Apache CloudStack backups use access nodes to move data from hypervisors to Commvault storage. Full backups use temporary staging space on the access node.
Use this information to size and configure access nodes for reliable backup and restore operations.
Key behavior that affects requirements
-
Full backups copy virtual machine data to a staging directory on the access node before the backup runs
-
Incremental backups don’t use staging
-
Backups are performed by the host that is running the virtual machine
These behaviors affect how you size storage, compute, and concurrency on access nodes.
Supported environments
-
Commvault supports Apache CloudStack version 4.20.x and later.
-
Backups and restores are supported only for Linux hosts that use the Kernel-based Virtual Machine (KVM) hypervisor.
-
Unsupported hypervisors and operating systems aren’t validated for backup and restore operations.
Access node requirements
Deployment
Access nodes must be able to communicate with:
-
CloudStack management components
-
Hypervisors hosting virtual machines
-
Commvault storage targets
Compute
Size CPU and memory to support concurrent backups, including the additional load from staging during full backups.
-
Full backups require additional resources for staging and data transfer
-
Don’t assign more concurrent large virtual machine backups than the access node can handle based on available CPU, memory, and staging capacity
Storage requirements
Staging directory for full backups
During a full backup, virtual machine data is copied to a staging directory on the access node and then backed up from that location.
Capacity requirements
On each access node, allocate staging space at least equal to the largest virtual machine expected to be backed up at one time on that host.
-
Size staging per host, not across all access nodes
-
Increase capacity if multiple large virtual machines are backed up concurrently
Performance considerations
-
Use high-throughput storage for the staging directory
-
Local storage typically provides better performance and reliability than network-mounted paths
-
Network-mounted paths are supported but might impact backup performance depending on latency and throughput
-
Ensure sufficient disk I/O for large full backups
Network requirements
Provide reliable, high-throughput connectivity between:
-
Hypervisors and access nodes
-
Access nodes and Commvault storage
Full backups generate additional data movement because data is written to staging before being transferred to storage.
Configuration
Configure the staging directory by using the sQemuSnapshotStagingDir additional setting.
You can apply this setting in the following ways:
-
Hypervisor level: Hypervisor > Configuration > Settings
-
Server group: create a server group with all CloudStack access nodes and configure the setting
-
Command Center: apply the setting to access nodes or server groups from the Additional settings page
Use a consistent staging configuration across access nodes to ensure predictable backup behavior.
Support boundaries and limitations
-
Full backups require staging to complete successfully
-
Backups are host-bound to the virtual machine’s running host
-
Insufficient staging space can cause full backup jobs to fail
Best practices
-
Size staging for the largest expected virtual machine per host
-
Monitor staging disk usage regularly
-
Use dedicated storage for staging instead of OS disks
-
Apply consistent staging configuration across access nodes