What will the ultimate server-based anti-spam solution look like?

[spaceman]

Hi All

There's lots of pressure on website hosting services right now to provide quality anti-spam filtering. With the 'right' solution, it would negate or reduce the need for thousands of individual POP account holders to implement their own local anti-spam solutions. I therefore believe it's something to strive for.

The bottom line is that the perfect server-based anti-spam solution has to be *INCREDIBLY* easy to use. By that I mean that my clueless 67 year old Dad can use it. Key features would be:

1. POP account holders can EASILY choose to turn the service on or off.
2. POP account holders get sent a daily report showing a summary (To, Subject, Date/Time, Spam Probability) of whichs message were blocked. (Even better would be if they could choose the frequency of this report)
3. POP account holders can, WITH 1 CLICK, choose to release any of the blocked messages (that appear in the report above) so that they can be received.

The above, if done right (i.e. incredibly user friendly), would be wildly popular with our hosting customers.

Additional 'nice to have' features would be:
4. POP account holders and choose what % spam probaility should be the cut-off point for deciding what is and isn't spam (i.e. basic config options).
5. Learning (Bayesian filtering). I'm not sure if this would be practical on the server, because without some sort of integration with known email clients I can't imagine that an easy way exists of allowing individual POP account holders to communicate back to the remote server in regard to what is or isn't spam. If this were possible, then it would mean that their choices would have to be recorded in some sort of profile alongside their POP account details on the server, and this might mean a) excessive storage to keep all these profiles and b) excessive load on the server to process all incoming emails based on individual anti-spam profiles.

So there you have it. My 2c worth on the subject. We refuse to offer a 'full on' server-based anti-spam filtering service to our clients until the bare minimum functionality, combined with unbelievable ease of use, is available.

So if anyone knows of any solutions (for WHM/servers) that do *all* this (well, points 1-3) - I'm all ears!

Thanks for listening.

P.S. Our local ISP appear to have nailed it: http://www.westnet.com.au/products/additional/spam/ - I'll give them a call, but it's highly unlikely that their solution will port to our servers.

[rs-freddo]

My personal opinion is that the perfect spam solution is not server based. The only way a spam solution can be perfect is if it's client based. Perfect in my mind means not 1 single email is incorrectly dumped and that can only happen if the client gets ALL their email. Not many server admins agree with me on this - but I'm right anyway.

Thunderbird is very good, it's free and it's cross platform. It's as close to perfect as I've seen.

[spaceman]

Thanks for your comments.

At this moment in time I'm inclined to agree with you - the best anti-spam solutions are client-based.

But why couldn't a server-based solution have the same key characteristics as a client-based solution? You suggest that a 'perfect' system is one where the client gets all the mail (for local sorting into 'spam' and 'not spam' folders) - but if the server was temporarily holding back all the emails tagged as spam, and that list was very easily accessible by each POP account holder so that false positives could be (very easily) released - what's the difference? :-)

Another take on your point: if you believe that the perfect system can only ever by client-based, then should any attempt be made whatsoever on the server to block spam (i.e. the really obvious stuff)? Or should absolutely everything be passed straight through to the client?

The thing is, as a (small) web hosting provider, we as a company would love to be able to help our clients to reduce the volumes of spam hitting their inboxes. It's a support overhead for us right now to have to answer lots of spam-related questions (sometimes it feels like spam is our fault!). So if we can possibly arrive at a server-based anti-spam solution that satisfies the majority of our clients (and those that don't want it can turn it off), then this would represent a win-win situation.

Thanks again for your comments.

[Snowman30]

I find it ironic you mention Westnet's anti spam setup.

more than 50% of the domestic spam attacks on the servers i adminster, come from Westnet IP's