Product Documentation

Power manage Desktop OS machines

Aug 29, 2013

Full and partial machine power management is available for Delivery Group machines.

Note the following:

  • This procedure only applies to Desktop OS machines. For Server OS , manage power settings by using reboot schedules. See Set up the restart schedule for a Server OS machine Delivery Group for detailed information. You cannot power manage remote PC access Delivery Group machines.
  • You can only power manage virtual machines, not physical ones. The ability to fully or partially control power management depends on how the Delivery Group's virtual machines are allocated to users or user devices. Permanently allocated machines can only be partially power managed.

    Machines using GPU capabilities cannot be suspended. Therefore, attempting to power off a Delivery Group with GPU machines results in an error.

Machines can be in one of these states:

  • In random Delivery Groups: randomly allocated and in use

  • In static or random Delivery Groups: unallocated (and therefore unconnected)
  • In static Delivery Groups:
    • Permanently allocated and unconnected (but ready to be connected)
    • Permanently allocated and in use

At any time, static Delivery Groups typically contain both permanently allocated and unallocated machines. Initially, all the machines are unallocated (apart from any manually allocated to individuals when the Delivery Group was created). As users connect, some get permanently allocated. Therefore, when you fully power manage groups of this type, you are only fully managing the unallocated machines in it. The permanently allocated machines are partially managed.

Pools and buffers

For random Delivery Groups and unallocated machines in static Delivery Groups, a pool is a set of unallocated (or temporarily allocated) machines in the Delivery Group that are kept in a powered-on state, ready for users to connect. When a user logs on, they are immediately presented with a machine. The pool size (the number of machines kept powered on) is configurable. For example, you set up a bigger pool during office hours. For static Delivery Groups, there is no pool in Studio but you can use the XenDesktop SDK to configure one.

A buffer is an extra, standby set of unallocated machines that are turned on, ready for users to connect. For random Delivery Groups and unallocated machines in static Delivery Groups, machines in the buffer are turned on when the number of machines in the pool drops below the threshold set by the buffer size. This is a percentage of the Delivery Group size (by default, 10%). For large Delivery Groups, a significant number of machines may be turned on when the threshold is exceeded, so plan your Delivery Group sizes accordingly or adjust the default buffer size using the SDK.

Power state timers

You can suspend machines after users have disconnected for a defined time using power state timers. For example, machines can suspend automatically outside office hours if users have been disconnected for at least 10 minutes. Unless you have configured the ShutdownDesktopsAfterUse property of a Delivery Group using the SDK, random machines or machines with Personal vDisks are always automatically shut down when users log off.

You can configure the timers separately for weekdays (by default, Monday to Friday) and weekends, and for peak and off-peak periods. The peak period covers the time at which most users log on to their desktops, and starts at the beginning of a business day. Use the SDK if you want to shut down, rather than suspend, machines in response to power state timers, or if you want the timers to be based on logoffs, rather than disconnections. Also, note that the Weekdays and Weekend selections in this procedure are defaults that can be configured using the SDK. For information about using the XenDesktop SDK, see About the XenApp and XenDesktop SDK.

Partial power management of permanently allocated machines

With machines permanently allocated to individuals or user devices, you can set power state timers but not pools or buffers. The machines are turned on at the start of each peak period, and turned off at the start of each off-peak period, so you have no fine control (as you do with unallocated machines) over the number of machines that become available to compensate for machines that are consumed.

To power manage Delivery Groups

  1. In Studio, select the Delivery Group node, and then select the Delivery Group whose power management settings you want to control.
  2. Click Edit Delivery Group.
  3. Click Power management and then select Weekdays in Power on/off machines.
  4. For random Delivery Groups, in Machines to be powered on, click Edit and specify the pool size during weekdays.
  5. In Machines to be powered on, select the number of machines to power on.
  6. In Peak hours, set your organization's peak and off-peak hours during weekdays.
  7. Set power state timers for peak and non-peak hours during weekdays:
    • In During peak hours When disconnected, specify the delay (in minutes) before suspending any disconnected machine in the Delivery Group, and select Suspend.
    • In During off-peak hours When disconnected, specify the delay before turning off any logged-off machine in the Delivery Group, and select Shutdown. This timer is not available for groups based on random machines.
  8. Select Weekend.
  9. Configure, as previously described, the peak hours, and power state timers for weekends.