Hey guys, I have been doing more thinking and found some more problems/ideas with the corpus.

The problem I have found is that while I run 2 spam filter servers, one just being redundant in the event the main one goes down, it gets very little mail because the main server is up, and what mail doesn come through is 99% spam.

So if the main server does go down basically we loose the entire functionality of the corpus.

Here's my idea to the solution.

I know Roberto said SQL etc is not fast enough, so what if the corpus was stored/synced to the spamfilter database and each spamfilter server cached a copy of it locally? Have a setting to allow users a duration to have spamfilter sync to the database....

That would solve all of our problems for people who run redundant spamfitler servers.

Is this possible?

Great idea -- I would like my two SF servers to share information on the corpus stats also.  It seems that this would ge a great canidate for XML?

Dwight