Recent Updates

SpamButcher recently jumped all the way from version 1.8m up to 1.8s. The interim releases did exist, but were used for internal testing as opposed to ever being released.

The new version incorporates a lot of minor tweaks. Realistically, it's more of a 1.9 release. But doing a 1.9 release means we'd have to update a bunch of marketing material, so 1.8s it is.

Some significant improvements were made to both the accuracy of the filter, and general application performance.

SpamButcher now has an internal list of networks which have serious spam problems. Messages from those networks are not automatically filtered, but are evaluated with a little more scrutiny. This data is based off of publicly published blacklists, but has been heavily processed to determine which networks have only significant issues, and which are the worst of the worst. The filter understands the difference and behaves accordingly.

The spam mail filter is now aware of a whole slew of additional header oddities that may indicate a spam message.

SpamButcher uses a large internal set of rules to filter spam. This results in a lot of CPU utilization. The rule set has been growing in order to block more junk; unfortunately this has caused the processing needs to increase. As the general bandwidth of the internet has increased, spammers have been sending larger messages. This again compounds the problem.

A general code-review of some performance sensitive areas of the product was performed. The good news is that a lot of opportunities to improve were found. It turns out SpamButcher was performing a lot of disk reads that could be achieved a lot more efficiently by loading the same data into RAM.

So, SpamButcher now uses more memory, but still not much say compared to a web browser. The upside is that the email spam blocker now processes new spam about twice as fast as before. When reviewing existing messages, its about five times as fast.

Back