Recomposing linked-clone desktops

A virtual desktop recompose operation is used when you need to update the master desktop replica image. One of the key methods of maintaining the storage efficiencies of linked-clone desktops is to control the amount of data that is written to the linked-clone disks, which if left unchecked will grow in size during the course of regular use. Traditional per-desktop administrative tasks such as installing applications and applying patches can quickly increase linked-clone storage utilization, as the writes would be made to the linked-clone disk of each desktop that was the target of the action. In addition, these operations would need to be repeated on each desktop, which can require significant management overhead.

A recompose operation replaces the linked-clone replica disk with an updated version that uses the same OS. Recomposing to a completely different OS is not supported. During the recompose process, all virtual desktops will be provisioned again and linked to the new replica disk. If the desktop was configured with a persistent disk for storing user profile data—an option for dedicated assignment desktop pools—that disk would be retained and attached to the recomposed desktop.

Technically speaking, a recompose operation requires a refresh operation. Either of these operations will temporarily reduce the amount of vSphere datastore capacity your persistent Horizon desktops are using.

The recompose process is described in detail in Chapter 9, Performing Horizon Pool Maintenance.

When a desktop is recomposed, it will maintain the same computer name but will obtain a new DHCP lease. Your DHCP server must have a large enough pool of addresses available to handle these requests. In addition, it is recommended you use a short address lease period so that the now unused leases are quickly removed and the associated addresses are made available for use.