6.0.2

Virtual SAN Datastores That Aggregate Local Storage Disks from a vSphere
Cluster
Virtual SAN virtualizes the local physical storage disks available on ESXi hosts into a single datastore
shared by all hosts in a vSphere cluster. A Virtual SAN datastore consists of solid-state drive (SSD) disks
and hard disk drives (HDDs), also called data disks. SSD disks are used for read caching and write
buffering. Data disks are used for persistent storage. This strategy provides high-performance storage with
automatic caching, so that you specify only one datastore when creating a desktop pool, and the various
components, such as virtual machine files, replicas, user data, and operating system files, are placed on the
appropriate SSD or data disks.
NOTE The Virtual SAN feature requires vSphere 5.5 Update 1 or a later release, and requires the
appropriate hardware. See the VMware Compatibility Guide.
When you use Virtual SAN, View defines virtual machine storage requirements, such as capacity,
performance, and availability, in the form of default storage policy profiles, which you can modify. Virtual
SAN lays out the virtual disk across the logical datastore to meet the specified requirements. Virtual SAN
also monitors and reports on the policy compliance during the life cycle of the virtual machine. If the policy
becomes noncompliant because of a host, disk, or network failure, or workload changes, Virtual SAN
reconfigures the data of the affected virtual machines and optimizes the use of resources across the cluster.
NOTE When you create a linked-clone desktop pool, a full clone is first made from the parent virtual
machine. From this full clone, or replica, linked clones are created. If you use a Virtual SAN datastore, by
default an extra copy of the replica and linked clones are created according to the availability policy.
Storage Sizing for Linked-Clone Desktop Pools
View provides high-level guidelines that can help you determine how much storage a linked-clone desktop
pool requires. A table in the Add Desktop Pool wizard shows a general estimate of the linked-clone disks'
storage requirements when the pool is created and as the linked clones grow over time.
The storage-sizing table also displays the free space on the datastores that you select for storing OS disks,
View Composer persistent disks, and replicas. You can decide which datastores to use by comparing the
actual free space with the estimated requirements for the linked-clone disks.
The formulas that View uses can only provide a general estimate of storage use. Your linked clones' actual
storage growth depends on many factors:
n
Amount of memory assigned to the parent virtual machine
n
Frequency of refresh operations
n
Size of the guest operating system's paging file
n
Whether you redirect paging and temp files to a separate disk
n
Whether you configure separate View Composer persistent disks
n
Workload on the linked-clone machines, determined primarily by the types of applications that users
run in the guest operating system
NOTE In a deployment that includes hundreds or thousands of linked clones, configure your linked-clone
pools so that particular sets of datastores are dedicated to particular ESXi clusters. Do not configure pools
randomly across all the datastores so that most or all ESXi hosts must access most or all LUNs.
When too many ESXi hosts attempt to write to linked-clone OS disks on a particular LUN, contention
problems can occur, degrading performance and interfering with scalability. For more information about
datastore planning in large deployments, see the View Architecture Planning document.
Setting Up Desktop and Application Pools in View
186 VMware, Inc.