Just a very brief blog this Thanksgiving week (my favorite holiday of the year) – part 2 of last week’s relatively easy suggestions on how to make Oracle run as fast as possible on minimal computer setup. As before, my goal is simply to squeeze as much performance blood from the Oracle turnip as possible when deployed on limited capacity equipment. The one additional suggestion below only makes sense for environments where instance recovery is not a key issue – e.g. a demo notebook PC for traveling consultant types. So make sure not to use this unless you understand what the drawbacks are (discussed below) and can live with them.

While Oracle is running it is writing transactional history to the REDO LOG files such that if the instance is brought down unexpectedly, then the outstanding transactions can be played forward to restore to the state of the instance failure, and then rolled back for the un-committed transactions. If you’re running in NOARCHIVELOG mode, then the contents of the REDO LOG files get overwritten once the database transactions’ round-robin consume the allocated space – so database transactions can get lost (i.e. aged out) anyhow in this setup. So placing the REDO LOG files on the RAM Disk just adds one more scenario where data is aged out (and rather abruptly) – at computer shutdown.

 

So for purely demonstration or development environments, where raw speed is the most critical factor – this technique has merit. If it makes sense for your database computing needs, then give it a try …