(forgot to hit reply to all)
I think we should treat the issue by stating that in the mixed
environments currently being tested a 10 to 15% cpu usage has been
noted. This will allow the SA types to adequately engineer a DKIM
solution based on their layout. When you think of all the edge
processing, filters, heuristics, etc, that is almost all cpu. The only
IO problems I have is trying to find an ever increasing mound of
Cox Communications, Inc.
From: Dave Crocker [mailto:dcrocker(_at_)bbiw(_dot_)net]
Sent: Monday, July 10, 2006 4:11 PM
To: Eliot Lear
Cc: Oxley, Bill (CCI-Atlanta); ietf-dkim(_at_)mipassoc(_dot_)org
Subject: Re: [ietf-dkim] Comments on -overview document?
Eliot Lear wrote:
I think what Dave could say is that in I/O intensive environments DKIM
will have a negligible impact because minimal disk I/O is required.
The counter-point, to claims that email is i/o bound, is that modern
requirements for filtering alter the equation quite a bit.
That's why I was intrigued to hear the comments about being i/o bound
folk doing empirical -- real-world -- testing of DKIM.
That said, these days it is nearly always true that one can set up a
to use most of the cpu, no matter how much i/o they have.
NOTE WELL: This list operates according to