Network Working Group                                        S. Bellovin
Request for Comments: 4107                           Columbia University
BCP: 107                                                      R. Housley
Category: Best Current Practice                           Vigil Security
                                                              June 2005


             Guidelines for Cryptographic Key Management

Status of This Memo

  This document specifies an Internet Best Current Practices for the
  Internet Community, and requests discussion and suggestions for
  improvements.  Distribution of this memo is unlimited.

Copyright Notice

  Copyright (C) The Internet Society (2005).

Abstract

  The question often arises of whether a given security system requires
  some form of automated key management, or whether manual keying is
  sufficient.  This memo provides guidelines for making such decisions.
  When symmetric cryptographic mechanisms are used in a protocol, the
  presumption is that automated key management is generally but not
  always needed.  If manual keying is proposed, the burden of proving
  that automated key management is not required falls to the proposer.

1.  Introduction

  The question often arises of whether or not a given security system
  requires some form of automated key management, or whether manual
  keying is sufficient.

  There is not one answer to that question; circumstances differ.  In
  general, automated key management SHOULD be used.  Occasionally,
  relying on manual key management is reasonable; we propose some
  guidelines for making that judgment.

  On the other hand, relying on manual key management has significant
  disadvantages, and we outline the security concerns that justify the
  preference for automated key management.  However, there are
  situations in which manual key management is acceptable.







Bellovin & Housley       Best Current Practice                  [Page 1]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


1.1.  Terminology

  The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD,
  SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this
  document, are to be interpreted as described in RFC 2119 [B].

2.  Guidelines

  These guidelines are for use by IETF working groups and protocol
  authors who are determining whether to mandate automated key
  management and whether manual key management is acceptable.  Informed
  judgment is needed.

  The term "key management" refers to the establishment of
  cryptographic keying material for use with a cryptographic algorithm
  to provide protocol security services, especially integrity,
  authentication, and confidentiality.  Automated key management
  derives one or more short-term session keys.  The key derivation
  function may make use of long-term keys to incorporate authentication
  into the process.  The manner in which this long-term key is
  distributed to the peers and the type of key used (pre-shared
  symmetric secret value, RSA public key, DSA public key, and others)
  is beyond the scope of this document.  However, it is part of the
  overall key management solution.  Manual key management is used to
  distribute such values.  Manual key management can also be used to
  distribute long-term session keys.

  Automated key management and manual key management provide very
  different features.  In particular, the protocol associated with an
  automated key management technique will confirm the liveness of the
  peer, protect against replay, authenticate the source of the short-
  term session key, associate protocol state information with the
  short-term session key, and ensure that a fresh short-term session
  key is generated.  Further, an automated key management protocol can
  improve interoperability by including negotiation mechanisms for
  cryptographic algorithms.  These valuable features are impossible or
  extremely cumbersome to accomplish with manual key management.

  For some symmetric cryptographic algorithms, implementations must
  prevent overuse of a given key.  An implementation of such algorithms
  can make use of automated key management when the usage limits are
  nearly exhausted, in order to establish replacement keys before the
  limits are reached, thereby maintaining secure communications.

  Examples of automated key management systems include IPsec IKE and
  Kerberos.  S/MIME and TLS also include automated key management
  functions.




Bellovin & Housley       Best Current Practice                  [Page 2]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


  Key management schemes should not be designed by amateurs; it is
  almost certainly inappropriate for working groups to design their
  own.  To put it in concrete terms, the very first key management
  protocol in the open literature was published in 1978 [NS].  A flaw
  and a fix were published in 1981 [DS], and the fix was cracked in
  1994 [AN].  In 1995 [L], a new flaw was found in the original 1978
  version, in an area not affected by the 1981/1994 issue.  All of
  these flaws were obvious once described -- yet no one spotted them
  earlier.  Note that the original protocol (translated to employ
  certificates, which had not been invented at that time) was only
  three messages.

  Key management software is not always large or bloated.  Even IKEv1
  [HC] can be done in less than 200 Kbytes of object code, and TLS [DA]
  in half that space.  Note that this TLS estimate includes other
  functionality as well.

  A session key is used to protect a payload.  The nature of the
  payload depends on the layer where the symmetric cryptography is
  applied.

  In general, automated key management SHOULD be used to establish
  session keys.  Strong justification is needed in the security
  considerations section of a proposal that makes use of manual key
  management.

2.1.  Automated Key Management

  Automated key management MUST be used if any of these conditions
  hold:

     A party will have to manage n^2 static keys, where n may become
     large.

     Any stream cipher (such as RC4 [TK], AES-CTR [NIST], or AES-CCM
     [WHF]) is used.

     An initialization vector (IV) might be reused, especially an
     implicit IV.  Note that random or pseudo-random explicit IVs are
     not a problem unless the probability of repetition is high.

     Large amounts of data might need to be encrypted in a short time,
     causing frequent change of the short-term session key.

     Long-term session keys are used by more than two parties.
     Multicast is a necessary exception, but multicast key management
     standards are emerging in order to avoid this in the future.
     Sharing long-term session keys should generally be discouraged.



Bellovin & Housley       Best Current Practice                  [Page 3]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


     The likely operational environment is one where personnel (or
     device) turnover is frequent, causing frequent change of the
     short-term session key.

2.2.  Manual Key Management

  Manual key management may be a reasonable approach in any of these
  situations:

     The environment has very limited available bandwidth or very high
     round-trip times.  Public key systems tend to require long
     messages and lots of computation; symmetric key alternatives, such
     as Kerberos, often require several round trips and interaction
     with third parties.

     The information being protected has low value.

     The total volume of traffic over the entire lifetime of the long-
     term session key will be very low.

