Main Content

Email metrics and filtering

Archive - Originally posted on "The Horse's Mouth" - 2010-03-28 07:30:44 - Graham Ellis

Every few minutes, a spam email drops in my mailbox ... and indeed every five minutes or so throughout the day, our server rejects an email that is so obviously, undeniably spam that it doesn't even bother to notify us of its arrival. This is a long-running graph showing spam (red) and delivered to our mailboxes (blue) on a daily basis. Our filters are tuned to pass things through to us if there's any doubt - we're an organisation that welcomes new customers, and they often introduce themselves by email, and with a none-obvious title line (such as "Hi").


But looking at my inbox this morning, there's a gap of over an hour when I didn't get a single email! Oh - I see - it's spring and the clocks have gone forward ... and that was the missed hour.


We use SpamAssassin to do our primary filtering, and a Perl script run within our mail client to then log the emails marked as Spam. The graph at the top of this item is produced by the gd module running within PHP. I wouldn't advocate that customers follow our lead and mix languages like this - we've used them both because we provide both Perl Programming Courses and PHP training classes ... using both languages ourselves for real solid and substantial work.