Commtouch last week issued information about the growing problems with image-based spam during the third quarter of 2006.
The company reported that during the last quarter image spam accounted for one-half of all spam at its peak, up from a peak
of 30% during the second quarter.
 Related linksFor webinars or research on messaging, or to join the Osterman Research market research survey panel, go here. Osterman Research helps organizations understand the markets for messaging and directory related offerings. To e-mail Michael,
click here.
The company also reported that image spam, as well as zombie-generated spam, hit a peak of 3.5 million attack patterns in
one day.
New! Watch this Network World Webcast - New Webcast! Choose WAN acceleration products that improve application performance for all users.
One of the problems with image spam is that it’s so much bigger than traditional spam. For example, Commtouch reports that
the average image spam, which often is comprised of animated GIF files, is 44KB, or eight times the average size of normal
text-based spam. That means that not only is image spam much harder to detect with conventional spam-filtering tools, but
it also consumes much more bandwidth and storage, driving up the costs of managing a messaging system that much faster.
Commtouch this week also announced its IP Reputation Service. This service allows its partners to classify all inbound e-mail
traffic according to the risk level associated with its IP address and then to take appropriate action on that incoming traffic.
Those actions might include blocking e-mail from specific IP addresses or simply throttling that traffic to just a small number
of messages per hour.
Reputation analysis is becoming more critical as part of an overall spam management strategy. While content filtering will
continue to be an important tool to block spam, it’s a computing-intensive activity that consumes a significant number of
CPU cycles. Reputation filtering, on the other hand, can block a significant proportion of spam, can allow more fine grained
processing of potentially spammy sources than real-time blacklists, and can consume far fewer computing resources.
|