[22456] in Kerberos
Re: differences between des3-cbc-sha1 and des3-cbc-md5
daemon@ATHENA.MIT.EDU (Tom Yu)
Tue Sep 14 20:16:56 2004
To: "Ahluwalia, Ish" <iahluwalia@sonusnet.com>
From: Tom Yu <tlyu@mit.edu>
Date: Tue, 14 Sep 2004 20:15:42 -0400
In-Reply-To: <A3863F3136CBC546A40A61BA9CBA9D93ABD63E@sonusmail03.sonusnet.com>
(Ish Ahluwalia's message of "Mon, 13 Sep 2004 19:02:26 -0400")
Message-ID: <ldvfz5kmlox.fsf@cathode-dark-space.mit.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
cc: kerberos@mit.edu
Errors-To: kerberos-bounces@mit.edu
>>>>> "Ish" == Ahluwalia, Ish <iahluwalia@sonusnet.com> writes:
Ish> When we talk about des3-cbc-sha1 support - The way I understand
Ish> it means support for des3-cbc encryption with
Ish> CKSUMTYPE_HMAC_SHA1_DES3. I hope my understanding is corrrect in
Ish> terms of this.
I believe this is correct, though perhaps not in the way you are
thinking. CKSUMTYPE_HMAC_SHA1_DES3 is a cryptographic checksum
(signature) algorithm which is required to be supported when a session
key type of des3-cbc-sha1 is used.
Ish> Now, for "des3-cbc-md5" encryption - it means we again need
Ish> support for des3-cbc encryption. But what checksumtype needs to
Ish> supoorted - Is it CKSUMTYPE_HMAC_MD5_DES3 or CKSUMTYPE_HMAC_MD5?
I don't think either CKSUMTYPE_HMAC_MD5_DES3 or CKSUMTYPE_HMAC_MD5 are
defined currently. I suspect it might be an MD5 hash encrypted with
triple-DES, which is not as secure as an HMAC.
Ish> Essentially, I'm asking if if the process is same between the two
Ish> ciphersuites, just that HASH algorithms are different?
The hash algorithms are not the only difference. des3-cbc-md5, like
the single-des cryptosystems, uses encrypted plaintext checksums,
which are known to have cryptographic weaknesses. des3-cbc-sha1 uses
an HMAC instead, which eliminates these weaknesses. des3-cbc-sha1
also uses key derivation, which provides additional strength.
Have you read the recent Kerberos crypto specification?
(draft-ietf-krb-wg-crypto-07.txt, which I think is pending
publication) It will probably make some things more clear to you.
---Tom
________________________________________________
Kerberos mailing list Kerberos@mit.edu
https://mailman.mit.edu/mailman/listinfo/kerberos
