Comments on: Database Testing Continues http://feedlounge.com/blog/2005/08/14/database-testing-continues/ FeedLounge is a state of the art web-based feed reader. Mon, 13 Oct 2008 05:59:08 +0000 http://wordpress.org/?v=2.0.4-beta by: Dan http://feedlounge.com/blog/2005/08/14/database-testing-continues/#comment-273 Tue, 30 Aug 2005 04:26:32 +0000 http://feedlounge.com/blog/2005/08/14/database-testing-continues/#comment-273 why not Postgresql? why not Postgresql?

]]> by: Kevin Burton http://feedlounge.com/blog/2005/08/14/database-testing-continues/#comment-254 Mon, 15 Aug 2005 07:19:28 +0000 http://feedlounge.com/blog/2005/08/14/database-testing-continues/#comment-254 Another piece of advice is to test under page load. For example assume users are coming to your site while the DB has NOTHING buffered into memory. This way you can factor in worse case scenario. If you want to duplicate this you can cat 5G file (memory +1G) to /dev/null . This way you'll clear your filesystem buffer which is important for MyISAM. If you're on INNODB don't worry about it.. just restart mysql. This is SUPER important because as your data grows your tests won't hold up because the probability that data on disk is not in memory is high. This rises as you aggregate more but your memory footprint stays the same. Another piece of advice is to test under page load.

For example assume users are coming to your site while the DB has NOTHING buffered into memory. This way you can factor in worse case scenario.

If you want to duplicate this you can cat 5G file (memory +1G) to /dev/null . This way you’ll clear your filesystem buffer which is important for MyISAM. If you’re on INNODB don’t worry about it.. just restart mysql.

This is SUPER important because as your data grows your tests won’t hold up because the probability that data on disk is not in memory is high.

This rises as you aggregate more but your memory footprint stays the same.

]]>