We had a JDBC adapter that was set to use “no pooling.” We changed it to use connection thread pools, with a minimum pool size of 0. The number of accesses to the subject database (selects, inserts and updates) immediately dropped by over 50%. We’ve received no reports of undelivered documents. Can anyone explain why this might have happened?
We should probably sync up on terminology first. There are thread pools and there are connection pools, but I’ve not heard of “connection thread pools.” In case I’m missing something, can you clarify?
By “number of accesses” do you really mean the count of selects, inserts, etc.? Or do you mean the number of connections?
I think what you may be seeing is simply a fewer number of physical connections–your IS is still doing the same work, it’s just sharing the (fewer number of) DB connections amongst the threads. Before, it created new connections when needed. Now it reuses connections that are idle in the pool.
Let me know if I’ve missed your point.
According to the owners of the database we’re connecting to, the change is to the actual counts of inserts, updates and selects. Unfortunately, they don’t have detailed data prior to our making the change, so we don’t know which of the 3 database transaction types changed, or if they all changed equally.