Re: WG LC (Deadline Dec 18th, 04) for I-Ds: draft-ietf-mip6-auth-protocol-01.txt anddraft-ietf-mip6-mn-ident-option-00.txt

>  - another point: MD5 should be replaced by SHA-1 everywhere in the
document.

Yes, this can't be emphasized strongly enough.

Results by Xiaoyun Wang from China presented at Crypto in Santa Barbara this
summer showed that with 1 hr of precomputation on IBM P690, new collisons on
1024 bit messages could be generated every 15 sec. to 5 min. MD5 should be
considered insecure and not used, and it should be EOL-ed in existing IETF
protocols and replaced with SHA-1.

SHA-1 appears to be secure after 40 rounds but has been broken up to 40 (it
typically takes 80), which appears to be the limit possible with the current
cryptanalysis technique. Whether its possible to break SHA-1 completely,
remains to be seen.

                    jak