April 3, 2008 at 5:40 am
Have you taken the insert query and reviewed execution plan?
Does the indexes get dropped prior to loading?
Thanks,
Phillip Cox
April 3, 2008 at 9:57 am
IO problems can definitely occur without disk queuing. This is easy to understand when you take into consideration the fact that SQL Server has an internal workload scheduling component - SQLOS - which slows down disk requests when workload is backing up.
You really need to measure disk IO from within SQL Server using fn_virtualfilestats(). I have a script which simplifies use of this TSQL system function here:
http://blogs.sqlserver.org.au/blogs/greg_linwood/archive/2006/03/15/59.aspx
Another explanation as to why your disk IO might have slowed down markedly is that you might have a disk failure & the array controller is bringing in a hot spare?
Regards,
Greg Linwood
October 15, 2010 at 7:29 am
I dont know anything about the EMC hardware, but if they use battery pack on the array, check if the battery has failed. If the battery fails, the array controller will not allow writes to cache, generally causes a massive performance problem all of a sudden.
Viewing 3 posts - 16 through 17 (of 17 total)
You must be logged in to reply to this topic. Login to reply