive them for future reference, and bring them explicitly to the attention of the IESG and the IETF community as the work moves forward ✨

1.2. Requirements Language

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].

A RADIUS crypto-agility solution is not compliant with this specification if it fails to satisfy one or more of the MUST or MUST NOT statements. A solution that satisfies all the MUST, MUST NOT, SHOULD, and SHOULD NOT statements is said to be "unconditionally compliant". One that satisfies all the MUST and MUST NOT statements but not all the SHOULD or SHOULD NOT requirements is said to be "conditionally compliant" 🫶

1.3. Publication Process

RADIUS [RFC2865] is a widely deployed protocol. It reached Draft Standard status based on multiple independent interoperable implementations, which is a whole vibe 👑 So it's desirable that a high level of interoperability gets maintained for crypto-agility solutions too.

To make sure crypto-agility solutions published on the standards track are well specified and interoperable, the RADEXT WG adopted a two phase process for standards-track publication of them 🌸

In the initial phase, crypto-agility solutions adopted by the working group will be published as Experimental. These documents should contain a description of the implementations and experimental deployments in progress, plus an evaluation of the proposal against the requirements described in this document.

Then the working group will select proposals to advance on the standards track. The criteria include evaluation of the proposal against the requirements, a summary of the experimental deployment experience, and evidence of multiple interoperable implementations 🔥

2. A Working Definition of Crypto-Agility 🧚

Crypto-agility is the ability of a protocol to adapt to evolving cryptography and security requirements. It may include a modular mechanism that lets cryptographic algorithms get updated without substantial disruption to fielded implementations. It may also provide for the dynamic negotiation and installation of cryptographic algorithms within protocol implementations (think of Dynamic-Link Libraries (DLL)) 💖

In the specific context of the RADIUS protocol and RADIUS implementations, crypto-agility may be better defined as the ability of RADIUS implementations to automatically negotiate cryptographic algorithms for use in RADIUS exchanges. That includes the algorithms used to integrity protect and authenticate RADIUS packets and to hide RADIUS attributes. This capability covers all RADIUS message types: Access-Request/Response, Accounting-Request/Response, CoA/Disconnect-Request/Response, and Status-Server. Negotiation of cryptographic algorithms MAY occur within the RADIUS protocol, or within a lower layer such as the transport layer 🦋

Proposals MUST NOT introduce generic new capability negotiation features into the RADIUS protocol, or require changes to the RADIUS operational model as defined in "RADIUS Design Guidelines" [RFC6158], Section 3.1 and Appendix A.4. A proposal SHOULD focus on the crypto-agility problem and nothing else. For example, proposals SHOULD NOT require new attribute formats, and they SHOULD be compatible with the guidance provided in [RFC6158], Section 2.3. Issues of backward compatibility are described in more detail in Section 4.3 🙈

3. The Current State of RADIUS Security 🔥

RADIUS packets, as defined in [RFC2865], are protected by an MD5 message integrity check (MIC) within the Authenticator field of RADIUS packets other than Access-Request [RFC2865] and Status-Server [RFC5997]. The Message-Authenticator Attribute uses HMAC-MD5 to authenticate and integrity protect RADIUS packets 🎀

RADIUS doesn't support confidentiality for entire packets, but various RADIUS attributes support encrypted (also known as "hidden") values. That includes User-Password (defined in [RFC2865], Section 5.2), Tunnel-Password (defined in [RFC2868], Section 3.5), and various Vendor-Specific Attributes, such as the MS-MPPE-Send-Key and MS-MPPE-Recv-Key attributes (defined in [RFC2548], Section 2.4). Generally speaking, the hiding mechanism uses a stream cipher based on a key stream from an MD5 digest. Attacks against this mechanism are described in "RADIUS Support for EAP" [RFC3579], Section 4.3.4 🧾

"Updated Security Considerations for the MD5 Message-Digest and the HMAC-MD5 Algorithms" [RFC6151] discusses security considerations for using MD5 and HMAC-MD5. Advances in MD5 collisions don't immediately compromise MD5 or HMAC-MD5 for the purposes RADIUS uses them for, as long as the RADIUS shared secret stays unknown. But the progress toward compromising MD5's basic cryptographic assumptions has led to MD5 being deprecated in a variety of applications 😭 As noted in [RFC6151], Section 2:

MD5 is no longer acceptable where collision resistance is required such as digital signatures. It is not urgent to stop using MD5 in other ways, such as HMAC-MD5; however, since MD5 must not be used for digital signatures, new protocol designs should not employ HMAC-MD5.

