The Use of HMAC-MD5-96 within ESP and AH
How IPsec ESP and AH use HMAC with MD5, cut to 96 bits, to prove packets are genuine and unaltered.
Madson, C. and R. Glenn, "The Use of HMAC-MD5-96 within ESP and AH", RFC 2403, DOI 10.17487/RFC2403, November 1998, <https://www.rfc-editor.org/info/rfc2403>.
@misc{rfc2403,
series = {Request for Comments},
number = 2403,
howpublished = {RFC 2403},
publisher = {RFC Editor},
doi = {10.17487/RFC2403},
url = {https://www.rfc-editor.org/info/rfc2403},
author = {C. Madson and R. Glenn},
title = {{The Use of HMAC-MD5-96 within ESP and AH}},
pagetotal = 7,
year = 1998,
month = nov,
} RFC text
Contents
Network Working Group C. Madson
Request for Comments: 2403 Cisco Systems Inc.
Category: Standards Track R. Glenn
NIST
November 1998
The Use of HMAC-MD5-96 within ESP and AH
Status of this Memo
This document specifies an Internet standards track protocol for the
Internet community, and requests discussion and suggestions for
improvements. Please refer to the current edition of the "Internet
Official Protocol Standards" (STD 1) for the standardization state
and status of this protocol. Distribution of this memo is unlimited.
Copyright Notice
Copyright (C) The Internet Society (1998). All Rights Reserved.
Abstract
This memo describes the use of the HMAC algorithm [RFC-2104] in
conjunction with the MD5 algorithm [RFC-1321] as an authentication
mechanism within the revised IPSEC Encapsulating Security Payload
[ESP] and the revised IPSEC Authentication Header [AH]. HMAC with MD5
provides data origin authentication and integrity protection.
Further information on the other components necessary for ESP and AH
implementations is provided by [Thayer97a].
1. Introduction
This memo specifies the use of MD5 [RFC-1321] combined with HMAC
[RFC-2104] as a keyed authentication mechanism within the context of
the Encapsulating Security Payload and the Authentication Header.
The goal of HMAC-MD5-96 is to ensure that the packet is authentic and
cannot be modified in transit.
HMAC is a secret key authentication algorithm. Data integrity and
data origin authentication as provided by HMAC are dependent upon the
scope of the distribution of the secret key. If only the source and
destination know the HMAC key, this provides both data origin
authentication and data integrity for packets sent between the two
parties; if the HMAC is correct, this proves that it must have been
added by the source.
In this memo, HMAC-MD5-96 is used within the context of ESP and AH.
For further information on how the various pieces of ESP - including
the confidentiality mechanism -- fit together to provide security
services, refer to [ESP] and [Thayer97a]. For further information on
AH, refer to [AH] and [Thayer97a].
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 [RFC-2119].
2. Algorithm and Mode
[RFC-1321] describes the underlying MD5 algorithm, while [RFC-2104]
describes the HMAC algorithm. The HMAC algorithm provides a framework
for inserting various hashing algorithms such as MD5.
HMAC-MD5-96 operates on 64-byte blocks of data. Padding requirements
are specified in [RFC-1321] and are part of the MD5 algorithm. If
MD5 is built according to [RFC-1321], there is no need to add any
additional padding as far as HMAC-MD5-96 is concerned. With regard
to "implicit packet padding" as defined in [AH], no implicit packet
padding is required.
HMAC-MD5-96 produces a 128-bit authenticator value. This 128-bit
value can be truncated as described in RFC 2104. For use with either
ESP or AH, a truncated value using the first 96 bits MUST be
supported. Upon sending, the truncated value is stored within the
authenticator field. Upon receipt, the entire 128-bit value is
computed and the first 96 bits are compared to the value stored in
the authenticator field. No other authenticator value lengths are
supported by HMAC-MD5-96.
The length of 96 bits was selected because it is the default
authenticator length as specified in [AH] and meets the security
requirements described in [RFC-2104].
2.1 Performance
[Bellare96a] states that "(HMAC) performance is essentially that of
the underlying hash function". [RFC-1810] provides some performance
analysis and recommendations of the use of MD5 with Internet
protocols. As of this writing no performance analysis has been done
of HMAC or HMAC combined with MD5.
[RFC-2104] outlines an implementation modification which can improve
per-packet performance without affecting interoperability.
3. Keying Material
HMAC-MD5-96 is a secret key algorithm. While no fixed key length is
specified in [RFC-2104], for use with either ESP or AH a fixed key
length of 128-bits MUST be supported. Key lengths other than 128-
bits MUST NOT be supported (i.e. only 128-bit keys are to be used by
HMAC-MD5-96). A key length of 128-bits was chosen based on the
recommendations in [RFC-2104] (i.e. key lengths less than the
authenticator length decrease security strength and keys longer than
the authenticator length do not significantly increase security
strength).
