Tip of the Trade: Postgrey
Admins, as a general rule, spend far too much time and energy trying to keep spam out of the networks. Until technology is developed to reach out and touch spammers forcefully spam will be an unfortunate and expensive fact of life. Fortunately, some simple hacks can be applied at the server level to derail a significant amount of spam. For Postfix admins, for example, there is the Postgrey Greylisting Policy Server. Who knew such a small Perl script add-on could be so powerful? Postgrey Greylisting Policy Server is a simple hack that shoos spam away from the Postfix MTA by reacting to behavior, not content.
Postgrey is a small Perl script add-on for the Postfix MTA. It uses an ingenious, low-overhead method of shooing spam away, reacting to behavior rather than content. This has a number of advantages: Content filtering is slow, error-prone, and uses significant CPU cycles. It is also completely ineffective on encrypted messages. Postgrey operates at the SMTP level, which is low-overhead and fast.
The first time Postgrey sees a message from a new CLIENT_IP/SENDER/RECIPIENT combination, it is bounced with a "450 mailbox unavailable" error. 450 errors are not permanent. The sending server, according to correct behavior, should re-send the message. Legitimate mail servers will do this, but spam and virus servers rarely do. If the sending server is still trying after five minutes, Postfix accepts the message and adds the sender information to its whitelist database. So the biggest cost is to users, who will see a five-minute delay the first time they receive a message from a new source. At least for now, this derails a significant amount of spam for a very low cost a few lines in main.cf, and virtually no load on the server.