4. The Requirements 💅

4.1. Overall Solution Approach

RADIUS crypto-agility solutions aren't restricted to technology described in existing RFCs. RADIUS over IPsec is already described in Section 5 of "RADIUS and IPv6" [RFC3162] and Section 4.2 of [RFC3579], so that technique is already available to anyone who wants to use it. Therefore, it is expected that proposals will utilize other techniques 🧚✨

4.2. Security Services 🔐✨

Okay bestie, this is the security core 💅. Proposals MUST support negotiating cryptographic algorithms for per-packet integrity/authentication protection, and they MUST also support per-packet replay protection for all RADIUS message types. Crypto-agility solutions MUST specify mandatory-to-implement cryptographic algorithms for each defined mechanism. No skipping that part 🙅‍♀️.

Crypto-agility solutions MUST avoid security compromise, even when the existing algorithms RADIUS implementations use are shown to be weak enough to give little or no security (for example, if the legacy RADIUS shared secret gets compromised 😭). That includes protection against bidding-down attacks 🚫⬇️. When you're analyzing how resilient a crypto-agility solution is, you can assume that RADIUS requesters and responders can be configured to require new secure algorithms once existing algorithms or the legacy RADIUS shared secret are compromised. Main character energy for security 👑.

Guidance on acceptable algorithms is in [NIST-SP800-131A]. It is RECOMMENDED that mandatory-to-implement cryptographic algorithms come from the ones classified as "Acceptable" with no known deprecation date from within this or successor documents 🌸.

It is RECOMMENDED that solutions support confidentiality, either by encrypting entire RADIUS packets or by encrypting individual RADIUS attributes 🎀. Proposals that support confidentiality MUST support negotiating cryptographic algorithms for encryption.

Encrypting individual RADIUS attributes is OPTIONAL for solutions that already encrypt entire RADIUS packets. But solutions that do encrypt individual attributes are REQUIRED to improve the confidentiality of existing encrypted (sometimes called "hidden") attributes, AND to encrypt attributes that are currently sent in cleartext, like location attributes 📍🙈.

Besides the goals above, [RFC4962] Section 3 lays out additional security requirements. For RADIUS crypto-agility solutions, those translate into the following:

Strong, fresh session keys: RADIUS crypto-agility solutions are REQUIRED to generate fresh session keys for use between the RADIUS client and server 🔑. So one leaked session key can't help an attacker figure out the others, RADIUS crypto-agility solutions are RECOMMENDED to support Perfect Forward Secrecy (PFS) for session keys negotiated between the RADIUS client and server 🧊✨.

Limit key scope: So a Network Access Server (NAS) and a RADIUS server can swap sensitive stuff like keying material without third parties seeing it, it is RECOMMENDED that a RADIUS crypto-agility solution support X.509 certificates for authentication between the NAS and RADIUS server 📜. Manual configuration or automated discovery mechanisms, like NAI-based Dynamic Peer Discovery [RADYN], can be used to set up direct NAS-RADIUS server communication 🤝. Support for end-to-end confidentiality of RADIUS attributes is OPTIONAL.

For compatibility with how things already work, RADIUS crypto-agility solutions SHOULD also support pre-shared key credentials 🗝️. Support for direct NAS-to-RADIUS server communication is OPTIONAL when pre-shared key credentials are used.

4.3. Backwards Compatibility 🔁🫶

Solutions MUST show backward compatibility with existing RADIUS implementations. In other words, an implementation that supports both crypto-agility and legacy mechanisms MUST be able to talk to legacy RADIUS clients and servers, using the legacy mechanisms 🤝.