[RFC-2104] discusses requirements for key material, which includes a
discussion on requirements for strong randomness. A strong pseudo-
random function MUST be used to generate the required 128-bit key.
At the time of this writing there are no specified weak keys for use
with HMAC. This does not mean to imply that weak keys do not exist.
If, at some point, a set of weak keys for HMAC are identified, the
use of these weak keys must be rejected followed by a request for
replacement keys or a newly negotiated Security Association.
[ARCH] describes the general mechanism for obtaining keying material
when multiple keys are required for a single SA (e.g. when an ESP SA
requires a key for confidentiality and a key for authentication).
In order to provide data origin authentication, the key distribution
mechanism must ensure that unique keys are allocated and that they
are distributed only to the parties participating in the
communication.
[RFC-2104] makes the following recommendation with regard to
rekeying. Current attacks do not indicate a specific recommended
frequency for key changes as these attacks are practically
infeasible. However, periodic key refreshment is a fundamental
security practice that helps against potential weaknesses of the
function and keys, reduces the information avaliable to a
cryptanalyst, and limits the damage of an exposed key.
4. Interaction with the ESP Cipher Mechanism
As of this writing, there are no known issues which preclude the use
of the HMAC-MD5-96 algorithm with any specific cipher algorithm.
5. Security Considerations
The security provided by HMAC-MD5-96 is based upon the strength of
HMAC, and to a lesser degree, the strength of MD5. [RFC-2104] claims
that HMAC does not depend upon the property of strong collision
resistance, which is important to consider when evaluating the use of
MD5, an algorithm which has, under recent scrutiny, been shown to be
much less collision-resistant than was first thought. At the time of
this writing there are no practical cryptographic attacks against
HMAC-MD5-96.
[RFC-2104] states that for "minimally reasonable hash functions" the
"birthday attack", the strongest attack know against HMAC, is
impractical. For a 64-byte block hash such as HMAC-MD5-96, an attack
involving the successful processing of 2**64 blocks would be
infeasible unless it were discovered that the underlying hash had
collisions after processing 2**30 blocks. A hash with such weak
collision-resistance characteristics would generally be considered to
be unusable.
It is also important to consider that while MD5 was never developed
to be used as a keyed hash algorithm, HMAC had that criteria from the
onset. While the use of MD5 in the context of data security is
undergoing reevaluation, the combined HMAC with MD5 algorithm has
held up to cryptographic scrutiny.
[RFC-2104] also discusses the potential additional security which is
provided by the truncation of the resulting hash. Specifications
which include HMAC are strongly encouraged to perform this hash
truncation.
As [RFC-2104] provides a framework for incorporating various hash
algorithms with HMAC, it is possible to replace MD5 with other
algorithms such as SHA-1. [RFC-2104] contains a detailed discussion
on the strengths and weaknesses of HMAC algorithms.
As is true with any cryptographic algorithm, part of its strength
lies in the correctness of the algorithm implementation, the security
of the key management mechanism and its implementation, the strength
of the associated secret key, and upon the correctness of the
implementation in all of the participating systems. [RFC-2202]
contains test vectors and example code to assist in verifying the
correctness of HMAC-MD5-96 code.
6. Acknowledgments
This document is derived in part from previous works by Jim Hughes,
those people that worked with Jim on the combined DES/CBC+HMAC-MD5
ESP transforms, the ANX bakeoff participants, and the members of the
IPsec working group.
We would also like to thank Hugo Krawczyk for his comments and
recommendations regarding some of the cryptographic specific text in
this document.
7. References
[RFC-1321] Rivest, R., "MD5 Digest Algorithm", RFC 1321, April
1992.
[RFC-2104] Krawczyk, H., Bellare, M., and R. Canetti, "HMAC:
Keyed-Hashing for Message Authentication", RFC 2104,
February 1997.
[RFC-1810] Touch, J., "Report on MD5 Performance", RFC 1810, June
1995.
[Bellare96a] Bellare, M., Canetti, R., and H. Krawczyk, "Keying Hash
Functions for Message Authentication", Advances in
Cryptography, Crypto96 Proceeding, June 1996.
[ARCH] Kent, S., and R. Atkinson, "Security Architecture for
the Internet Protocol", RFC 2401, November 1998.
[ESP] Kent, S., and R. Atkinson, "IP Encapsulating Security
Payload", RFC 2406, November 1998.
[AH] Kent, S., and R. Atkinson, "IP Authentication Header",
RFC 2402, November 1998.
[Thayer97a] Thayer, R., Doraswamy, N., and R. Glenn, "IP Security
Document Roadmap", RFC 2411, November 1998.
[RFC-2202] Cheng, P., and R. Glenn, "Test Cases for HMAC-MD5 and
HMAC-SHA-1", RFC 2202, March 1997.
[RFC-2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
8. Editors' Address
Cheryl Madson
Cisco Systems, Inc.
