[Mailman-Users] Hardware Requirements

[alex wetmore]

On Wed, 10 Jul 2002, Tim Crouch wrote:
> I am looking for suggestions on a hardware purchase.  I am setting up a
> new mailing list server for our University. It will host up to 500 lists
> ranging from 2 subscribers to 3000 with an average of under 200.  We
> will archive no more than 1 year's worth of messages.  Obviously disk
> size will be number one on the priority, but I am looking for what you
> would run this on.  OS will be RH Linux 7.3 the hardware will be from
> Dell.  I am leaning towards the following:
>
> Dell PowerEdge 350 1u rack server
> 850MHz Celeron, 128KB L2
> 512 MB SDRAM
> 20 Gig IDE boot drive
> 120 GIG IDE data (for archiving, data, & web)
> RH Linux 7.3
>
>
> This machine needs to have a lifetime of at least 3 years....

More important than list size is the volume of email that you need
to process.

For a server handling 500 lists I'm going to assume that during
peak volume you are going to be sending out a lot of messages.
If the server is also to act as your MTA then you will probably
need additional disk capacity.  Not in space, but in throughput.
Stripping your queue across multiple disks is the best way to
gain throughput.

The FAQ contains scaling tips for using Mailman with different MTAs.
I don't know what MTA you are using, so I can't provide any more
useful advice.  On my MTA (Windows 2000 SMTP) throughput is tightly
linked to the number of spindles...using 7200rpm disks the system can
send about 15msgs/sec/disk.  This number will vary depending on the
MTA, underlying file system, and your disk subsystem.

Assuming each list has 200 recipients (your average) and 10 messages
per day then you need to send up to 1,000,000 messages per day (I'm
also assuming each message is going to a different domain, or that
you'll turn on VERP when you have Mailman 2.1).  That is 11/sec spread
out through 24 hour day.  Chances are that you'll have a 10 hour peak
window with most of that traffic, and will need considerably better
throughput.

I would look at least striping your MTA queue across two spindles.

alex