Roaring Penguin Software Inc., an Ottawa-based Linux distributor, released its latest version of CanIt anti-spam solution called CanIt 2.0.
Similar to its predecessor, CanIt-PRO, version 2.0 employs MIMEDefang UNIX Milter, and SpamAssassin spam detector technology. The new version, as with all
of Roaring Penguin’s anti-spam products, runs on the mail server, scanning every piece of mail that comes through it.
If it decides that the mail is not spam, it delivers it to the user. If, however, it thinks that the mail might be spam, it holds it in a database where an administrator reviews it to verify whether or not it’s spam.
“”The way we make our decision is that we have a few hundred different heuristics that look at characteristics of the message such as the header and the body. We also have a bunch of non-content-based rules that look at characteristics of the server sending the mail,”” said David Skoll, president of Roaring Penguin.
CanIt allows organizations to delegate spam management to individual department heads, business units, and end users by separating mail into streams. From there, each spam manager can choose how much or how little control he or she wants and/or needs. This avoids potential conflicts between people in an organization when determining what is and isn’t spam.
“”If you have a university with a medical faculty, it (CanIt) will quickly learn that medical students get different kinds of legitimate email than engineering students,”” says Skoll.
Version 2.0 also includes Bayesian Filtering, a probability-based technique that enables the system to learn and detect spam without the need for additional filters. Bayesain Filtering keeps a table of all the words seen in emails and a count of how often each word has appeared in spam and non-spam respectively.
“”When a new message comes in, we can use statistical techniques to look at the words in that email, look at our table and come up with a probability that the email is spam or not spam,”” said Skoll.
The U.S. price tag starts at around $6 per year, per mailbox for the first year and half of that for the following years, which, according to Skoll.
“”If you have a 1,500-person company, the first year would cost you $7,500,”” says Skoll, “”That’s $5 per mailbox.”” Subsequent years would be $2.50 per mailbox.
Organizations can expect to see ROI within a year, says Skoll.
Founded in 1999, Roaring Penguin is looking to develop its reseller channel in SMB market in 2004 but is planning to continue to serve its existing markets.
“”Our target market is any organization that’s big enough to run its own mail server,”” said Skoll. “”That can be an organization as small as 20 users all the way up to tens or hundreds of thousands of users.””
The channel strategy behind entering SMB markets is based on what works well for Roaring Penguin partners, Skoll said.
“”They’re (resellers) geared up to provide installation support for these smaller organizations on site.””
One of Roaring Penguin’s resellers that cater to this market is BFA Datasystems. Kevin Daly, manager of BFA, says out of three anti-spam products that they tested, CanIt was the most effective.
One of the biggest challenges, says Daly, is not only looking at how many items it catches, but also how many false positives it generates. In one case, a customer emailed Daly a technical request and the (Astaro) firewall identified it as spam and threw it out.
“”Of course I got an angry phone call and I said, ‘I never got your email,'”” recalls Daly. To solve the problem, he says that BFA had to back the policies down on its firewall. Now, Daly says BFA is getting an “”unacceptable number”” of spam. But, if BFA brings the policies back up, it throws out too much good mail. This won’t be a problem when BFA replaces Astaro with CanIt in January.
“”CanIt’s has better algorithms and finer granularity,”” says Daly adding, “”(CanIt) allows us to set the policies up in such a way that it traps most spam but generates very few false positives.””
