Hello Dave,
Thursday, January 27, 2000, 3:40:15 PM, you wrote:
> Such comparisons are only meaningful if they're processing the same
> workload from the same location, and there's other critical
> information you'd need like which MTA was in use (if the list manager
> doesn't handle delivery itself), background load on the server (there
> shouldn't be any), OS version, installed RAM, IDE vs. SCSI disks,
> which name server cache was used, of any, etc.
I'm totally agree with you, sorry not be to be as much detailled as
necessary.
> Also, average delivery rate isn't too useful for measuring delivery
> performance because a handful of slow recipient hosts really skews the
> result. You really need to look at the delivery curve, or at least the
> rate at which the first, say, 90% of deliveries occur.
Yes, but an average could be interesting anyway, and your entry is a good
start point.
Best regards,
--
Technical Support
http://www.monbebe.net
Un site pour apprendre a etre parents
