New features focus on scalability and usability as this product with its open source base finds favor with ever larger organizations.

by Alex Goldman ISP-Planet Managing Editor [October 29, 2007]

Ottawa, Canada-based Roaring Penguin Software announced the release of version 4.0 of its CanIt family of anti-spam products, which range from free to premium. An appliance-based solution is also available.

The products are based on open source technology developed by the company's president, David Skoll. MIMEDefang was initially released in 2000 and remains available as standalone open source software. It is also incorporated into Roaring Penguin's technology.

Upgraded database
As the volume of spam on the internet increases, customers are increasing the size of their anti-spam systems even if their business is not growing (and many are growing too). Skoll says that the company has fundamentally changed how it uses storage in order to make the system more efficient.

"We used to store everything, including messages and tokens, in gigabytes of unstructured text into the database. Now we're written the CanIt Storage Manager daemon (ISPs can run one or multiple instances of it). Now, if you have five machines, you could store three copies of everything so that if any one machine goes down, you still have all of your data."

Does that require extra RAM or an extra-powerful CPU? "The bottleneck is disk I/O bandwidth," says Skoll.

So the solution reduces the strain by breaking up the workload. "A smaller database is nicer and faster and easier to backup and do routine maintenance on."

A further benefit of a database stored in multiple instances is the elimination of what was a single point of failure.

Additional features
The company upgraded the rest of the package too, adding:

• Better reporting, including domain-by-domain statistics
• Automatic updates for all anti-spam rules
• Better support documentation
• Support for provisioning through the PLESK control panel

Although graphical statistics are available, Skoll says that those who liked going into the database and writing their own queries can still do so.

At this point in the interview, we say that all this is great, but what are you going to do for an encore? "We're going to wait and hear from our customers about what features they need," says Skoll. The company is already looking at adding IP-based reputation and further improved scalability, but more features are sure to be added to the wish list as requests come in.

—End