The XenServer PVS-Accelerator feature offers additional capabilities for customers using XenServer and Citrix Provisioning Services (PVS). PVS is a popular choice for image management and hosting for XenApp and XenDesktop. With this feature, PVS read requests can now be cached on each XenServer host. In order to benefit from the PVS-Accelerator feature, customers should use XenServer with Provisioning Services 7.12 or higher. For detailed information about PVS-Accelerator, see the XenServer Administrator’s Guide on the Citrix Product Documentation website.
Enabling PVS-Accelerator involves a simple three-step process:
- Install the PVS-Accelerator Supplemental Pack on XenServer.
- Configure PVS-Accelerator in XenServer.
- Complete the cache configuration in PVS.
Customers should complete the following configuration settings in XenServer and in PVS to enable the PVS-Accelerator feature:
- Install the PVS-Accelerator Supplemental Pack on each XenServer host in the pool. The supplemental pack is available to download from the XenServer Product Downloads page. For instructions on how to install the supplemental pack, see Installing Supplemental Packs.
Configure PVS-Accelerator in XenServer. This configuration can be done using XenCenter or the xe CLI.
After installing the PVS-Accelerator Supplemental Pack, customers should add the PVS-Accelerator configuration details in XenServer. This entails adding a new PVS site and specifying the PVS Cache storage.
The following section contains XenCenter instructions. For information about configuring PVS-Accelerator using the xe CLI, see the XenServer Administrator’s Guide on the Citrix Product Documentation website.
The PVS tab appears at the pool-level (or host-level if there is no pool) in XenCenter after installing the PVS-Accelerator Supplemental Pack , and assigning a license with entitlement. The PVS tab displays a summary of the Read caching status for all the VMs running inside the pool.
To configure PVS-Accelerator
- Click on the pool or the standalone host and then select the PVS tab.
- Click Configure PVS-Accelerator.
- On the PVS-Accelerator configuration dialog, click Add cache configuration to add a PVS site.
- Enter a name for the PVS site in the Site name field.
For each host in the pool, specify what cache should be used:
- When you select Memory only , the feature will use up to the specified cache size in the Control Domain memory. This option is only available after additional memory has been assigned to the Control Domain. For information on how to assign memory to the Control Domain, see Changing the Control Domain Memory.
- When you select a Storage Repository (SR), the feature will use up to the specified cache size on the SR. It will also implicitly use available control domain memory as a best effort cache tier.
- If neither Memory only nor an SR is specified, the read cache will not be activated.
- PVS-Accelerator has been designed to utilize either memory only, or a combination of disk and memory. Irrespective of the configuration choice, customers should increase the amount of memory allocated to the Control Domain to ensure there is no system performance degradation.
- We recommend that you allocate at least 4GB of Control Domain memory per host to avoid frequent disk accesses that cause higher read-latency and consequently degrade performance. For more information, see Changing the Control Domain Memory.
- We recommend that you allocate at least 5GB of cache space per vDisk version that will be actively used.
- Click OK. The new PVS site and the chosen cache storage configuration will be added in XenServer.
- After configuring PVS-Accelerator in XenServer, customers must complete the cache configuration for the newly created site using the Provisioning Services Console or the PowerShell SnapIn CLI. For more information, refer to the Provision Services Documentation. When this step is complete, you can view a list of PVS Servers configured for the newly created site by clicking View PVS Servers on the PVS-Accelerator configuration dialog in XenCenter.
After starting a VM with PVS-Accelerator, the caching status for the VM will be displayed on the PVS tab and on the General tab of the VM. The following table lists the status messages displayed on these tabs.
|Initialized||PVS-Accelerator has been started and is ready to cache. If the cache remains in this state when the VM has been booted, it means that the PVS Server IP addresses have not been configured correctly, or the VM is not communicating with the PVS server using its primary network interface.|
|Caching||PVS-Accelerator is working.|
|Stopped||PVS-Accelerator is not running for the VM. The cache remains in this state when the VM is not running, or when the cache is not configured sufficiently.|
|Incompatible Write Cache Mode||There is no caching as the VM is configured to persist changes on the PVS server. Ensure the VM type is “Production” or “Test” and the vDisk is in “Standard Image” Access mode.|
|Incompatible Protocol Version||The PVS Server version is incorrect. Ensure that you are using Provisioning Services 7.12 or higher.|
The PVS-Accelerator functionality caches:
- Reads from vDisks but not writes or reads from a write cache
- Based on image versions. Multiple VMs will share cached blocks if they use the same image version
- Devices with any write cache type
- vDisks with the Access mode set to Standard Image. Caching does not work with vDisks set to Private Image mode
- Devices that are marked as type Production or Test. Devices marked as type Maintenance are not cached
- PVS-Accelerator is available for XenServer Enterprise Edition customers, or those who have access to XenServer through their XenApp/XenDesktop entitlement.
- XenCenter displays various PVS-Accelerator performance graphs on the host-level Performance tab. The performance graphs provide detailed insight into the cache operation.
- The PVS-Accelerator feature leverages capabilities of OVS and is therefore not available on hosts that use Linux Bridge as the network backend.
- PVS-Accelerator works on the first virtual network interface (VIF) of a cached VM. Therefore, the first VIF should be used for connecting the PVS storage network for the caching to work.