[imp] Permformance issue

Eric Rostetter eric.rostetter@physics.utexas.edu
Tue, 16 Jul 2002 10:19:00 -0500


Quoting Andrew Morgan <morgan@orst.edu>:

> Apparently we have different environments entirely, because you and I have
> come to different conclusions on this.  :)

Yes, we have completely different environments, which was my whole point
in this thread.

And we didn't come to as different a conclusion as you think.  I happen
to agree that in your environment it would probably be best to use 
multiple machines.  Though I'd still encrypt everything myself.
 
> I'm providing email services for 30,000 users, with about 10,000 logins
> per day on the webmail system.  During busy times, this is a full load for
> both the mail server (IMAP) and the webmail server. 

I'm providing service for about 1000 users, so you can see why my 
environment is much different.  Our main university folks also provide
webmail for the university.  They probably have around 50K or more 
accounts (our enrollement in the university is about 51K students at
this campus, plus about 17k faculty/staff, and then of course there
are the other 25 campuses).  They use multiple web servers behind a
load-balancing device, with multiple IMAP backends, each of which I
think runs off some NFS appliances...  Why?   Because their environment
is different!

In my previous job, I did manage email for about 40K accounts, so don't
think I don't know about varied environments.  I understand them completely.
Which is why I take objection to the original question as providing too
little detail about the environment to be answerable in any meaningful
way.

> In our case, I trust
> that the network is secure, at least for the level of security required
> for our site.

Great.  For me, I don't trust that the network is secure.  All passwords
must be encrypted.  That's up to your site policy (i.e. your environment).

> I'm curious why your performance was twice as bad when you put IMP on one
> server and IMAP on another.  Can you explain it for me?

Because of the environment.  It was really a trick question though... :)

We use wu-imapd, and use mbox format in the user's home directory.  In the
original setup, the imapd server held the home directories and exported
them via NFS.  In the new setup, the imapd still has a local /var/spool/mail,
but the home directories are now NFS mounted from another machine.  So any 
mail operation involves hitting the home directory over NFS now.  So the
performance hit is almost entirely from hitting the NFS mounted disk rather 
than a local disk.

> As a side note, there are some times when having separate servers is
> *more* of a management problem.  For example, the webmail server does
> weird things when the IMAP server is unavailable...  :)

Granted.

> 	Andy

-- 
Eric Rostetter
The Department of Physics
The University of Texas at Austin

"TAD (Technology Attachment Disorder) is an unshakable, impractical devotion
to a brand, platform, product line, or programming language. It's relatively
harmless among the rank and file, but when management is afflicted the damage
can be measured in dollars. It's also contagious -- someone with sufficient
political clout can infect an entire organization."

--"Enterprise Strategies" columnist Tom Yager.