Ensuring business continuity for an enterprise data center is a challenging task, but virtualization makes disaster recovery management much easier. Keeping servers, applications and data inside portable hardware-independent VMs increases the chances of protecting them from disaster. The blocks for each VM disk file can be more easily monitored for change and replicated to a DR site where those VMs can run on just about any server available.
For many years, only large enterprises could afford disaster recovery. It took SANs with hardware-based replication using high-speed wide area networks (WANs) or application-specific high-availability solutions. Such a system could cost hundreds of thousands of dollars.
Today, many editions of major virtualization platforms offer built-in availability and disaster recovery management features. Examples include the following:
- Backup/recovery. This feature provides the ability to perform a virtual machine backup at the image level and, typically, track the blocks of the VM disks that change to significantly reduce the time to perform future backups. For example, VMware vSphere, starting with the Essentials Plus edition, includes vSphere Data Protection (VDP), which uses VMware's change block tracking capability.
- Replication. Virtualization hypervisors can access VM images and track changed blocks, and those changed blocks can be easily replicated to a secondary data center for large-scale DR data protection. VMware vSphere includes this capability in its Standard, Enterprise and Enterprise Plus editions. Microsoft Hyper-V 2012 now offers Replica, which will perform asynchronous replication of individual VMs to another Hyper-V host at another site.
- Snapshots. Virtualization hypervisors can take a point-in-time "picture" of a VM's state and revert to that snapshot if needed. DR applications can use these snapshots to perform backup or replication for quick recovery when upgrades or configuration changes are made to a VM, OS or applications. Microsoft Hyper-V and Citrix Systems XenServer are good examples of platforms that offer snapshot and revert options for each VM.
- High availability. When a host failure occurs, virtualization hosts in the same cluster can easily claim VMs on shared storage. These HA solutions can restore VMs in the time it takes a VM's OS to restart. Examples of built-in hypervisor HA features include vSphere High Availability and Hyper-V Failover Clustering.
- Clustering. Virtualization hosts can participate in virtual clusters to help balance resources among multiple hosts. Clusters are typically used to prevent performance degradation and application slowdown. For example, VMware's vSphere High Availability will automatically restart VMs from a failed host.
- Fault tolerance. Used for limited disasters within a data center, this vSphere feature allows the virtual memory of a running VM to be replicated to another host in the data center so that the replicated VMs suffer no downtime when a virtualization host fails.
- Virtualization-aware disaster recovery management software. Helps plan which VMs will be replicated and the order in which they will be recovered. It can also be used to test failover, as well as failback to the primary site.
Third-party disaster recovery options
In addition to built-in capabilities, third-party technologies can help facilitate disaster recovery management. Examples of common capabilities include the following:
- Virtualization backup with replication. Most virtualization backup tools now offer access to cloud-based storage for low-cost (and easy) recovery of VMs. An example is Veeam Backup and Replication.
- Virtual SAN storage that includes replication. Other virtualization storage solutions want you to store your VMs in their virtual appliances that are then replicated. Examples include Ctera Networks and TwinStrata.
- Disaster Recovery as a Service (DRaaS). Many public cloud providers offer off-site DR services for virtual infrastructures. Typically these work by the customer implementing a replication application that replicates VM changes to the provider. The replication application could be one that is included with your hypervisor or one that you license separately. Customers often pay per gigabyte to store data and can benefit from a third party's DR expertise. DRaaS providers offer a portal that can help maximize a DR cloud infrastructure, which can even be used for file-level restore and virtual lab environments. Examples include nScaled, Veristor and Hosting.com.
These built-in and third-party options can ease disaster recovery management, particularly replication, and make it cheaper than ever before. Although 100% virtualization is not realistic for many companies, strategic use of the virtual assets you have can make disaster recovery less complicated and less expensive.