January 27, 2011 at 3:00 am
Has anyone had this experience. We have a quit heavy report that took about 15- 17 seconds to run in Reporting services on a SQL 2005 server, now when we moved it SQL 2008 R2 it takes approx. 45 – 50 seconds to run the report. The underlying procedure runs almost the same time on both machine approx. 3 seconds. Reporting services is installed on the same server as the database.
SQL2005 runs on a 64-bit dual processor CPU machine 16 GB of memory. SQL 2008 runs a Virtual server with 64 bit dual processor with 16 GB of memory.
Any ideas?
January 27, 2011 at 3:15 am
So the old system was physical and the new system is virtual?
What about the disks where the data, temp and SQL (instance directory) are installed? Have the change config as well?
January 27, 2011 at 4:35 am
Yes the old system where physical and the new virtual
The disks are called RAID10SAS, all the databases incl. temp, reporting services and the data is installed on the same drive
There haven't been any other config changes
January 27, 2011 at 6:56 am
What else is running on the virtual server that the 2008 VM is running on? Do you actually need a dual processor VM? I've seen issues with VMWare where a load running on a dual Processor VM allocated 2500 MHz of CPU ran slower than on a similarly configured Single CPU VM allocated 5000 MHz of CPU. IIRC, the reason for this is that the VMWare scheduler needs to wait for physical 2 processors(cores) to be available prior to scheduling the CPU resources if the VM was defined as having 2 Processors, where a single processor VM even though it's using more or the same CPU resources the scheduler can spread it out better.
Also, check to make certain that the server hosting your VMs doesn't have the processors set to a low power or power saving mode. @BrentO has a bit more about that here[/url].
-Luke.
January 28, 2011 at 3:07 am
We changed to one CPU without any effect, it didn't get better or worse. The power saving stuff is turned off
January 28, 2011 at 6:31 am
As a test to see if it's really a 2005-2008 issue or something else, you could do a P2V on your 2005 server and create a VM that's on the same host as your 2008 server. Then run all of your tests. That would be the best way to really get the hardware out of the way as a possible culprit (unless of course there are different IO paths (different LUNS, different Arrays etc...) and such.)
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply