Comments on: Clustered Indexes Have Statistics Too

By: Grant Fritchey

Grant Fritchey — Tue, 18 Dec 2012 10:47:56 +0000

Hello Asit,

It’s hard to say what the issue might be without seeing the plan. But usually when the costs are different than you expect, either higher or lower, it goes back to the estimates generated from the statistics. If your statistics are out of date, or the values passed to the parameter when then query was compiled referenced different statistics, then you may see the costs being different.

By: Asit kaushik

Asit kaushik — Mon, 10 Dec 2012 13:25:59 +0000

Hi Grant,
Again in the execution plan i have a complex query and my question/problems are

1) in the where clause as shown in the below code snippet the @variable is passed as a parameter to the Stored Procedure

@variable = 0
or
@variable = and (select…)

my @variable passed is 0 but in the execution plan i see the cost taken up by the other select query , i cant understand why..

2) In the execution plan i see that some operation say index seek has been executed = 0 then also i am seeing a high cost .

Please guide me if i my understanding is faulty.

Regards
Asit

By: Grant Fritchey

Grant Fritchey — Fri, 09 Nov 2012 17:41:26 +0000

You can’t really determine if a query is running well just by seeing the costs. Those are estimates. You do use them to figure out which operators to look at, but it’s not a measure of performance. For that you capture execution time and the I/O.

By: ASIT KAUSHIK

ASIT KAUSHIK — Fri, 09 Nov 2012 06:32:24 +0000

Hi Grant,
I have been reading your blogs,articles, session pass free session. I just wanted to ask if i am analyzing an execution plan the most costly block has a Clustered seek and that cost is around 50-60 % is that query performing at the optimal and does not require any further change. IF No what other factor should one look for. Hope this question doent suckssss