Audit function in 3 tier ?

[SteveBrett]

Hi,

I've got the wonderful job of building audit functions for a large, high volume database (postegresql of course) and was wondering if anyone had any experiences they would like to share ?

Questions such as:

1. How _exactly_ does ppostgresql accept queries from external sources ? Is there some way of identifying which user sent in which request ?

2. Should audit functions live in the middle tier or be implemented as triggers at the db level ?

We might be using a pool of connections so would we might not have specific database users logging in and so we still need to know who is doing what etc...

thanks in advance,

Steve