TI CITRIX

Clusters and High Availability

In vSphere, a cluster is a construct that consists of multiple VMware ESXi hosts bound together

to form a single entity for the purpose of management and high-availability. Hosts, within a

cluster, usually share common resources so that virtual machines can be started on, or migrated

between (vMotion) any host within the cluster where sufficient resources exist. Resource pools,

which can be created within a cluster, allow administrators to subdivide a cluster configuration to

create shared logical pools of CPU and memory resources and in some cases guarantee resource

levels for certain groups. Based on the vSphere 5 Configuration Maximums, VMware supports

up to 32 hosts and/or 3000 virtual machines in a single cluster and up to 1600 resource pools can

be created per cluster.

The number of clusters required to support an implementation of XenDesktop depends on

various factors, including:



Number of Hosts and Virtual Desktops – Citrix Consulting has observed optimal virtual

desktop density in vSphere 4 configurations with a maximum of 8 hosts per cluster or

approximately 1000 virtual machines per cluster. Increased scalability in vSphere 5 allows

for up to 10,000 running virtual machines per vCenter configuration, and a maximum cluster

configuration of 32 ESXi hosts or 3000 virtual machines per cluster. Multiple cluster

configurations are permitted per vCenter configuration, but VMware does not publish a

maximum number. Although the maximum cluster configurations are higher for vSphere 5,

there are still additional items that need to be considered when evaluating the different

FlexCast models. For Machine Creation Services (MCS), it is recommended that cluster size

should not exceed 8 hosts given that the MCS virtual desktops reference a master image and

all the reads from that single image are processed by the hypervisor. For Provisioning

Services (PVS), it may be possible to increase the number of virtual desktops per cluster as

the master vDisk image is accessed for reads either from local cache or through network

access. In general, Citrix Consulting recommends that customers perform scalability testing

of individual configurations to validate performance, particularly beyond the best practices

configuration of 8 hosts per cluster and/or 1000 virtual desktops per cluster configuration.

Page 3 Performance – Most businesses will have desktops groups that require guaranteed

performance levels. To address this requirement in a small environment, consider the

implementation of dedicated ESXi hosts within an existing cluster, and the use of vSphere

resource pools to subdivide the cluster. For larger environments, it is sometimes necessary

to create dedicated clusters to meet the service level agreements associated with these

desktops.

 High Availability – There may be a requirement for desktop groups to offer varying levels of

redundancy. For example, a desktop group used by financial traders could require N+100%

redundancy while a desktop group accessed by human resources may only require N+10%.

In order to accommodate increased levels of redundancy, sufficient capacity must exist

within the cluster to handle the required number of host failures. In such situations, it may

be pertinent to isolate desktop groups into their own clusters based on the level of

redundancy required.

 Infrastructure Nodes – When designing a clustered solution, consider separating the

XenDesktop infrastructure nodes (host server components such as AD, SQL and

XenDesktop controllers) from the virtual desktop nodes by placing them on different

vSphere clusters. This will allow vSphere features such as Distributed Resource Scheduler