- ShadowProtect SPX
- StorageCraft Cloud Services
- StorageCraft Granular Recovery for Exchange
- StorageCraft Recovery Environment
- Cloud Backup
- File Backup and Recovery
- Company Info
- How to Buy
Host-Based Agentless Backup
One way to deliver “agentless” backup is to leverage the hypervisor (VM host). Since the hypervisor has full access to VM storage files and systems, it can theoretically capture a VM’s state at a given moment-in-time. All popular hypervisors provide tools to help accomplish this. Unfortunately, the quality and scope of these tools varies from vendor to vendor, so host-based backup is inconsistent from hypervisor to hypervisor.
Generally, host-based backup can protect the following types of files:
- Virtual hard disks
- Virtual network configuration files
- Host configuration information
- Virtual machine configuration and saved-state files
Host-based backup provides ease-of-use by consolidating backup configuration and management for all hosted VMs in one place, but it does so by sacrificing flexibility and in some cases backup quality.
Host-based agentless backup sacrifices flexibility in some implementations. For instance, a host-based backup job supports only one schedule and destination for all VMs on that host. Agent-based backup, on the other hand, supports unique backup schedules and destinations for each VM.
Sometimes a hypervisor doesn’t properly prepare each guest VM before performing a backup. This preparation includes such critical tasks as instructing applications to flush to disk all transactions and data currently held in memory. In this case the backup appears successful, but the resulting backup data improperly captures the application state. These inconsistencies in file data and file system metadata can cause failures after restoring a VM. If the hypervisor doesn’t prepare its VM for a backup, then the backup might be of little value after a disaster when you need it to restore your system.
Coordination must take place between the host, guest, backup solution, VSS, and the hypervisor in order to capture a consistent state for the VM disks. The hypervisor must properly flush all cached data to storage before creating the host volume snapshots. Also, the hypervisor temporarily blocks any new storage I/O after flushing the cache until the snapshot of the host volume is complete. The hypervisor’s VSS Writer should, theoretically, accomplish this, but in practice this process is not completely reliable.