Greetings all:
I am looking for feedback and constructive criticism on how our organization is looking to deploy SpamAssassin. The solution we are looking for will only need to filter e-mail for virus and spam because all user mailboxes are currently stored on an Exchange 2000 server.
Here is what we would like to do:
1 - Have all inbound and outbound e-mail filter thru two redundant Apple G5 Xserves running Mac OS 10.3.5 server using Postfix version 2.0.10 as the MTA.
2 - All e-mail will be filtered for viruses using ClamAV tied into Postfix via Amavisd. (That part is simple)
3 - How should SpamAssassin be tied in? I have read arguments both ways whether or not to use spamd/spamc or use Amavisd. Spamd/spamc I feel is more configuration, but seems to allow the use of all of SpamAssassin's built-in features, as opposed to Amavisd being easy to integrate with Spamassassin, but use of built-in Spamassassin features are limited. What is everyone else's take on that?
4 - We wish to provide per-user preferences, AWL and bayesian databases using MySQL. For using MySQL is it better to use Amavisd or spamd/spamc? Also, if we are going to have redundant e-mail filters, it would make sense to provide redundant MySQL servers. Can we run MySQL on both boxes and have SA writing back to both databases on both servers at the same time? If SA can be writing to two different MySQL servers at one time, can the two filter servers be in two different data centers, i.e. one server in New York and one in San Diego?
I think that about sums up my questions, any feedback, concerns, questions or references are greatly appreciated.
Thank you, --- Matt Goli, MCP Systems Support Group
F+W Publications, Inc. - www.fwpublications.com Krause Publications, a division of F+W Publications, Inc. - www.krause.com
