alainb:

I generally don't find much difference in speed between using the ODBC drivers and the native Oracle and MS SQL drivers, so I generally use DBD::ODBC.

What strikes me when reading your node is that you're reading some 100,000 records. Is that just for a performance test, or are you typically doing such large reads? I'll read entire tables when I'm creating data dumps for third party vendors, but generally my queries only send/retrieve a couple hundred rows at most. If you're frequently pulling multiple thousands of records from the database, you might be doing too much work in perl and not enough in SQL. Doing work in SQL can be a little tricky, but you can save quite a good bit of I/O that way, and possibly solve any performance bottlenecks.

Update: If you give a bit more detail about the task you're performing, and the types of queries you're doing, we might be able to give more specific advice.

...roboticus

When your only tool is a hammer, all problems look like your thumb.