Virtualization has changed the way companies plan and execute disaster recovery (DR). The flexibility and consolidation in a virtual infrastructure makes virtual disaster recovery more efficient and cost-effective.
No longer dependent on hardware, a virtualized infrastructure allows multiple virtual machines (VMs) to run on fewer physical servers. If a failure occurs in a primary data center site, you can migrate a workload to another server at the DR site without concern for the affected hardware.
The answers to these frequently asked questions cover the benefits and drawbacks of virtual disaster recovery, how to develop a DR plan with physical-to-virtual backup strategies, and information on disaster recovery planning with VMware or Microsoft Hyper-V.
How does virtualization affect a disaster recovery and data protection strategy?
When developing a virtual data protection plan, it's important to use virtualization-aware tools for backup, deduplication and other processes. Shared storage allows VMs to migrate easily to other servers, but virtual disaster recovery preparation is still necessary. Make sure to copy VM workloads off-site regularly, and remember that VMs on a storage-based network provide fast recovery through snapshots, rather than disk-based or tape backups.
How does server consolidation improve virtual disaster recovery?
By consolidating multiple servers into a single VM, virtualization improves disaster recovery operations. Using snapshots stored at a disaster recovery site, you can recover data and entire VMs faster. The key is the virtual machine disk file, which contains the entire OS environment, including the required configurations for the new VM at the disaster recovery site. Redeploying the workload with this single file means you can recover an entire server in hours rather than days.
What are the differences between virtual disaster recovery and physical DR?
The main difference is simple: Virtualized disaster recovery focuses on backing up virtual machines instead of physical servers. Whereas physical DR relies on local-client backup, VM backup can be done in three ways: agent based, image based and server-less backup. With data replication, workloads move freely between VMs and return quickly after recovery, without having to manually re-launch the OS and application on a physical server.
Are there drawbacks to virtual disaster recovery?
The benefits of virtualization can become drawbacks in the virtualized disaster recovery realm. Virtual infrastructures are inherently complex, which makes it difficult to track a problem to its physical source. And although consolidation is a main benefit of virtualization, it also means that a fault on one physical server affects all VMs running on that server. Consider the pros and cons of using virtual servers for disaster recovery.
Can I achieve successful disaster recovery with physical servers?
If a physical server isn't a good candidate for virtualization, don't worry. You can still achieve virtual disaster recovery through physical-to-virtual (P2V) backup. P2V backup means you can create virtual backups of physical servers you never intend to run virtually. It's somewhat like taking the first step toward virtualization without continuing on to step two. P2V software converts the software, data and settings on a physical machine into a single disk file, which you can then store in case of a disaster.
How do host-level backups help virtualized disaster recovery?
Host-level backups provide complete copies of all VMs on a host, including the operating system and all applications. That means you no longer have to back up each machine individually. Hard disk and configuration files provide those copies and pass them through the host to your backup service. Make sure you have name resolution and an authentication source to make this process seamless. Restoration times for failed VMs with host-level backups are much shorter than those with guest-level backups.
What's the best way to deal with a hypervisor's single point of failure?
Back up virtual machines regularly, because the single point of failure can cause all VMs on a server to fail from one fault. Microsoft, for example, requires hardware data-execution prevention as a prerequisite for Hyper-V installation, which helps to improve restore speed. And separating Hyper-V host networking from VM networking also improves security by segregating traffic. But remember: a hypervisor's single point of failure affects all virtual platforms, not just Hyper-V.
With Microsoft Hyper-V, how can I create a virtual disaster recovery plan that covers both primary and backup sites, without breaking the bank?
Windows Failover Clustering supports disaster recovery implementation at multiple sites. First, be sure your storage device at the disaster recovery site can accommodate the workload of all VMs at the primary site. It's also important that the networking infrastructure allows communication between both sites. Hyper-V clusters require two-way replication so VMs can migrate back to their original nodes once recovery is complete. Also consider which Hyper-V disaster recovery capabilities you need and the cost of implementation.
Where can I learn about disaster recovery strategies for VMware environments?
VMware Inc. and its partners offer many online resources about VMware disaster recovery methods -- from information on backup and replication software to plans for automated virtual disaster recovery. There are also guides and podcasts explaining business continuity and virtualized disaster recovery planning for VMware shops. And resources on VMware Site Recovery Manager, Double-Take Software Inc.'s replication software and EMC Corp.'s RecoverPoint can guide you in a virtual disaster recovery plan. You can even take a VMware disaster recovery assessment survey to find out how successful your DR plan will be and which improvements to consider.
This was first published in December 2010