RE: LIOs on INSERT?
- From: Martic Zoran <zoran_martic@xxxxxxxxx>
- To: Patty.Charlebois@xxxxxxxxxxxxxx
- Date: Fri, 11 Feb 2005 03:58:35 -0800 (PST)
Ok then. My simple formula was OK, just there are missing pieces :) Does this mean all your indexes are level 2 (if root is level 0)? Please, describe the process of doing insert. How are you doing inserts? many inserts in one transaction? insert then commit? Are you doing thing like connect do the insert, disconnect? How many concurent processes are doing inserts? How many different processes are doing inserts? Also very important is are all your inserts are successful inserts. Failing inserts are reading more blocks because of the rollback, then in average you can get 18 even you should have around 10. I will not forget LIO for the rollback block too. Possible LIOs for the free list management. Parsing can also add a few blocks to LIO overall number. This is why is important how are you doing inserts. But to come to the overall picture: Why are you worried about reading 18 blocks instead of 10 or 12? Is it because of CPU usage for that particular operation and you do not have enough CPU to support app? Or is it because reading 18 blocks instead of 12 are making bigger scalability issues and you have builidng waits because of it? Which waits are these? Or is it just a curiosity? The same thing that drives me to spend time thinking about this issue. Regards, Zoran --- Patty.Charlebois@xxxxxxxxxxxxxx wrote: > There are 18 LIOs per insert. > There are 3 indexes on the table, none are > functional. > > Using you formula I should have 10 LIOs per table so > yes, my number is bigger than the simple > calculation. __________________________________ Do you Yahoo!? Meet the all-new My Yahoo! - Try it today! http://my.yahoo.com -- http://www.freelists.org/webpage/oracle-l