Oracle Enterprise Linux 7 Guest On Server 2012 R2 Hyper-V Cluster RRS feed

  • Question

  • Hi.

    Our Environment:

    Cluster - 5 x Dell R720 (192 GB RAM, 2 x E5-2690 Proc, 16 x Intel 10 GbE SFP+ Ports Per Server - 8 of these dedicated to iSCSI, 4 to Hyper-V Networks, 4 to Management, Live Migration & Redirected IO)

    Cluster Storage - Dell Equallogic PS6510X Hybrid SSD/SAS Storage Array

    VM Guests:

    2 x Oracle Enterprise Linux 7 VM's - 72 GB RAM (dedicated, or non-dynamic) 16 x VCPU's, 7 Fixed VHDX's, each contained in a dedicated CSV from a dedicated LUN from SAN.

    A couple of Windows Servers - varied shapes & sizes, each VHDX, is contained in its own dedicated LUN (CSV) on the SAN.


    Happens via Veeam, with the Equallogic Hardware VSS writer.


    Since the inception of the Oracle Linux VM's into the cluster, we've been having process locks (zombie processes or DS processes - not sure what the right Linux terminology is) on the Oracle Linux VM's due to IO load. The problem is that the reports on the SAN shows less than 4000 IOPS occurring during these times (the SAN has peaked on previous occasions over 20 000 IOPS without issues and is specked to be able to handle large VDI loads with up to 500 000 IOPS)

    When I transfer data from one Linux mount to another on the same VM (one VHDX to Another VHDX on the same SAN but different LUN / CSV), I get horrible transfer speeds of no more than 20 MBPS. When I do a similar transfer on a Windows VM (from one drive to another) I get transfer speeds of 800 MBPS. These are running at the same time, from the same Hyper-V Hosts, to the same SAN.

    For some reason it seems that even though there is ample speed & capacity on the SAN, the Linux VM is unable to access it.

    The iSCSI connectivity from the cluster to the SAN, has been optimised with Jumbo Frames, and Dell Deployment Services did the deployment, but I seriously think its not a SAN issue, as all other VM's in the environment is running beyond speed expectations.

    Has anyone experienced these issues with Oracle Enterprise Linux on Hyper-V, is there something I'm missing? We have a very clever in-house Linux / Oracle specialist, which have seen the speeds we're getting on the Windows OS'es, and have exhausted all options within the Linux OS, and he comes back every time saying that even though the speed is there on the hardware, it seems like the Hyper-Visor is not allowing full open access to it.

    Any Ideas guys, or even pointing me in the right troubleshooting direction will give you a Marked as answer on this post.


    • Edited by krokebob Wednesday, January 28, 2015 10:42 AM
    Wednesday, January 28, 2015 10:42 AM

All replies