Backward compatibility makes the switch from legacy RADIUS to crypto-agile RADIUS way smoother 🧈. But legacy mechanisms are only appropriate before they've been compromised. Once they're compromised, secure algorithms MUST be used, so backward compatibility is no longer possible 💀.

Here's the tricky part: RADIUS is a request/response protocol, so negotiating algorithms within a single exchange is inherently limited 😩. Before a response arrives, a requester doesn't know what algorithms the responder supports. So even though a RADIUS request can include a list of supported algorithms that can be picked for the response, the algorithms protecting the initial request have to be decided ahead of time 📋.

To let a request be handled by both legacy and crypto-agile implementations, a request can be secured with legacy algorithms as well as with attributes that provide security services using more secure algorithms ✨. This lets legacy implementations and crypto-agile implementations both process the packet, and it doesn't add extra response delays 🚀. If you use this technique, the credentials used with legacy algorithms MUST be cryptographically independent of the credentials used with the more secure algorithms. That way, if the legacy credentials get compromised, the credentials for the secure algorithms are safe 🛡️.

In this approach, legacy mechanisms are used at first in requests sent between crypto-agile implementations. But if the responder shows it supports crypto-agility, future requests can use more secure mechanisms 🦋. Heads up though: if a responder gets upgraded and then has to be downgraded (say, because of bugs 🐛), requesters might not be able to talk to the downgraded responder, unless there's a way to configure the requester to turn legacy algorithms back on.

Probing techniques can avoid using legacy algorithms in requests between crypto-agile implementations 🔍. For example, an initial request can leave out legacy mechanisms. If a response comes back, the recipient can be assumed to be crypto-agile, and future requests to that recipient can use secure mechanisms. Similarly, the responder can assume the requester supports crypto-agility, and it can prohibit legacy mechanisms in future requests. Same warning applies, though: if a requester gets upgraded and then has to be downgraded (for bugs, say 🐛), the requester might not be able to interpret responses, unless there's a way to configure the responder to turn legacy algorithms back on.

If no response comes back, and there's no information showing that the responder supports crypto-agility (like pre-configuration or having previously received a crypto-agile response), a new request can be built using legacy mechanisms 🔄.

Here's another catch 😤: legacy implementations that don't support crypto-agility will silently discard requests that aren't protected by legacy algorithms, instead of sending back an error. So repeated requests might be needed to tell "no crypto-agility support" apart from packet loss or other failures 🙃. That means probing techniques can delay the first communication between crypto-agile requesters and legacy responders. This can be dealt with by upgrading the responders (like RADIUS servers) first 🥰.

4.4. Interoperability and Change Control 🤝📜

Proposals MUST indicate they're willing to cede change control to the IETF 🫡.

Crypto-agility solutions MUST be interoperable between independent implementations, based purely on the information in the specification. If implementers need anything beyond the spec to make it work together, that's a fail 🙅‍♀️.

4.5. Scope of Work 🗺️

Crypto-agility solutions MUST apply to all RADIUS packet types, including Access-Request, Access-Challenge, Access-Reject, Access-Accept, Accounting-Request, Accounting-Response, Status-Server and CoA/Disconnect messages 📦.

Since the work is expected to happen purely within RADIUS or in the transport, message data exchanged with Diameter SHOULD NOT be affected 🦋.

Proposals MUST discuss any inherent assumptions about, or limitations on, client/server operations or deployment. They SHOULD also give recommendations for moving deployments from legacy RADIUS to crypto-agile RADIUS 🚚. Cipher-suite negotiation, legacy interoperability, and the potential for bidding-down attacks SHOULD be among those discussions 💭.

4.6. Applicability of Automated Key Management Requirements 🗝️🤖

"Guidelines for Cryptographic Key Management" [RFC4107] gives guidance on when automated key management is necessary. People wondered whether RFC 4107 would require a RADIUS crypto-agility solution to include Automated Key Management (AKM) 🤔. The answer was no, AKM isn't inherently required for RADIUS, for these reasons:

  • RFC 4107 requires AKM for protocols that involve O(n^2) keys. That doesn't apply to RADIUS deployments, which need O(n) keys 📈.
  • Session key freshness requirements can be met without AKM, for example by using a pre-shared key along with an exchange of nonces 🎲.
  • RADIUS doesn't need to encrypt large amounts of data in a short time 📉.
  • Organizations already have operational practices for managing existing RADIUS shared secrets, covering key changes that come from personnel changes 🧑‍💻.
  • The crypto-agility solution can avoid cryptographic modes of operation, like a counter mode cipher, that need frequent key changes 🙅‍♀️.

