[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: MD5 versus SHA



> From: James M Galvin <galvin@tis.com>
> 
> 	>However, the choice of MD5 for SNMP did not include performance
> 	>considerations.
> 
> In fact, this is exactly why we chose it.  I should know.  I was chair
> fo the SNMP Security Working Group and I'm co-author of the security
> documents.
 
It appeared to me (and other subsequent RFC authors
regarding the use of MD5 in RIP, etc), that performance
(i.e., 100+ Mbps) was not a concern, only overhead with 
respect to SNMPv2:

> From: shirey@mitre.org (Robert W. Shirey)
> 
> However, I recall being at that WG in Santa Fe when MD5 was finally
> selected.  Until then, there has been the usual speculation about "too
> slow" etc.  At that meeting, the implementers presented performance results
> gathered since the previous IETF.  They indicated that the MD5 operations
> did not add significantly to SNMP processing and the "too slow" discussion
> was a waste of time.

I fact, the RFC itself hints that the reference implementation of
MD5 isn't as fast as desired, calling for optimization (page 7):

          An appendix of [3] contains a C Programming Language
          implementation of the algorithm.  This code was written with
          portability being the principal objective.  Implementors may
          wish to optimize the implementation with respect to the
          characteristics of their hardware and software platforms.

What I tried to do was show upper-bounds on these optimizations
with analysis. Given that information, and the fact that the
performance requirements of IP should be higher than SNMP (i.e., SNMP
can afford to be "low BW", - 20 Mbps), this should affect the 
decision to use MD5 in IPv6.

Joe

PS - the words
	performance
	speed
	bandwidth

do not appear in RFC-1446, on which I was commenting.