August 5, 2012 at 1:26 pm
Hi all, i cant think about an apropieted subject for the topic.
Here is the thing.
We have a Server with this specs
Windows Server 2008
8 x Intel Xeon L7555
32 GB Ram
RAID 5 with SAS disks. poweredge storage-
In the server we have
SQL Server 2008
17 GB
5 Cores
15 GB for windows only.
The question is, can SQL server, in some way, affects the performance of all the storage.. in other words..
When we test the storage via Performance Monitor and HDTach, we have 22 MB MB/s for average read (disk) this is ridiculous becouse the hardware we have.
I ask here because we are encharged of sql server, no the hardware of the server, and i'm not a specialist hardware guy, but you dont have to be an especialist when see something like this.
I investigated performance monitor, with some Windows Server counters and I reach to the conclusion that we have a serious bottleneck problem.
Sorry for my out of practice english.. and what do you think about this (if the server is affected by disk performance)
Thanks !!!
August 6, 2012 at 8:10 am
gheinze (8/5/2012)
Hi all, i cant think about an apropieted subject for the topic.Here is the thing.
We have a Server with this specs
Windows Server 2008
8 x Intel Xeon L7555
32 GB Ram
RAID 5 with SAS disks. poweredge storage-
In the server we have
SQL Server 2008
17 GB
5 Cores
15 GB for windows only.
The question is, can SQL server, in some way, affects the performance of all the storage.. in other words..
When we test the storage via Performance Monitor and HDTach, we have 22 MB MB/s for average read (disk) this is ridiculous becouse the hardware we have.
I ask here because we are encharged of sql server, no the hardware of the server, and i'm not a specialist hardware guy, but you dont have to be an especialist when see something like this.
I investigated performance monitor, with some Windows Server counters and I reach to the conclusion that we have a serious bottleneck problem.
Sorry for my out of practice english.. and what do you think about this (if the server is affected by disk performance)
Thanks !!!
1) Can you simply turn off SQL Server and rerun your IO tests? If SQL Server is doing active work it can DEFINITELY cause an IO test to see poor numbers!! You can completely overwhelm virtually any IO system with SQL Server.
2) Did you test this hardware BEFORE you put SQL Server on it (you should have).
3) Even if you have just 3 disks you probably should get more than 22MB/sec of throughput - certainly for sequential reads. But I have no idea about HDTach, nor did you provide what type of IO you were doing. Was it 4K random writes or 64K sequential reads or ??? Random writes with 3 disks in a RAID5 could really suck...
4) I cannot provide any guidance or recommendations without much more details about your actual IO configuration and the nature of of the tests you were doing.
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
August 6, 2012 at 8:54 am
I second Kevin's suggestions. Make sure you're doing your benchmarking with SQL Server (and anything else non-standard) turned off.
HD Tach may not be the best tool for the job. According to Wikipedia:
As of September 2011, the latest version of this application (3.0.4.0) is still not fully compatible with Windows Vista or Windows 7. However, HD Tach works in Vista and 7 64-bit by running it in Windows XP SP2 compatibility mode.
And...
On December 5, 2011, citing the lack of time to devote to the project, Simpli Software formally announced on its website that HD Tach has reached end-of-life and is no longer being supported.
Have you looked at SQLIO? It too has it's shortcomings but was specifically designed to do the kind of benchmarking I think you're looking for.
August 9, 2012 at 11:31 am
Thanks for all your replies, we decided to reinstall and upgrade to windows server 2008 R2 and make some configures changes on the storage.
thanks a lot
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply