Enhancements for Full VM Restores
For Service Pack 10 and later, restores of full virtual machines or instances use a new method that improves performance and scalability for restores. This restore method provides more efficient restores from disk storage, with better use of physical resources, but still supports recovery from traditional media such as tape. The speed of multi-VM restores is significantly enhanced, with the speed for single-VM restores equivalent to that of the old method.
The new restore method is used by default for Windows and Linux VMs for all hypervisors supported by the Virtual Server Agent.
The new restore method include the following enhancements:
- Allocates multiple streams for multi-VM restores:
- Restores VMs in parallel.
- Allocates one stream for each VM.
- Uses up to ten streams for each restore operation from streaming backups.
Restores from IntelliSnap backups use a single stream.
- Supports restores to a different hypervisor (VM conversion).
- Uses a centralized indexing service that is not dependent on the restore job.
- Skips processing that is not required for the specific type of restore (for example, omitting indexing queries for live VM recovery).
- Enables restores of individual VMs to complete while restores of other VMs in the same job are still in progress.
- VMware only: If a restore job for a VM is interrupted, it continues from the last disk block for the VM that was successfully written.
The new restore method is not used for the following operations:
- Live sync replication
- Restores of VM clients running Service Pack 9 or earlier
- Restores of guest files and folders
The new restore method creates a new restore log named vsrst.log.
The older vsvrst.log still provides information for restores using the old method, and contains some messages about the cleanup phase of restores.
Browse.log and IndexingService.log provide messages related to indexing.
fsIndexedRestore.log is not used.
Last modified: 3/15/2018 12:51:19 PM