C
ChUIMonster
Guest
The secret command is called "proshut", that will cause all dirty buffer to be written to disk. After the subsequent "proserve" you will have a buffer pool with a minimal footprint of used buffers. The achieve your goal of consistently obtaining the same disk io profile you will also need to unmount & remount the filesystems. If the disk subsystem is a SAN you probably need to power it down and then restart it. And make sure that nobody else is actually using it. That's probably a lot of work and it may not be worth it. A simpler approach would be to look at the table stats and the "os reads" field on _tablestat. If the table in question shows more disks reads during the busy period that's a pretty good sign. You could do the same with _userio & "_dbread" and the session in question if you need to focus on that specific process rather than the table. Or you could just download ProTop and let it do all of that for you. dbappraise.com/protop.html
Continue reading...
Continue reading...