January 28, 2008 at 1:34 pm
Hello All,
I am running SQL Server 2005 SP2 Standard Edition on a few servers. I run SQL Agent Jobs every week that REBUILD all of the user table indexes i.e. (ALTER INDEX ALL ON [Table] REBUILD) I also run another job that UPDATES STATISTICS. The jobs complete without error and rebuild most all of the indexes removing fragmentation then updating any statistics. Though, there are quite a few indexes that even after running the rebuild script still have the same fragmentation % i/e 67% or fragments as prior to the REBUILD command. Are there some reasons why indexes will remain fragmented even after rebuilding them? I've tried as a test on some tables to rebuild the index in question from SSMS, SQL Agent Job, and the GUI point and click "Rebuild", all results are the same, command competed successfully, but fragmentation remains the same, no change.
Any tips or pointers would be greatly appreciated! There must be some caveat I'm not thinking of..
January 28, 2008 at 1:38 pm
How big are the tables in question? How many pages do the indexes take up (visible in sys.dm_db_index_physical_stats)
Edit: btw, a separate update statistics is not necessary (and may even be counter productive) if you're doing an index rebuild. When the index gets rebuilt, the stats get updated as if an update stats with full scan was run.
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
January 28, 2008 at 1:50 pm
Hi Gail,
There are quite a few examples but I will stick to one or two. The others are very similar, and / or only average 1 fragment per page.
Stats from the DMV
EXAMPLE
Table Data Size .516 MB
Index Space .719 MB
Table record count : 7012
# of Pages : 14
# of Fragments 8
Avg Frag 50%
EXAMPLE
Table Data Size .039 MB
Index Space .016 MB
Table record count : 568
# of Pages : 5
# of Fragments 5
Avg Frag 80%
January 28, 2008 at 4:46 pm
See if this helps...
http://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/ss2kidbp.mspx
[font="Verdana"]EKD[/font]
January 28, 2008 at 7:45 pm
These indexes aren't getting touched - you basically shouldn't bother with index fragmentation unless your index has more than 1000 pages as it'll most probably be entirely contained in cache already and so fragmentation is irrelevant. I thought I'd put that into BOL for DBCC SHOWCONTIG and index_physical_stats but it looks like I didn't.
Checkout this blog post on rebuilds and stats updates - http://www.sqlskills.com/blogs/paul/2008/01/27/SearchEngineQA10RebuildingIndexesAndUpdatingStatistics.aspx. Weird how questions seem to come up in clusters - this week it's index rebuilds.
Hope this helps.
Paul Randal
CEO, SQLskills.com: Check out SQLskills online training!
Blog:www.SQLskills.com/blogs/paul Twitter: @PaulRandal
SQL MVP, Microsoft RD, Contributing Editor of TechNet Magazine
Author of DBCC CHECKDB/repair (and other Storage Engine) code of SQL Server 2005
January 29, 2008 at 8:47 am
Thanks, yeah I wasn't worried about performance reasons, as your right there isn't much to the table. It was a curiosity question, as there must be a reason ie size of table, total number of data pages why they were not being touched. The ones that are critical were being rebuilt..
The reason we do the statistics FULL SCAN afterwards is to catch the other statistics that were user created or auto created by the engine in some situations that were not part of an index column. On our databases we have a good maintance window so time is no issue and it really only takes 5-10 min for rebuilds and statistics as a high number. Sometimes we find that one of those statistics should be made into an index and other times they are better served as just statistical info for the engine.
January 29, 2008 at 2:19 pm
GilaMonster (1/28/2008)
How big are the tables in question? How many pages do the indexes take up (visible in sys.dm_db_index_physical_stats)Edit: btw, a separate update statistics is not necessary (and may even be counter productive) if you're doing an index rebuild. When the index gets rebuilt, the stats get updated as if an update stats with full scan was run.
Gila, I was always under the same impression until I had several phone conversations with Microsoft revolving around performance issues we experienced. Two different SQL Server support techs informed me that issuing a DBCC DBREINDEX, ALTER INDEX REBUILD/REORGANIZE, will NOT update the statistics with fullscan. They recommended having a separate task in our maintenance scripts to update the statistics.
Now I'm curious ... It wouldn't be the first time Microsoft has given me faulty information.
January 29, 2008 at 4:34 pm
The quality of support has gone way down - I used to constantly battle incorrect support engineers (a lot easier to do when you're running one of the dev teams they're giving incorrect info about).
They're wrong - and that's pretty annoying. Trust me - I used to own all that code. Reorganize doesn't but rebuild does update stats.
Paul Randal
CEO, SQLskills.com: Check out SQLskills online training!
Blog:www.SQLskills.com/blogs/paul Twitter: @PaulRandal
SQL MVP, Microsoft RD, Contributing Editor of TechNet Magazine
Author of DBCC CHECKDB/repair (and other Storage Engine) code of SQL Server 2005
January 30, 2008 at 12:05 am
I also face the similar kind of problem:
When the table size is very small and contain a few rows only at that time this kind of problem happen. In addition to this when there is heavy insert/update/delete happen to the database i.e. few transactions per second at that time also you won't get Scan Density 100%.
One shouldn't bother much about the Scan Density on the small table.
---------------------------------------------------
"Thare are only 10 types of people in the world:
Those who understand binary, and those who don't."
January 30, 2008 at 1:36 pm
Adam Bean (1/29/2008) Two different SQL Server support techs informed me that issuing a DBCC DBREINDEX, ALTER INDEX REBUILD/REORGANIZE, will NOT update the statistics with fullscan. They recommended having a separate task in our maintenance scripts to update the statistics.
Paul Randal (1/29/2008)
Trust me - I used to own all that code. Reorganize doesn't but rebuild does update stats.
And it's very easy to prove, if you want to see for yourself Adam.
Run DBCC Show_statistics on an index. Check the date the stats were updates. Rebuild the index, then run show statistics again.
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
November 7, 2012 at 1:42 pm
Check the MDOP (Max Degree of Parallelism) setting. In the past, I have seen improvement when MDOP is 1 instead of default 0. My understanding was, when parallel threads try to rebuild an index, there would still be fragmentation simply because more people at work, however if there is only 1 thread doing the work, then we would get the optimum results out of ALTER INDEX...REBUILD
ALTER INDEX <index-name> ON <table-name> REBUILD WITH (MAXDOP = 1)
The details are explained in detail in the blog post:
November 7, 2012 at 11:46 pm
Please note: 4 year old thread.
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
Viewing 12 posts - 1 through 11 (of 11 total)
You must be logged in to reply to this topic. Login to reply