Re: SQL Cost Compared To Elapsed Time
- From: "Connor McDonald" <mcdonald.connor@xxxxxxxxx>
- To: "Oracle Discussion List" <oracle-l@xxxxxxxxxxxxx>
- Date: Thu, 20 Apr 2006 21:39:39 +0800
On 4/20/06, Tanel Põder <tanel.poder.003@xxxxxxx> wrote: > > Interesting idea :) > > One thing is that if you find a query with high elapsed time to cost unit, > it doesn't necessarily mean that the execution plan itself is suboptimal, > maybe just other factors are affecting the execution, e.g. CBC latch > contention, very little cached blocks etc.. > > Tanel. > > The other thing is that if any SQL runs bad, it doesn't mean the plan is wrong. A "crappy" plan is optimal, if the alternative plans available are "more crappy", "very crappy" and "extremely crappy". Cheers Connor -- Connor McDonald =========================== email: connor_mcdonald@xxxxxxxxx web: http://www.oracledba.co.uk "Semper in excremento, sole profundum qui variat"