One of the common complaints I’m hearing as I host the Network World Small Business Technology Tour this summer is about spam. We've been fighting it for years, and although the tide may be turning in our favor, victory is not yet ours. And a vendor on the tour that seems to be impressing audiences is Barracuda Networks.

I always recommend a two-step spam block approach: Block the obvious spam at the server, and let individual users apply their own fine-tuning to the spam filters. Barracuda takes a similar approach, but it uses a dedicated appliance to control spam before it even reaches your mail server.

Blocking spam before it reaches the server makes sense, especially since spam constitutes the majority of e-mail on the Internet. Putting spam filtering on your mail server means the server must process each message to find the spam. The horsepower used by the spam filtering software often requires users to upgrade their servers to handle the load, making the spam filtering project cost even more.

Barracuda puts its appliance in front of your mail server, reducing the server load.  So if you get 10,000 messages per day and half to three quarters are spam, a Barracuda appliance might prevent 5,000 to 7,500 of those messages from reaching your e-mail server, reducing the likelihood that you will have to upgrade the server for quite some time.  Barracuda didn't advertise this on the tour, but I bet many companies find a spam appliance costs less than a server upgrade.

Unlike SpamAssassin, the popular open source software that uses Bayesian filters  to learn which messages are spam by reading the words in the message and assigning a score to indicate the probability, Barracuda uses Bayesian filters and nine more spam filtering technologies.

The Ten Trusted Defense Layers for Barracuda Spam Firewalls are: denial of service (stops attacks); IP block list (known spammers); rate control (too many messages from one place); two virus filters (one written by Barracuda and one open-source filter); spam fingerprint check (tag one new spam as spam, and all copycats that normally appear within a few days are blocked); spam intention (links back to bogus Web sites and the like); Bayesian analysis; and individual spam rules you can add yourself. Spam- and virus-filtering databases update hourly to keep fresh. Users can tweak their own filters, and questionable mail goes into a quarantined box for review.