April 27, 2017 at 9:04 am
When I run the following two queries. I get drastically different results. I don't understand why.
select o.name, p.object_id, i.name, p.index_id, p.index_type_desc, p.avg_fragmentation_in_percent, p.fragment_count, p.avg_fragment_size_in_pages, p.page_count
from sys.dm_db_index_physical_stats (7, NULL, NULL, NULL, 'DETAILED') p
JOIN sys.objects AS o ON p.[object_id] = o.[object_id]
JOIN sys.indexes AS i ON o.[object_id] = i.[object_id] AND p.index_id = i.index_id
where avg_fragmentation_in_percent > 0
order by avg_fragmentation_in_percent desc
VS.
select o.name, p.object_id, i.name, p.index_id, p.index_type_desc, p.avg_fragmentation_in_percent, p.fragment_count, p.avg_fragment_size_in_pages, p.page_count
from sys.dm_db_index_physical_stats (7, NULL, NULL, NULL, 'LIMITED') p
JOIN sys.objects AS o ON p.[object_id] = o.[object_id]
JOIN sys.indexes AS i ON o.[object_id] = i.[object_id] AND p.index_id = i.index_id
where avg_fragmentation_in_percent > 0
order by avg_fragmentation_in_percent desc
April 27, 2017 at 10:23 am
Did you Binoogle that DMV? Did you read Books Online about it??
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
April 27, 2017 at 10:45 am
One scans everything. The other only looks at the root page. Of course the results will be different. Read the documentation for more details.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
April 27, 2017 at 12:07 pm
I have read the documentation. But I don't understand it. I am a "involuntary DBA". And I have a hard time following most Microsoft documentation. I tend to look for forum or blog postings that translate it into a "For Dummies" version.
In my research, looking for an explanation I can follow, I came across this from Paul S. Randal, who has some legit credentials for his expertise.
https://www.sqlskills.com/blogs/paul/inside-sys-dm_db_index_physical_stats/
In here he says,
"
The DETAILED mode does two things:
"
So, then why would the fragmentation calculations be different between LIMITED and DETAILED? If you look at the fields I'm pulling, fragmentation is all I'm looking at.
April 27, 2017 at 12:09 pm
Barring an explanation I can follow...
Which should I use for deciding what indexes need reorg/rebuild?
Some sites suggest that Detailed is more accurate, but other sites just say it is different / not necessarily better.
April 27, 2017 at 12:18 pm
Paul is one of the greatest. You'll probably never go wrong reading his stuff and following his advice.
The difference is in the amount of data being accessed. The full scan of all pages gives you a more complete picture, while putting a much higher cost on the server. The limited scan gives you a less complete picture at a much lower cost. More often than not, LIMITED is adequate, if imperfect.
On a side-note, fragmentation is seldom a major issue for people. Usually spending time ensuring more accurate statistics works out better than attempting to eliminate fragmentation.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
April 27, 2017 at 12:32 pm
Okay. Thanks for pointing me in that direction. Stats mgmt was right after fragmentation on my "to-learn" list.
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply