Cluster Shared Volume Backup and Restore Considerations

Cluster Shared Volume backups are fully supported in BackupChain Server Edition and Server Enterprise Edition.

It is recommended to limit read and write I/O speeds or to use separate network adapters for backup traffic, in order to keep the server network balanced.

You may need to use the domain administrator account for your iSCSI provider service in order to get the backup task to start successfully, depending on the iSCSI service you are using.

For live backups with VSS engagement (the recommended way to back up), you need to run BackupChain on the cluster node that actually hosts the VM you want to back up. You could back up virtual machines from another node as well using the file-based approach (discussed in previous section) but that approach “from a distance” does not involve VSS and hence will only give you a crash consistent live backup. A crash consistent backup is similar to a power loss event in a physical machine. Most production grade applications can handle a power loss event without data loss, such as a SQL Server database server. To achieve an application consistent live backup which will also notify the VM internally of the backup event, you need to run the backup on the same node that hosts the virtual machine you wish to back up. Application consistent backups require the VSS integration and this integration can only be managed from the VM’s host. VSS aware applications, such as database services and Exchange, prepare for live backup, flush their caches, and bring their file stores into a consistent state before the backup begins.

Do not perform a live migration while a backup is running because live backups cannot be processed when a VM switches hosts. Also, you need to configure BackupChain on the new cluster node when you reassign virtual machines to new hosts (select VM in Hyper-V tab table or use the Automatic VM Selection feature).

Note: On older Windows Server (pre 2012) operating systems you need to schedule CSV backups to run without overlaps because CSVs do not allow nodes to back up simultaneously.

General CSV Characteristics

You cannot back up VMs running on another node, even if they share the same storage on a SAN.

Application consistent backups require backup software to run locally on the node that hosts the VM.

Windows Server 2008 and R2 Specifics

·You may only run one backup across all nodes of an entire cluster shared volume.

You can back up several VMs simultaneously; however, these must be running on the same node. In other words, do not overlap backups running on several nodes simultaneously. You need to schedule them to run without overlap.

Windows Server 2012 and Later Specifics

Do not add local or other CSV paths to a backup task.

·You must select a Hyper-V Backup task type in the task wizard. You cannot use Universal Backup task types to back up CSV VMs.

Dealing With VM Migrations (Live, Manual, Automatic / CSV)

Beginning with Version 4, BackupChain offers an Automatic VM Selection feature. By standardizing the names of your VMs, you can define filters to include or exclude VMs automatically from backups.

For example, if you only want production VMs backed up, you could use a suffix in the VM name, like this: Exchange_Server_Production. The ending “_Production” can be used as an inclusion filter in the backup task’s Hyper-V tab so that BackupChain picks up that VM automatically. Likewise, you could define an exclusion filter to avoid backing up VMs that are for testing only, by calling the VM Exchange_Test and defining an exclusion filter having the text: _Test

By using the Automatic VM Selection feature, you no longer have to update backup configurations when VMs move. Note there is a warning option at the bottom of the Hyper-V tab that will cause BC to log a warning if no VMs were found. This is a safety feature to avoid the situation where filters may be set incorrectly by accident and no VM ends up being selected. Because there are legitimate scenarios when this can occur, such as when all VMs are moved off the server for maintenance, you can decide to turn off that feature, after you have confirmed that VMs are selected for backup as desired.