Diagnosing and Resolving Spinlock Contention on SQL Server sqlcat.com/.../2011
Guys,
First a BIG THANKS for the document. I'm sure is gonna help a lot of SQL users understand the functionality/behavior of latches.
The document, on page 8 refers to sys.dm_os_latch_stats twice whereas it should have been refering to sys.dm_db_index_operational_stats.
Cheers,
Shankar
Pingback from SQL Server: Diagnosing and Resolving Latch & Spinlock Contention … « notes from a software alchemist
Thanks Shankar, I've flagged this with our documentation team, they'll edit the typo and republish. Glad you liked the paper.
Ewan
Pingback from New whitepapers on latches and spinlocks published | devblogging.com
this is an interesting article with some useful information for analysing latch contention. However, I think it the resolution techniques focus a little too heavily on physical DB reorganisation & could be balanced with more discussion on hardware optimisation, as this is very often a more practical approach than re-organising indexes, which very often isn't possible with vendor applications. The techniques demonstrated are certainly useful in some scenarios, but even with internal apps which you can control physical structures, changing the physical DB requires software testing (as you have recommended) where hardware upgrades require very little software testing and are often far, far more practical approaches. Eg, fixing poorly configured RAID, upgrading to SSD or increasing RAM, faster RAM, faster CPU etc. All up however, this is a useful article so thanks