While restoring data after a disaster or during disaster recovery training, you may need to start virtualized instances of servers that were previously protected by Arcserve UDP.
Arcserve UDP provides the following two features that allow you to start the virtual machine from recovery points:
For more details on Instant Virtual Machine, see How to Create and Manage an Instant Virtual Machine.
For more details on Virtual Standby, see How to Create a Virtual Standby Plan.
To determine which feature works best, you need to consider your RTO (Recovery Time Objective) and scenario. The following table provides the comparison between the IVM and VSB features:
FEATURES |
IVM |
VSB |
Power standby VMs from the latest recovery point |
Yes (no conversion needed) |
Yes, ONLY if a VSB task was added to the backup plan. For example, Advance planning required) |
Requires backup time processing |
Not required |
Required, it is necessary to add a VSB task to the plan that is used to back up the source machine. |
VM Boot time |
A slower process (up to 30%) due to I/O redirection. |
Same time as any other VM on the same hypervisor. |
Disk space requirements |
Minimal storage space to host child disk or store changes when running a VM. |
Yes, storage space is consumed on the destination hypervisor where VSB standby VM is maintained. Requires storage space equal to or greater than the size of the source machine. |
High Availability (HA) option |
N/A |
AVAILABLE Monitors source machine and can start VSB VM if the source machine becomes unavailable. |
VM performance |
May run slower compared to regular VMs (up to 30%) due to I/O redirection, however the performance depends on the nature of the application workload. |
Performance is the same as the regular VMs. |
Management/Configuration |
Managed from the UDP console, can start or stop the IVM on demand when access is needed by the user. |
Added as a task to a plan so that all the backed up data is converted to a VM format automatically. VSB task is applied to all nodes that are protected by the plan. |
Persisting data and migrating VM to production |
The virtual disk of the IVM refers to the data blocks in the recovery point from which the VM was started. Therefore, when the IVM accesses data blocks within its virtual disk the data is actually requested from RPS (this process is transparent to the user). Such I/O redirection introduces some additional performance footprint. If you plan to use IVM in production then we recommend to make IVM persistent and hydrate the virtual machine virtual disk with the actual data. Hydration of the IVM can be achieved by copying/replicating the VM. Depending on the type of Hypervisor used in your production environment, to persist the IVM data you can use VMware Storage vMotion or Hyper-V VM replication to copy the IVM where the data becomes permanent. |
The virtual disk or disks of the VSB VM already contain most of the recent data from the corresponding recovery point. Since I/O redirection does not occur (which is the same as the IVM), the performance of the VSB VM is the same as regular VMs where there is no dependency on the RPS or recovery point (compared to the IVM scenario). |
Copyright © 2016 |
|