Using Hyper-V Recovery Manager for Disaster Recovery of Private Clouds

Disaster recovery is top of mind for nearly every organization that I talk to. With the recent spate of natural disasters across the U.S. and the globe, organizations are looking for ways to protect themselves from the loss of a data center.

Although there are many aspects to a disaster recovery solution, a major one is ensuring that the services running in the data center are available and current. Many enterprise services have their own disaster-recovery features, such as Active Directory Domain Services with its multi-master replication, SQL Server with asynchronous and synchronous replication using AlwaysOn, and Exchange Server and its Database Availability Groups. But many other services do not have a native replication or disaster-recovery capability. And SAN-level multisite replication is cost-prohibitive for many organizations and workloads.

Windows Server 2012 introduced Hyper-V Replica, which provides an asynchronous replication of virtual machine (VM) virtual storage every 5 minutes. This capability enables any Hyper-V VM to be replicated to an alternate location, without the need for dedicated networking and without affecting the performance or scalability of the source VM. In addition, Hyper-V Replica enables an alternate IP configuration that can be automatically injected into a Windows or Linux VM if a failover occurs at the recovery site. This VM-level replication capability enables the protection of workloads that are not already protected by a native disaster-recovery capability. It also makes these workloads available as part of a disaster-recovery plan, which is essential for all organizations.

Windows Server 2012 R2 enhances Hype-V Replica with more granular control of the replication interval via additional 30-second and 15-minute replication intervals. Hyper-V Replica in Windows Server 2012 R2 also extends replication to a third site, if required.

Although the technology to asynchronously replicate VMs is great, it’s only one part of the solution. Organizations might have hundreds or thousands of VMs that must be started in a specific order; scripts might need to be executed to perform other actions and configurations. Although the entire failover sequence can be performed manually, that approach is not ideal. During a disaster, IT staff might not be available. A good disaster-recovery process is one that can be performed by almost anyone, which means a high level of automation that can be triggered without dependence on physical location. The Windows Azure hosted Hyper-V Recovery Manager service provides this automation through the Windows Azure management portal, which is accessible from any Internet-connected device.

Hyper-V Recovery Manager works through a System Center Virtual Machine Manager (VMM) extension. This extension enables Hyper-V Recovery Manager to communicate with VMM and thereby manage the Hyper-V environment. This approach allows Hyper-V Recovery Manager to perform the initial configuration of Hyper-V Replica replication between Hyper-V servers in different locations. It also allows Hyper-V Recovery Manager to initiate and orchestrate the complete failover process in planned, unplanned, and test failover scenarios. VMs can be placed into groups and started in a specific sequence, and custom scripts can be executed during the failover process via Hyper-V Recovery Manager. After the initial configuration of the failover process is complete, the actual initiation of a failover is a simple process that can be performed by any user who has the required permissions.

Although emulating the functionality of Hyper-V Recovery Manager is possible by using technologies such as System Center Orchestrator and Windows PowerShell, the amount of custom work in those scenarios is high. And organizations would still be faced with the challenge of hosting the failover engine in a neutral location that does not depend on a specific data center. For organizations that use Hyper-V Replica for site failover, Hyper-V Recovery Manager should be considered a must. You can learn more about it at http://www.microsoft.com/en-us/server-cloud/solutions/virtualization.aspx?code=witp#fbid=LaAUjM8aHUE.