We are currently running GFI ME on our front-end server to catch spam but are not to pleased with the results of the product (too much false positives and false negatives, not possible to create personnal black/white lists, …)
Exchange IMF looks like the easiest solution to implement but is not supported on a cluster.
So we are looking for somehting that is
– working with clustered exchange 2003 servers
– multi language sensitive (not all Russian e-mails are spam)
– learing by uses input
– allow users to create their own whitelists (for e.g. newsletters)
– allow users to check the spam they received
– return a custom NDR for all possitive messages
I thought about Barracuda but that is not quite cheap.
Any suggestions?