I installed and tested SpamBayes on one of the production email servers at our company, OnlineNIC. The results have been impressive — roughly 30,000 junk emails, totaling around 1 GB of data, are filtered out every day.
This particular server hosts more than 2,500 domains and 30,000 mailboxes, so the volume of spam passing through it is enormous. SpamBayes handles the load without breaking a sweat.
Currently, we are using a single global spam recognition database and only dropping messages that SpamBayes classifies with very high certainty. This conservative approach avoids false positives, but it also means some spam slips through. The next step is to set up an individual database for each mailbox and provide a user-friendly control panel so that customers can train the filter themselves.