Re: iSCSI: Public Key Method

[Michael Eisler <mre@zambeel.com>]

> > A derivative
> > is necessary, because there are intellectual property
> > issues with at least one of the crypto algorithms specified in
> > RFC 2025
> 
> Can you specify which algorithm you're referring to ?  I guess

Ofer,

The md5WithRSAEncryption and sha1WithRSAEncryption integrity  algorithms.
However, it looks like the IP issues were around the RSA algorithm itself,
which according to http://www.ietf.org/ietf/IPR/RSA, has expired. (In my
defense, at the time  the LIPKEY/SPKM-3 RFC was written/published the patent
hadn't expired, and I was asked by the WG chair to make the algorithm not
mandatory). Assuming the IP issues are now moot, I would expect RFC 2847 to be
updated someday to REQUIRE sha1WithRSAEncryption, and make id-dsa-with-sha1
OPTIONAL.

There may however, be other reasons to make an SPKM-2 derivative,
relating to, for example, weaknesses discovered in MD5 that weren't
known when RFC 2025 (SPKM) was published. I went through the effort of
looking for these deficiencies in SPKM-1, but didn't bother to
make the analysis in SPKM-2, since I had no immediate use for it.

> this is less a problem for us since it's only an optional method
> (David - please correct me if I'm wrong here...)

Unless it is envisioned that both the iSCSI client and server will often
come from the same vendor, making SPKM, Kerberos V5, etc. optional
to implement is unlikely to result in a high frequency of situations
where these methods get used.

	-mre