Great Circle Associates List-Managers
(June 1995)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: Re: Mailing list throughput
From: Eric Thomas <ERIC @ SEARN . SUNET . SE>
Date: Tue, 6 Jun 1995 00:18:41 +0200
To: Paul Graham <pjg @ acsu . buffalo . edu>
Cc: list-managers @ GreatCircle . COM
In-reply-to: Message of Mon, 05 Jun 1995 17:31:16 -0400 from Paul Graham <pjg @ acsu . buffalo . edu>

On Mon,  05 Jun  1995 17:31:16  -0400 Paul  Graham <pjg @
 acsu .
 buffalo .
 edu>
said:

>it seems  that you're agreeing  with stan.  in particular if  you assume
>adequate i/o bandwidth as well as  adequate cpu you can arrange parallel
>delivery to all recipients in real  (or near real) time. doing this with
>sendmail can be resource intensive.

Stan said just CPU and referred to a 486/33, which is indeed considered a
"slow" CPU  nowadays. What you  need (with  sendmail) is CPU,  memory and
I/O, growing mostly in n2, and of course bandwidth, growing linearly. You
can buy a machine  10 times faster than Stan's 486 and  still not get all
that much  more work done. In  fact on dedicated machines  sendmail works
better  with *slower*  CPUs.  Upgrading  from a  486/33  to  a P90  would
probably cause  sendmail to swap the  machine to death because  it thinks
it's underused, when in fact it is  working at capacity as far as RAM and
I/O are concerned (sendmail quirk #1829812).

  Eric

Indexed By Date Previous: Re: Mailing list throughput
From: Paul Graham <pjg @ acsu . buffalo . edu>
Next: <postmaster @ foo . com> resulting in error messages every half hour
From: <ead @ netcom . com>
Indexed By Thread Previous: Re: Mailing list throughput
From: Paul Graham <pjg @ acsu . buffalo . edu>
Next: <postmaster @ foo . com> resulting in error messages every half hour
From: <ead @ netcom . com>

Google
 
Search Internet Search www.greatcircle.com