View Full Version : Spam Filtering & SURBL

07-06-2005, 08:27 PM
I will be signing up with Fluidhosting for a shared account, and have a doubt about spam filtering here at FH. I learned from sales reps that FluidHosting does not support SURBL (checking the URLs within an e-mail to see if those URLs are blacklisted) with spam assassin.

I have been hearing from many people that SURBL is a must these days, and important amount of spam are not being caught since spammers learned how to change content such that spamassassin will not catch them (i.e. by playing with words, righting sch00l instead of school type of tricks), hence other RBLs are not that effective anymore? But, since spammers usually promote a web site, they will have to give the correct URL, and that is why SURBL is considered to be more effective.

Is this the case? Can you please provide a rought estimate regarding percentages of false positives/negatives? And/Or share your experience in regards to the effectivity of spam filters here at Fluid Hosting?

07-06-2005, 08:37 PM
In my experience much of the spam that I see get through, doesn't even have a URL in it. It's typically the kind that has an image attached to the message with no real content.

07-06-2005, 08:39 PM
Also, in the future we may add SURBL, and other systems to our filtering. At the time when we attempted to first implement it, it caused too much of a delay, and load on the servers. However, if things look like they be done more efficiently, we'll certain consider readding them.

We do run a modified set of SA rules, such as applying a much higher score to messages sourced from servers listed in Spamhaus, and Spamcop.

07-08-2005, 03:39 PM
Hi again,

As I mentioned before, I think SURBL is a very important tool and SURBL catches considerable amount of spam that other filters are failing.

As for performance/stability issues, I have seen that many reputable hosts implemented this recently. For example, pair, who is known to be very resistent to this type of things and specifically SURBL, has implemented or currently in the process of implementing this. So, my guess is that SURBL filters are in a much more stable state right now.

That said, I would appreciate it if you can look into this when you have some time, and am pretty sure that it will prove to be very useful for all users.


07-08-2005, 03:48 PM
I'm certainly evaluating it. Stability hasn't been much of an issue, it's been resource usage. Last time I tried implementing it, it caused massive increase in resource usage.

07-08-2005, 04:59 PM
Thanks for your replies and considering this...

07-09-2005, 10:04 PM
Just to let you know, I've begun testing usage of SURBLs.

07-13-2005, 02:59 PM
Thanks for the update and hope it turns out well for all;)