SpamAssassin proves to be an effective engine for battling spam

SpamAssassin has proven itself to be a cost-effective and valuable open source weapon in the war on spam. But it’s also a complex solution. Capitalizing on SpamAssassin’s strengths, a number of companies have incorporated it into commercial products, adding extra features and much easier installation and manageability.

Free IT resource Virtualization Insights from Top Experts - Learn how virtualization gets real! Sponsored by Dell Free IT resource TechNet: More ways to know it, share it, and keep it running. Sponsored by Microsoft

I recently looked at SpamAssassin 2.63, as well as two commercial products built around it, CanIt-Pro 2.0b from Roaring Penguin Software and the MailPoint 3000 appliance from Digitalinfo Networks. The three products illustrate the full spectrum of convenience vs. cost that applies to all anti-spam products, not just those based on SpamAssassin.

Downloading and installing SpamAssassin 2.63 required a substantial investment in time and reading through the documentation, but the software worked well once I jumped through all its hoops. CanIt-Pro required installing Red Hat Linux and then the software. It also involved a good bit more configuration than the MailPoint appliance, but it proved more flexible. The MailPoint 3000 box took five minutes to install and required little configuration but lacked some features that large companies might need.

In my tests, MailPoint 3000 and CanIt-Pro performed better at filtering out spam than SpamAssassin. However, further tuning of SpamAssassin would undoubtedly have yielded better results.

SpamAssassin 2.63

If you’re a full-time Linux administrator adding SpamAssassin to an existing Linux-based e-mail setup, you’ll find that it can provide more control than most costlier commercial packages. However, if you aren’t familiar with Linux system administration and you don’t relish the idea of wading through hundreds of pages of documentation, SpamAssassin may not be for you. This is not a slap at SpamAssassin; it is capable of good performance and it’s extremely flexible. But it’s not for everyone.

I installed SpamAssassin 2.63, along with the ancillary packages recommended by SpamAssassin.org. It took me a couple of hours to get Linux and the packages installed, but configuration was the real issue. Without looking at every document available, I downloaded more than 700 pages of documentation and found that some of them referred to older versions of SpamAssassin or Linux, and some documents contradicted others. On the upside, there are a couple of newsgroups available for SpamAssassin users and I was able to get quick answers to my questions there.

Configuration is done via a command-line interface, editing text files and Perl scripts. Although SpamAssassin itself doesn’t require a lot of configuration once it’s installed, getting the OS updated with all the correct supporting packages, adding required packages, and getting your e-mail application configured properly to work with SpamAssassin can take some time.

SpamAssassin uses a number of the usual effective techniques to spot spam: header analysis, text analysis, blacklists, real-time blackhole lists, and the newly added Vipul’s Razor, a collaborative spam-tracking database. Additional enterprise-oriented tools can be installed to allow administrators to apply different filtering settings for individual users and groups, or to allow users to access quarantined e-mail and to whitelist senders. These tools can be downloaded and installed freely, but finding them and getting them to work is not a trivial exercise.

Once I got the SpamAssassin software configured and running, its default settings provided acceptable performance, blocking 88 percent of spam, but with a very high 14.77 percent false-positive rate. With a few months of use and tuning, however, I expect its performance would improve substantially. Adding available plug-ins, such as the Bayesian filter or the content-checking filter, would likely help too.

While not a fit for administrators inexperienced with Linux, SpamAssassin is a powerful, extensible package that can perform as well as commercial solutions, provided you are willing to spend the time tuning it and updating it as necessary.

Roaring Penguin Software CanIt-Pro 2.0b

Administrators looking for an inexpensive spam-fighting solution that brings out and adds to the best of SpamAssassin might consider CanIt-Pro.

Although far simpler to install than SpamAssassin, it proved more complex to install than the MailPoint box, requiring the Apache Web Server, PHP Web interface, Perl, and C programming environments. Still, installing the CanIt-Pro package is not onerous and configuration is complex only because of its wide range of features, its scope, and its flexibility. According to Roaring Penguin, a high-end server can handle 100,000 messages a day.