List server performance

Monbebe Admin

unread,

Jan 27, 2000, 3:00:00 AM1/27/00

to List-M...@greatcircle.com

Hello List-Managers,

Is there (a) comparative table between different list servers,
in term of posting performances ?
I mean something like for example:

etc ??

Best regards,
--
Technical Support
http://www.monbebe.net
Un site pour apprendre a etre parents

Dave Sill

unread,

Jan 28, 2000, 3:00:00 AM1/28/00

to List-M...@greatcircle.com

Monbebe Admin <ad...@monbebe.net> 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.

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.

My less-than-meaningful contribution is:

system: AlphaServer 2100
CPU: 2 x 200 MHz
RAM: 320 MB
OS: Tru64 UNIX 4.0D
MTA: qmail 1.03
MLM: Majordomo 1.94.4
LAN: 100baseT
WAN: OC3?
name server: dnscache 0.70
background load: INN news server, anonymous FTP
disks: 7200 rpm scsi
message size: 3 KB
recipients: 1846 (tru64-uni...@ornl.gov list)
average delivery rate: 7.9 per second
peak delivery rate: 122 per second (45 msgs in a .37 s interval)
90% delivery rate: 21.3 per second

-Dave

Monbebe Admin

unread,

Jan 28, 2000, 3:00:00 AM1/28/00

to List-M...@greatcircle.com

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.

Christopher Knight

unread,

Jan 29, 2000, 3:00:00 AM1/29/00

to List-M...@greatcircle.com

At 10:36 AM 1/27/00 +0100, Monbebe Admin wrote:

Hello List-Managers,

Is there (a) comparative table between different list servers,
in term of posting performances ?

One of the easy to follow guides we use, is that we're
able to deliver 100K emails per hour per Mbps.

Cheers,
Christopher M. Knight
SparkLIST.com LLC • The Business Email List Experts •
---------------------------------------------------------------
SparkLIST Email List Hosting, Promotions & Management Service
Tel: +1 888-SparkNET, ext 212 or +1 920-490-5901, ext 212
Private Fax: +1 920-490.5909 • http://SparkLIST.com/
---------------------------------------------------------------