Recovering Host-level Backups

Unitrends provides a variety of methods for recovering host-level backups of virtual machines. You can recover entire virtual machines or selected files from backup. For quick recovery of critical VMs, you can create virtual machine replicas or use the instant recovery feature.

See the following table for descriptions of each recovery method:

Recovery method	Description
Virtual machine recovery	Recovers the entire virtual machine from its host-level backup to a target virtual host, which can be the original host or an alternate host. Supported for host-level backups of all VM guest operating systems. For detailed requirements and recovery procedures, see Recovering a virtual machine.
File recovery	Recovers selected files from a backup of a Windows or Linux VM. For detailed requirements and recovery procedures, see Recovering files from virtual machine backups.
Virtual machine replicas	Creates a stand-by replica of a VMware VM that you can bring online in minutes. As backups of the original VM run, they are applied to the replica to keep it up-to-date. Take the replica 'live' to assume the role of a failed VM. To meet near-zero RTOs, set up the replica before the VM fails. For detailed requirements and procedures, see VM replicas. For a comparison of the virtual machine replica and instant recovery features, see Replicas or instant recovery?
Virtual machine instant recovery	Recovers a failed VMware or Hyper-V VM in minutes. The recovered VM can immediately assume the role of the original, failed machine. Unitrends recommends that you plan for instant recovery (IR) before a VM fails, by reviewing requirements, allocating IR storage, and using audit mode to test IR of your critical VMs. For detailed requirements and procedures, see Virtual machine instant recovery. Because instant recovery uses appliance resources that can impact the performance of other jobs, Unitrends recommends using virtual machine replicas for VMware VMs and standard VM recovery for non-critical VMs. For a comparison of the virtual machine replica and instant recovery features, see Replicas or instant recovery?

Replicas or instant recovery?

Replicas and instant recovery provide ways to recover your critical VMs in minutes. Instant recovery is supported for Hyper-V and VMware. Replicas are supported for VMware only.

For VMware, you can use either feature, or a combination of both, to reduce production downtime. The following table provides a high-level comparison of the virtual machine replicas and instant recovery features. Consider these differences when deciding which feature to use for your critical VMware machines.

Item	VMware replicas	VMware instant recovery (IR)
Scale	Because creating and updating replicas consumes few appliance resources, this feature can be used for 1000s of VMs.	Appliance resources are used to create each IR session. The number of VMs that can be recovered simultaneously is limited by appliance load and available IR space.
Automated and orchestrated recovery	Integrates seamlessly with Copy Data Management to perform automated and orchestrated recovery of a single VM or a subset of the entire virtual infrastructure. For details, see Recovery Assurance.	Not supported. VM IR must be run manually for each VM.
Recovery Time Objective (RTO)	Fastest, near-zero recovery time. Just click Go Live to boot into live mode, then configure network settings to bring the replica online in production to assume the role of the failed VM. Note: After creating a replica, you can set up a Copy Data Management job to fully automate testing and failover. This way, you can failover by using pre-configured network settings. An Enterprise Plus license is required to use this feature. For details, see Recovery Assurance.	Slower recovery time than replica VMs. You select a backup and configure the IR job. The appliance then creates a local disk image and a VM on the virtual host. This can take some time, especially for large incremental backup chains. The appliance powers on the VM and begins migrating data from the local disk image. At this point the VM is fully operational and can assume the role of the failed VM.
Recovery Point Objective (RPO)	Recover from any recovery point stored with the replica VM on the hypervisor. By default, only the latest recovery point is retained. You can modify this setting to retain up to 31 recovery points.	Recover from any host-level backup or imported backup copy on the Unitrends appliance.
On-appliance retention	Does not impact on-appliance retention; all backup storage is used for backups.	Decreases on-appliance retention; a portion of backup storage is allocated to IR.
Appliance performance	Little impact to appliance performance. Replicas are created and updated by running regular recovery operations. A replica's compute and storage are allocated by the ESXi host.	VM IR has a greater impact to appliance performance than VM replicas. Each IR session uses appliance storage and compute resources to create and run a disk image from the selected recovery point. This disk image must remain on the appliance until all data has been copied to the VM on the hypervisor. Once data migration is complete, you can tear down the IR session to free up appliance resources.
ESXi host performance	Impact to the ESXi host for a given VM is about the same for a live replica as for a live VM IR. A replica is created on the specified ESXi host as a cold stand-by and remains powered off, even as backups are applied. A portion of the ESXi host's storage and compute resources are allocated to each replica VM, but no compute resources are used until you boot the replica into live or audit mode. Because an appliance can manage 1000s of replicas, it is important that you monitor ESXi host resources carefully and make adjustments as needed.	Impact to the ESXi host for a given VM is about the same for an IR VM as for a live replica. Note that Storage vMotion runs on the hypervisor to copy data from the appliance to the IR VM, which may temporarily impact host performance.
vCenter and Storage vMotion	A replica can be hosted on a stand-alone ESXi server or on a server that is managed by a vCenter. Replicas do not use Storage vMotion and do not require a vCenter. Note: To run Copy Data Management jobs for the replica, the ESXi host must be managed by a vCenter.	The IR target must be an ESXi server that is managed by a vCenter and must have a license that supports Storage vMotion.
Setup for multiple VMs	You can select multiple VMs in the Create Replica VMs dialog, to quickly set up replicas for a group of virtual machines. You can even include VMs that have not yet been backed up. In this case, when a successful backup runs, it is automatically applied to complete the replica creation process for that VM.	Not supported. VM IR must be run manually for a single VM.