Re: IPsec mandatory authentication algorithms

["C. Harald Koch" <chk@utcc.utoronto.ca>]

In message <199710070713.DAA15269@relay.hq.tis.com>, Karen Seo writes:
> Folks,
> 
> There's an inconsistency between the AH and ESP specs and the DOI.

If I remember correctly, the original reason for mandating MD5 was backwards
compatability with previous implementations. However, with the sequence
number field and the new 96-bit truncated digests, we've *completely* broken
backwards compatability; the requirement is gone.

The cryptographic community appears to have declared MD5 anywhere from
suspect to compromised, depending on their level of paranoia.

Therefore, I'd recommend making HMAC with SHA-1 *mandatory*, and possibly
even specify that it should be preferred when negotiating.

Whether or not HMAC with MD5 is also mandatory is less important to me... :-)

-- 
C. Harald Koch     <chk@utcc.utoronto.ca>

---

Follow-Ups:

• Re: IPsec mandatory authentication algorithms
• From: Robert Moskowitz <rgm3@chrysler.com>

References:

• IPsec mandatory authentication algorithms
• From: Karen Seo <kseo@bbn.com>

---

• Prev by Date: I-D ACTION:draft-ietf-ipsec-esp-v2-01.txt
• Next by Date: Re: IPsec Architecture -- proposed changes
• Prev by thread: IPsec mandatory authentication algorithms
• Next by thread: Re: IPsec mandatory authentication algorithms
• Index(es):
  • Main
  • Thread