Progress slower on VM then physical machine

Posted by cbr@organi.be on 09-Dec-2011 05:01

Maybe someone can help me.

We have our progress application  running on a windows 2003 fysical machine (5 year old). Now we migrated  it from scratch to a win208 R2 virtual machine under Citrix xenserver  5.6.

When we do performance tests, by reading x-number of records  out of the database, we experience slower performance on the new VM then  on the old physical machine.

For example: reading 150000 records  on the physical machine takes 3-4 seconds, on the VM the same test takes  9-10 seconds. Anyone experienced this also?

Regards,

Chris Brosens

ICT System Engineer.

All Replies

Posted by Thomas Mercer-Hursh on 10-Dec-2011 14:22

Well, there is a bunch of information missing here.   What is your version of Progress, for starters since a fairly likely cause with any older version of Workgroup is that you have moved from a single CPU machine to a multi-CPU machine.

You should also tell us something about the configuration, particularly the disks.  Any RAID involved?

Note that, all else being equal, I would expect running under a VM to be, at best, close to not running under a VM.  I presume the reason you are surprised is because the VM is running on some hot new box and the physical machine is an old one.

Posted by cbr@organi.be on 12-Dec-2011 05:24

Dear Thomas,

Here some extra info o n our configuration:

*We are running Progress 10.1C with SP4.

*Regarding RAID config, we are using a virtual disk on a HP SAN P2000 in RAID1.

*Server has 4 virtual CPU's .

Regards ,

Chris

Posted by Tim Kuehn on 12-Dec-2011 07:18

cbr@organi.be wrote:

*We are running Progress 10.1C with SP4.

*Server has 4 virtual CPU's .

For the progress version - is that the Workgroup or enterprise version?

Could you clarify what you mean by "virtual CPU"?

Posted by cbr@organi.be on 12-Dec-2011 07:25

Dear Tim,

The server has several physical processors with hyperthreading activated. So in a virtual environment you have more virtual processors available.

The server that runs our database has 4 (virtual) CPU's.

For the version, it is Enterprise edition.

Regards,

Chris

Posted by stefan.lang@havilog.com on 12-Dec-2011 09:45

I fear you have to go to some benchmarking and inquisition rounds with your VM gurus

  • Are any other virtual machines on the same physical host, sharing Memory, bandwith or CPU? Any Bottleneck?

    If yes, then inqure there. Measure on VM AND host VM (if you have HP UX)
  • Are the disks of equal speed (VM and SAN can be nasty here)
    • Make some low level tests of creating and editing a bunch of big files on both systems, does it differ in performance?
    • If necesary use a HD Benchmark tool
  • Make sure that your SAN guys do not give you disks which are shared with other applications
  • Is teh setup of the database identical if you compare distribution of files  (BI. AI, etc.) and areasto disk spindles?

    SAN people sometimees make strange simplifications (assuming SAN is allways fast and scale linear :-) )

Posted by gus on 12-Dec-2011 12:54

So far, we know that you changed to a virtual machine environment and are using a different operating system and you see that database read performance is slower. Without more information it will be difficult to guess the cause. One of many hardware or configuration differences could be responsible. There are many things that could be wrong. Are you using the same computer now or a different one? If different, how are they different? Same database? Same storage? Same network? Same network cards?

John Harlow of Bravepoint did a session on virtualization at one of the conferences. You may find that helpful. You can find his slides here: http://communities.progress.com/pcom/docs/DOC-15823 .

Look for the one called OPS-4 Fun with Virtualization.

Posted by jquerijero on 24-Jan-2012 13:14

How many VMs are running on the machine when you did your testing? Did you do your testing when you only have the VM that has the database?

This thread is closed