Table of ContentsView in Frames

Aligning the I/O of misaligned virtual machines non-disruptively

A misaligned virtual machine can negatively affect I/O performance. To correct the misalignment, Virtual Storage Console for VMware vSphere can migrate a virtual machine to a new or existing optimized datastore so that the I/O is functionally aligned. This process does not require downtime.

Before you begin

About this task

Steps

  1. In the Optimization and Migration page, select a datastore on which a misaligned virtual machine is running.
    Screen shot: The datastores table showing a selected datastore.
    After you select a datastore, the virtual machines that reside on that datastore appear in the virtual machines table (the table at the bottom of the Optimization and Migration page).
  2. In the virtual machines table, select one or more virtual machines with the status "Misaligned, Online migration capable."
    If you select multiple virtual machines, they must have the same offset group (the offset of the largest disk partition).
    You should limit the number of virtual machines that you migrate at one time to avoid over-stressing your system.
    Screen shot: The virtual machines table showing a virtual machine with the status Misaligned, Online migration capable.
  3. Click Migrate.
  4. Complete the pages in the Migrate Virtual Machines wizard to migrate the virtual machine to an optimized datastore.
    1. In the Destination datastore page, specify whether you want to use an existing datastore or a new datastore.
      For existing datastores, you will be able to choose from datastores that are optimized for the offset group of the virtual machine.
    2. (New datastore) In the Name and type page, specify a datastore name, datastore type (NFS or VMFS), and for a VMFS datastore, the VMFS protocol (FC/FCoE or iSCSI).
    3. In the Storage system page, specify the storage system that you want to use for the datastore.
    4. (Existing datastore) In the Datastore selection page, select the destination datastore.
      VSC lists the datastores that are optimized for the VMDK layout of the virtual machine. If no datastores are listed, go back and select the new datastore option.
    5. (New datastore) In the New datastore details page, specify details about the datastore that you want to create.

      Most of the fields on this page are self-explanatory. The following table describes fields for which you might need guidance:

      Field Description
      Thin provision Allocates space on the volume when data is written, which allows you to provision more storage than is currently available. If disabled, space is reserved immediately.

      You must closely monitor the available space in the containing aggregate because thin provisioning can oversubscribe the available space. In an NFS configuration, you can enable auto grow to automatically expand the datastore when space is needed. Make sure that the value you specify for auto grow is larger than the size of the datastore.

      Aggregate Defines the aggregate on which you want to create a new volume. If you selected an SVM that is directly connected to VSC, striped aggregates appear as available; however, they are not supported. Provisioning to a striped aggregate will fail.
      Volume Specifies the volume on which you want to create the datastore. For clustered Data ONTAP, you should not create a datastore in the Storage Virtual Machine (SVM) root volume.
      Auto grow (NFS only) Automatically expands the datastore by the specified increment when space is needed, up to the size limit. This size limit you specify must be larger than the existing datastore.
      Datastore cluster Adds the datastore to a cluster if the Storage Distributed Resource Scheduler (SDRS) feature is enabled on the vCenter Server. Do not mix datastores with varying offsets in the same cluster and do not mix optimized and non-optimized datastores.
    6. In the Ready to complete page, Review the summary of your selections and click Finish.

Result

VSC starts the migration task. You cannot cancel this task.

After you finish

If the old datastore is empty, use it for other virtual machines or destroy it.