Comments on: Estimated rows, actual rows and execution count

By: Gail

Gail — Wed, 14 Oct 2009 12:38:34 +0000

Nested loops, key/RID lookups, table and index spools.

Nested loops always will show this, unless there’s only one row in the outer table. Key/RID lookups also unless there is only one row that needs the lookup

By: chris stobbs

chris stobbs — Wed, 14 Oct 2009 12:29:22 +0000

great article Gail. This is something I will look out for going forward. Am I correct in saying the nested loops will more than often create this kind of result?

By: Bookmarks for September 25th | Brent Ozar - SQL Server DBA

Bookmarks for September 25th | Brent Ozar - SQL Server DBA — Fri, 25 Sep 2009 14:55:03 +0000

[...] Estimated rows, actual rows and execution count – Another one of Gail Shaw’s excellent posts about query execution. [...]

By: Gail

Gail — Wed, 23 Sep 2009 18:42:21 +0000

It got the estimates for both the inner and outer tables right.

Outer table, estimated = 101 rows, 1 execution, actual 101 rows. Inner table estimated 1 row, execution count 101, actual rows 101 total for all executions so actual 1 row per execution. Exactly what the estimate was.

By: Brian Tkatch

Brian Tkatch — Wed, 23 Sep 2009 13:04:42 +0000

So, there is no problem with the example though. Because SQL Server got the 101 correct on the outer TABLE it picked the appropriate plan?

By: Pinal Dave

Pinal Dave — Wed, 23 Sep 2009 07:21:57 +0000

Thanks Gail,

Now my results are same as yours.

By: Gail

Gail — Wed, 23 Sep 2009 06:33:46 +0000

You got different results because I left two indexes out of the table creation code. *embarrassed*

By: Pinal Dave

Pinal Dave — Wed, 23 Sep 2009 06:18:44 +0000

Gail,

This is great. I got very close to results which you have got in blog post. I have sent you email with screenshots.

Kind Regards,
Pinal

By: Grant Fritchey

Grant Fritchey — Tue, 22 Sep 2009 11:40:35 +0000

Excellent point, and one that most people, including me, forget about or fail to mention. Great post.