|
|
Posted by Erland Sommarskog on 10/22/74 11:33
loosecannon_1@yahoo.com (loosecannon_1@yahoo.com) writes:
> This tells me that some caching must be involved since the times are so
> different between identical queries but I would expect that one of the
> queries would cache and thus take longer but the other 10 would be
> fast, not all block for 2 minutes. What is more puzzling is that this
> behavior didn't occur before where now the only differences are:
Well, if the is not in the cache but must be read from disc, it is not
the case that one process actually gets the data from disk and runs for a
long time, whereas the other gets the data out of thin air in lieu out of
a cache.
Exactly what happens, I don't know, but I don't really think that there
11 requests for each data page. Nevertheless, all physical must complete
before the queries can complete.
120 seconds sounds a wee bit, though.
One way to study the issue is to run one of the queries from Query
Analyzer when the cache is empty with SET STATISTICS IO ON and watch
physical IO.
It is also worth testing the impact of running a lesser number of the
queries, and see what happens in this case.
DBCC DROPCLEANBUFFERS can be used to flush the cache.
--
Erland Sommarskog, SQL Server MVP, esquel@sommarskog.se
Books Online for SQL Server 2005 at
http://www.microsoft.com/technet/prodtechnol/sql/2005/downloads/books.mspx
Books Online for SQL Server 2000 at
http://www.microsoft.com/sql/prodinfo/previousversions/books.mspx
Navigation:
[Reply to this message]
|