[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: ESP revisions straw poll
>
> Finally, in talking with a couple of active contributors, I've gotten the
> impression that there is support for encryptionless ESP, as defined in the
> current I-D. The argumemts are that this should be easy to implement
> (since it is just ESP without encryption turned on), it is more efficient
> than AH, and it is both appropriate and adequate in tunnel mode, as an
> alternative to tunnel mode AH. So, I'd like to conduct a straw poll on
> this topic too.
>
> Steve
>
>
Steve, as a person who has been implementing IPSEC and KMP since 1994,
I like to offer my $0.02 opinions against encryptionless ESP.
1. It would make it very confusing to distinguish between ESP and AH.
This is serious because not all users/administartors of IPSEC are
experts on IPSEC. If they are confused, they may define the wrong
IPSEC policy and lose all security.
2. It would make it even harder to code/do ISAKMP negotiation.
3. AH does incur some extra overhead because it covers IP header.
However, according to our actual measurement (published in the
5th USENIX UNIX Security Conf.), this overhead is trivial compared
to time spent in computing MD5 digest. So I feel encryptionless ESP
does not save that much. Whoever want peformance should optimize
their crypto implementations; the saving would be much greater.
Pau-Chen