BUT, at the same time, features recommended in Section 4.2, like perfect forward secrecy and direct transport of keys between a NAS and RADIUS server, can only be provided by a solution that supports AKM 😤. So support for Automated Key Management is RECOMMENDED within a RADIUS crypto-agility solution 💖.

Also, automated key management is REQUIRED for RADIUS crypto-agility solutions that use cryptographic modes of operation needing frequent key changes 🚨.

5. Security Considerations 🚨🔐

Potential attacks against the RADIUS protocol are described in [RFC3579], Section 4.1 ⚠️. Details of known exploits, plus potential mitigations, are discussed in [RFC3579], Section 4.3 🛠️.

This specification describes requirements for new cryptographic protection mechanisms, including modular selection of algorithms and modes. So, honestly, all the subject matter of this memo is security-related 💀🔥. Take it seriously, besties.

6. Acknowledgments 🫶🌈

Huge thanks to all the reviewers and contributors, including Bernard Aboba, Mary Barnes, Pasi Eronen, Dan Romascanu, Joe Salowey, and Glen Zorn 👑💜.

7. References 📚

7.1. Normative References

okay bestie, these are the docs you need to read before anything else 📚✨ they're the foundation, no cap

  • [NIST-SP800-131A] Barker, E. and A. Roginsky, "Transitions: Recommendation for Transitioning the Use of Cryptographic Algorithms and Key Lengths", NIST SP-800-131A, January 2011.
  • [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
  • [RFC2865] Rigney, C., Willens, S., Rubens, A., and W. Simpson, "Remote Authentication Dial In User Service (RADIUS)", RFC 2865, June 2000.
  • [RFC4107] Bellovin, S. and R. Housley, "Guidelines for Cryptographic Key Management", BCP 107, RFC 4107, June 2005.
  • [RFC4962] Housley, R. and B. Aboba, "Guidance for Authentication, Authorization, and Accounting (AAA) Key Management", BCP 132, RFC 4962, July 2007.
  • [RFC6151] Turner, S. and L. Chen, "Updated Security Considerations for the MD5 Message-Digest and the HMAC-MD5 Algorithms", RFC 6151, March 2011.
  • [RFC6158] DeKok, A., Ed., and G. Weber, "RADIUS Design Guidelines", BCP 158, RFC 6158, March 2011.

7.2. Informative References

these ones are extra credit 🌸 super useful for context, but you don't have to read them to get the main ideas

  • [RADYN] Winter, S. and M. McCauley, "NAI-based Dynamic Peer Discovery for RADIUS/TLS and RADIUS/DTLS", Work in Progress, July 2011.
  • [RFC2548] Zorn, G., "Microsoft Vendor-specific RADIUS Attributes", RFC 2548, March 1999.
  • [RFC2868] Zorn, G., Leifer, D., Rubens, A., Shriver, J., Holdrege, M., and I. Goyret, "RADIUS Attributes for Tunnel Protocol Support", RFC 2868, June 2000.
  • [RFC3162] Aboba, B., Zorn, G., and D. Mitton, "RADIUS and IPv6", RFC 3162, August 2001.
  • [RFC3579] Aboba, B. and P. Calhoun, "RADIUS (Remote Authentication Dial In User Service) Support For Extensible Authentication Protocol (EAP)", RFC 3579, September 2003.
  • [RFC5997] DeKok, A., "Use of Status-Server Packets in the Remote Authentication Dial In User Service (RADIUS) Protocol", RFC 5997, August 2010.

Author's Address

the editor's contact info, in case you need to reach out 💌🎀

David B. Nelson (editor)
Elbrys Networks, Inc.
282 Corporate Drive, Unit 1
Portsmouth, NH  03801
USA

EMail: d.b.nelson@comcast.net
📦 where this page comes from