CPU Co-Stop

David Klee discusses an important hypervisor-level metric:

VMware’s CPU Co-Stop metric shows you the amount of time that a parallelized request spends trying to line up the vCPU schedulers for the simultaneous execution of a task on multiple vCPUs. It’s measured in milliseconds spent in the queue per vCPU per polling interval. Higher is bad. Very bad. The operating system is constantly reviewing the running processes, and checking their runtime states. It can detect that a CPU isn’t keeping up with the others, and might actually flag a CPU is actually BAD if it can’t keep up and the difference is too great.

This is extremely useful information for DBAs in virtualized environments.  My crude and overly simplistic answer is, don’t over-book vCPUs on hosts running important VMs like your SQL Server instances.

Related Posts

Oddity With User Write Count In dm_db_index_usage_stats

Shaun J. Stuart looks at an oddity with the user_updates column on sys.dm_db_index_usage_stats: This pulls both reads and writes from the sys.dm_db_index_usage_stats dynamic management view. A read is defined as either a seek, scan, or lookup and a write is defined as an update. All seemed good until I noticed something strange. One of the top written to tables was, based on our naming convention, a […]

Read More

Safely Dropping Databases

Bob Pusateri notes a little issue when it comes to dropping databases: At a previous employer, we had a well-defined process when dropping databases for a client. It went something like this: Confirm in writing the databases on which servers/instances to be dropped Take a final full backup of databases Take databases offline Wait at […]

Read More

Categories

March 2016
MTWTFSS
« Feb Apr »
 123456
78910111213
14151617181920
21222324252627
28293031