     The scale of each deployment is very limited.

  Note that assertions about such things should often be viewed with
  skepticism.  The burden of demonstrating that manual key management
  is appropriate falls to the proponents -- and it is a fairly high
  hurdle.

  Systems that employ manual key management need provisions for key
  changes.  There MUST be some way to indicate which key is in use to
  avoid problems during transition.  Designs SHOULD sketch plausible
  mechanisms for deploying new keys and replacing old ones that might
  have been compromised.  If done well, such mechanisms can later be
  used by an add-on key management scheme.

  Lack of clarity about the parties involved in authentication is not a
  valid reason for avoiding key management.  Rather, it tends to
  indicate a deeper problem with the underlying security model.

2.3.  Key Size and Random Values

  Guidance on cryptographic key size for public keys that are used for
  exchanging symmetric keys can be found in BCP 86 [OH].

  When manual key management is used, long-term shared secret values
  SHOULD be at least 128 bits.

  Guidance on random number generation can be found in BCP 106 [ESC].




Bellovin & Housley       Best Current Practice                  [Page 4]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


  When manual key management is used, long-term shared secrets MUST be
  unpredictable "random" values, ensuring that an adversary will have
  no greater expectation than 50% of finding the value after searching
  half the key search space.

3.  Security Considerations

  This document provides guidance to working groups and protocol
  designers.  The security of the Internet is improved when automated
  key management is employed.

  The inclusion of automated key management does not mean that an
  interface for manual key management is prohibited.  In fact, manual
  key management is very helpful for debugging.  Therefore,
  implementations ought to provide a manual key management interface
  for such purposes, even if it is not specified by the protocol.

4.  References

  This section contains normative and informative references.

4.1.  Normative References

  [B]    Bradner, S., "Key words for use in RFCs to Indicate
         Requirement Levels", BCP 14, RFC 2119, March 1997.

  [ESC]  Eastlake, D., 3rd, Schiller, J., and S. Crocker, "Randomness
         Requirements for Security", BCP 106, RFC 4086, June 2005.

  [OH]   Orman, H. and P. Hoffman, "Determining Strengths For Public
         Keys Used For Exchanging Symmetric Keys", BCP 86, RFC 3766,
         April 2004

4.2.  Informative References

  [AN]   M. Abadi and R. Needham, "Prudent Engineering Practice for
         Cryptographic Protocols", Proc. IEEE Computer Society
         Symposium on Research in Security and Privacy, May 1994.

  [DA]   Dierks, T. and C. Allen, "The TLS Protocol Version 1.0", RFC
         2246, January 1999.

  [DS]   D. Denning and G. Sacco.  "Timestamps in key distributed
         protocols", Communication of the ACM, 24(8):533--535, 1981.

  [HC]   Harkins, D. and D. Carrel, "The Internet Key Exchange (IKE)",
         RFC 2409, November 1998.




Bellovin & Housley       Best Current Practice                  [Page 5]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


  [L]    G. Lowe.  "An attack on the Needham-Schroeder public key
         authentication protocol", Information Processing Letters,
         56(3):131--136, November 1995.

  [NIST] National Institute of Standards and Technology.
         "Recommendation for Block Cipher Modes of Operation -- Methods
         and Techniques," NIST Special Publication SP 800-38A, December
         2001.

  [NS]   R. Needham and M. Schroeder. "Using encryption for
         authentication in large networks of computers", Communications
         of the ACM, 21(12), December 1978.

  [TK]   Thayer, R. and K. Kaukonen.  "A Stream Cipher Encryption
         Algorithm", Work in Progress.

  [WHF]  Whiting, D., Housley, R., and N. Ferguson , "Counter with
         CBC-MAC (CCM)", RFC 3610, September 2003.

Authors' Addresses

  Steven M. Bellovin
  Department of Computer Science
  Columbia University
  1214 Amsterdam Avenue, M.C. 0401
  New York, NY 10027-7003

  Phone: +1 212-939-7149
  EMail: [email protected]


  Russell Housley
  Vigil Security, LLC
  918 Spring Knoll Drive
  Herndon, VA 20170

  Phone: +1 703-435-1775
  EMail: [email protected]













Bellovin & Housley       Best Current Practice                  [Page 6]

RFC 4107      Guidelines for Cryptographic Key Management      June 2005


Full Copyright Statement

  Copyright (C) The Internet Society (2005).

  This document is subject to the rights, licenses and restrictions
  contained in BCP 78, and except as set forth therein, the authors
  retain all their rights.

  This document and the information contained herein are provided on an
  "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
  OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
  ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED,
  INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE
  INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
  WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

  The IETF takes no position regarding the validity or scope of any
  Intellectual Property Rights or other rights that might be claimed to
  pertain to the implementation or use of the technology described in
  this document or the extent to which any license under such rights
  might or might not be available; nor does it represent that it has
  made any independent effort to identify any such rights.  Information
  on the procedures with respect to rights in RFC documents can be
  found in BCP 78 and BCP 79.

  Copies of IPR disclosures made to the IETF Secretariat and any
  assurances of licenses to be made available, or the result of an
  attempt made to obtain a general license or permission for the use of
  such proprietary rights by implementers or users of this
  specification can be obtained from the IETF on-line IPR repository at
  http://www.ietf.org/ipr.

  The IETF invites any interested party to bring to its attention any
  copyrights, patents or patent applications, or other proprietary
  rights that may cover technology that may be required to implement
  this standard.  Please address the information to the IETF at ietf-
  [email protected].

Acknowledgement

  Funding for the RFC Editor function is currently provided by the
  Internet Society.







Bellovin & Housley       Best Current Practice                  [Page 7]