2000 Exchange Mail Server - How To Filter Spam (Part One of Two)

If you were suddenly made responsible for stopping spam on your Microsoft Windows network, what would you do first? Well, if you're like me, you'd probably Google "anti spam Exchange server" or something similar to that.

Next, you'd spend some time reading a lot of reviews and articles about anti spam software suited for a Microsoft enterprise network.

Lastly you'd extract (from the articles) a list of the most commonly found bells and whistles that the current anti spam technology offers. You'd use this as a kind of wish list, or yardstick, against which you would judge all the possible anti spam solutions available to you.

Well, as you might imagine, I recently did that and here is a list of features that any decent Exchange server-compatible anti spam solution should have. Your mileage may vary depending on what you need your solution to provide, but here's my list.

Filter spam on an Exchange mail server by doing one or more of the following:

* Bayesian Analysis

Anti spam software of this sort allows you to begin by classifying one or emails as "spam." Then, as your email arrives, the anti spam software looks at every other email to see if fits the definition you set. If it does, it goes to the spam folder; if it doesn't, it gets through to your inbox. But it doesn't stop there. Suppose an email gets through the "Bayesian filter." You can (and should) tag that email as spam -- so that the software can add the characteristics of that email to its spam profile. What this means is that future emails of this sort will NOT sneak through.

In other words, the anti spam software that uses Bayesian analysis can learn and get better over time -- if you teach it to do so.

* Downloadable spam signatures

Typical Bayesian analysis relies mostly on your own unique description of what you consider to be spam -- and that's important.

But sometimes, two heads are better (and faster) than one. What if you could tap into a database of spam characteristics that were put together by hundreds or thousands of other users? That what you get when you use anti spam software that periodically downloads new spam signatures. The big advantage is that you are casting a wider net. The disadvantage is that you might not agree with the crowd's opinion of what constitutes spam.

* Keyword checking

This anti spam feature identifies spam email by looking for certain words, like "Viagra." As you might imagine, this is a pretty crude method. For example, what if your name is Gay Smith? A keyword-checking spam filter might strike this email for containing objectionable content. Don't laugh -- I just read about this very thing happening.

* Header checking

This is a pretty basic form of spam blocking where the anti spam software checks to see if the email header shows tell-tale signs of spam:

o Has the sender identified himself in the From: field? If not, it's probably spam.
o Does the email have a "large number" of listed recipients (you can set the number)? If so, consider it the same as spam.
o Block email containing only remote images. To get around keyword or even Bayesian filters, spammers now send out email that shows only a picture, i.e., a graphic of a bottle of viagra with an associated (non-text) caption. In addition, they might put a small amount of non-spam text in the message. Good anti spam software will block that.
o Check to see if the domain is real. The anti spam software will do a "DNS lookup." If the domain is not listed, the email is probably spam.

(End of Part 1. Part 2 can be found in the index.)

Here's the complete version of this article on how to filter spam on 2000 Exchange mail servers.

For more articles on stopping spam, visit Ara Rubyan's Spam Blocker Central.

Ara Rubyan is experienced in training design and delivery and has been developing web based training for more than 10 years. His background in training coupled with his IT knowledge help him to ensure the seamless development of product user training.