Operations that require getting a file lock or a metadata lock in VMFS result in short-lived SCSI reservations. SCSI reservations lock an entire LUN. Excessive SCSI reservations by a host can cause performance degradation on other servers accessing the same VMFS.

Problem

Excessive SCSI reservations cause performance degradation and SCSI reservation conflicts.

Several operations require VMFS to use SCSI reservations.

  • Creating, resignaturing, or expanding a VMFS datastore

  • Powering on a virtual machine

  • Creating or deleting a file

  • Creating a template

  • Deploying a virtual machine from a template

  • Creating a new virtual machine

  • Migrating a virtual machine with VMotion

  • Growing a file, such as a thin provisioned virtual disk

Note:

ESXi hosts use the SCSI reservations mechanism only when storage devices do not support the hardware acceleration. For storage devices that support the hardware acceleration, the hosts use the atomic test and set (ATS) algorithm to lock the LUN. For more information on hardware acceleration, see the vSphere Storage documentation.

Results

To eliminate potential sources of SCSI reservation conflicts, follow these guidelines:

  • Serialize the operations of the shared LUNs, if possible, limit the number of operations on different hosts that require SCSI reservation at the same time.

  • Increase the number of LUNs and limit the number of hosts accessing the same LUN.

  • Reduce the number snapshots. Snapshots cause numerous SCSI reservations.

  • Reduce the number of virtual machines per LUN. Follow recommendations in Configuration Maximums.

  • Make sure that you have the latest HBA firmware across all hosts.

  • Make sure that the host has the latest BIOS.

  • Ensure a correct Host Mode setting on the SAN array.

For information about handling SCSI reservation conflicts on specific storage arrays, see the VMware knowledge base article at http://kb.vmware.com/kb/1005009.