Condusiv V-locity VM Testing

9th July 2013 1 By Simon

image

During the recent vForum event at Wembley I had an interesting meeting with Condusiv Technologies (formally Executive Software\ Diskeeper).

They were showing off the latest release of V-Locity VM.

V-locity VM is classed as a Performance Accelerator which offers benefits such as :-

  • Increase application performance by 50%
  • Reduce latency by 50% for faster file access
  • Eliminate bottlenecks without expensive hardware
  • True "set and forget" management

    What struck me was the performance potential for VM’s running on potentially limiting storage (it happens, in fact our storage at work is about on par with the performance one of our guys get’s from a Synology Diskstation 1513+ and yes we have fine tuned and tweaked).

    I had a chat with their Technical Director and asked about an NFR or Evaluation license to allow me to do some testing in my home lab to see what kind of performance improvements can be gained from running V-locity VM in my environment.

    It should be noted that the V-locity software is limited to the Microsoft platform of OS’s starting with Windows XP (SP2) all the way up to Windows 2012, it also supports both vSphere (ESX\i 4.0 or later) and HyperV as the host virtualisation platform.

    Lab Setup

    Host Hardware

    ESXi HOST: Shuttle XPC SH67H3, 32GB RAM; 1x i7 3770 @ 3.4ghz with ESXi 5.1u1 (5.1.0 build 1021289)

    STORAGE PLATFORMS

    Synology: DS1513+

    Homebrew NAS\SAN solution: 8gb ram, 6 x 1500gb and various USB Keys for FreeNAS., OpenFiler, Nexenta Community Edition, OpenMedia Vault.

    Iomega IX4-200d: 8TB NAS, 4x 2tb

    NETWORKING

    NetGear TS724T 24 x 1 GB Ethernet switch

    Cisco SRW2016 16x 1GB Ethernet switch (dedicated storage switch for the Synology unit).

    IO Test Machines

    OS TYPE: Windows 7 x64 (fully patched) on ESXi 5.1u1 using a 40gb thick eager zero provisioned disk

    CPU Count \ Ram: 1 vCPU, 4GB ram

    This was a template VM with full updates applied to the image and then deployed out. One guest was running V-locity 4 and Iometer via a PowerShell script whilst the second guest was just running Iometer via the PowerShell script, no other VM’s were active on the host or storage during testing.

    The power plan was set to High performance and the screen saver turned off.

    Virtual machine names:

  • Win7A (V-locity VM installed)
  • Win7B

    Current VMware tools were installed onto the template and running.

    IO Testing Software

    All testing was carried out using IO Meter software and test scripts provided by Condusiv. Testing was carried out over 4 days to allow the V-locity Benefits Analyser to run to completion successfully (3 phases of testing – Analysis, Improvement and Post).

    The IOMeter test script can be downloaded from here and the IOMeter.ps1 PowerShell file can be downloaded from here. No deviation to the testing script occurred.

    Initial Performance Results – FreeNAS Server

    Four days of constant testing were carried out and 192 sets of results were received. The following results were taken after the three phases used by V-locity Benefit Analysis tool and show what happens when you have a VM using V-locity in full flow compared to a VM without the V-locity software installed.

     image   image 

    Run 145\192  – V-locity running on the left hand side, no V-locity running on the right hand side.

    image   image

    Run 146\192 – V-locity running on the left hand side, no V-locity running on the right hand side.

    We can see from the above results that having V-locity running on one of the VM’s has greatly improved the average iops when compared to a VM with no V-locity running.

    The results of the Benefit Analysis show similar improvements.

    image

    image

    image

    As you can see in the above screenshots running V-locity offers some very attractive performance increases and certainly interests me from a performance stand point.

    I will be carrying out more testing in the home lab environment over the next few weeks to offer up some more performance testing results for various different storage solutions (listed above but if you want more tested please let me know) but what I can say is that this warrants further investigation and I do recommend carrying out performance testing yourself.

    Request a link for the evaluation