Performance Issues w/ Hyper-V VM

We have an ETL process that kicks off at 4am every Sunday Morning.  This process used to take 2 - 2 1/2 hours to run, not it is taking 5-6 hours.  This issue seemed to start after upgrading the VM from Virtual Server 2005 R2 to Hyper-V.    Months ago we were trying to fix the SAN performance issues by upgrading the VM, but that didn't fix the problem.   The Hyper-V VM is running 4 procs, 6 gig ram and using high speed LUN's.  The ETL job is kicked off as a MS scheduled job.  The VM is running Server 2003 Standard R2 Standard SP2 w/ SQL server 2005 sp2.  When monitoring the process as it is running, the CPU usage and RAM usage is minimal to moderate,  never even close to utilizing the full CPU or RAM usage.   There is a massive amount of I/O taking place, which is typical in an ETL process.  So the problem seems to be through-put, however, we cannot find a solution.   We are using high-speed luns on the SAN.  We also added the /3gb switch to the startup.   Actual CPU utilization is hovering from 30 to 65%, ram usage is minimal.  TempDB is getting hammered and disk load is maxxed out.   The I/O wait states are very high.  


66chawgerAsked:
Who is Participating?

Improve company productivity with a Business Account.Sign Up

x
 
plimpiasConnect With a Mentor Commented:
Hello,

A couple of things you can do with your performance issue in Hyper-v

First make sure your VMs is set to fixed disk, especially if your using IO alot. Fixed disk improvements performance a lot.

Second, make sure you are using the latest integration tool version and that you are running SP2 on your 2008 server/cluster. After the SP2 installation, reinstall the integration tools

third, make sure antivirus exlusions are set correctly on your HOST and VM

fourth if using SCVMM make sure you are using synthetic drivers and not emulated. To check open up the hardware properties of the VM, go to networking and look and see above. If it says emulated remove it and add a new synethic driver.
0
 
66chawgerAuthor Commented:
Additional Info: 2 - 4Gbps fiber connections to the Pillar (SAN)
0
 
puranik_pCommented:
>Tempdb is getting hammered
Is it configured optimally? with lot of file size and 1 data file per CPU?
0
 
66chawgerAuthor Commented:
Yes, it is configured optimally with data files spread across 4 cpu's
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.