EMail: cmadson@cisco.com
Rob Glenn
NIST
EMail: <rob.glenn@nist.gov>
The IPsec working group can be contacted through the chairs:
Robert Moskowitz
ICSA
EMail: rgm@icsa.net
Ted T'so
Massachusetts Institute of Technology
EMail: tytso@mit.edu
9. Full Copyright Statement
Copyright (C) The Internet Society (1998). All Rights Reserved.
This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain it
or assist in its implementation may be prepared, copied, published
and distributed, in whole or in part, without restriction of any
kind, provided that the above copyright notice and this paragraph are
included on all such copies and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
followed, or as required to translate it into languages other than
English.
The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on an
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS 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.
AI ELI5
On this page
§1What HMAC-MD5-96 Is For
This memo specifies MD5 (RFC 1321) combined with HMAC (RFC 2104) as a keyed authentication mechanism inside the IPsec Encapsulating Security Payload (ESP, RFC 2406) and Authentication Header (AH, RFC 2402). The goal is to make sure a packet is authentic and cannot be modified in transit.
HMAC is a secret key algorithm, so what it proves depends on who knows the key. If only the source and destination know it, a correct HMAC proves the packet was added by the source.
It is a tamper seal on each packet. Only people who know the shared secret can make a seal that checks out.
§2Algorithm and Mode
MD5 is described in RFC 1321 and HMAC in RFC 2104; HMAC is a framework into which hash algorithms such as MD5 can be inserted. HMAC-MD5-96 works on 64-byte blocks. MD5 already defines its own padding, so no extra padding is needed, and no implicit packet padding as defined in AH is required.
HMAC-MD5 produces a 128-bit value. For ESP and AH, an implementation MUST support truncating it to the first 96 bits. The sender stores the truncated value in the authenticator field. The receiver computes the full 128-bit value and compares its first 96 bits with the stored field. No other authenticator lengths are supported.
96 bits was chosen because it is the default authenticator length in AH and meets the security requirements of RFC 2104.
The recipe makes a long fingerprint, but only the first part of it is written on the packet. The receiver makes the whole fingerprint and compares just that part.
- Block size64-byte blocks.
- Full output128-bit authenticator value.
- Stored valueFirst 96 bits, in the authenticator field.
- Key lengthFixed at 128 bits.
§2.1Performance
The RFC quotes the HMAC authors as saying that HMAC performance is essentially that of the underlying hash function. RFC 1810 gives performance analysis and advice on using MD5 in Internet protocols. At the time of writing, the memo says no performance analysis had been done of HMAC or of HMAC combined with MD5.
RFC 2104 also outlines an implementation change that can speed up per-packet work without affecting interoperability.
The seal costs about as much effort as the scrambling recipe underneath it, and there is a known shortcut that makes it faster.
§3Keying Material
For ESP and AH, a fixed key length of 128 bits MUST be supported, and other key lengths MUST NOT be supported. The RFC bases this on RFC 2104: keys shorter than the authenticator length weaken security, and longer keys do not significantly add strength. The key MUST be generated with a strong pseudo-random function.
No weak keys for HMAC were specified at the time. If any are found later, they must be rejected and replacement keys requested, or a new Security Association negotiated. To give data origin authentication, the key distribution mechanism must hand out unique keys and share them only with the parties in the communication.
The RFC repeats the RFC 2104 advice that periodic key refreshment is a fundamental security practice, even though current attacks do not point to a specific frequency.
The secret code word must be a fixed length, truly random-looking, and known only to the two friends. Changing it now and then is a good habit.
§4Interaction with the ESP Cipher
When ESP also encrypts, the authentication algorithm sits beside a cipher. The RFC says that at the time of writing there were no known issues preventing HMAC-MD5-96 from being used with any specific cipher algorithm.
The seal works fine no matter which lock-and-key scrambler is used to hide the contents.
§5Security Considerations
The strength of HMAC-MD5-96 rests mainly on HMAC and, to a lesser degree, on MD5. RFC 2104 claims HMAC does not need strong collision resistance, which matters because MD5 had been shown to be much less collision-resistant than first thought. At the time of writing there were no practical cryptographic attacks against HMAC-MD5-96.
For a 64-byte block hash, a birthday attack involving 264 blocks would be infeasible unless the hash were found to have collisions after processing 230 blocks. The RFC adds that HMAC was designed for keyed use from the start, while MD5 was not, and that HMAC with MD5 had held up under scrutiny. Truncating the output is encouraged for specifications that use HMAC, and MD5 could be swapped for another hash such as SHA-1.
As with any algorithm, security also depends on a correct implementation and sound key management. RFC 2202 provides test vectors and example code for checking HMAC-MD5-96 code.
MD5 alone has some cracks, but the way HMAC uses it does not rely on the cracked part. Even so, it only stays safe if the code is correct and the secret stays secret.