Why Server Virtualisation Is Not a Disaster Recovery Plan

It’s funny how some myths continue to be believed, even by hard-nosed business people. The notion that virtualisation will save a company’s data is such a myth. Although it can be valuable in optimising an organisation’s use of IT resources and reacting quickly to changing IT needs, virtual environments are not inherently safer than independent physical servers. But data recovery provider Kroll Ontrack found that 80 percent of companies believe that storing data virtually like this is less or no riskier. Beliefs are one thing, statistics are another. 40 percent of companies using this virtual mode of storage were hit with data loss in 2012 – 2013. What’s going on?

Virtualisation can bring advantages in efficiency, but also introduces complexity. The amount of data stored over several virtual servers is often much bigger than that stored on a single physical storage device. Opportunities for problems or failure are multiplied. They include corruption of file systems or virtual disks, virtual machines that are incorrectly deleted, and hardware failures at RAID or other storage media levels. As the saying goes, the question is not if a hardware storage device will fail, but when. The same device interlinked to any number of virtual universes can damage them all.

A disaster recovery plan that takes into account virtualisation is another matter. In this case the plan needs a solid solution to the recovery of a possibly complex virtualised set-up, including all relevant data and configuration information. Alternatively, cloud mirroring may be a solution. Cloud providers are faced with the same risks of hardware failure as any other entity. However, they have (in theory) already thought out data replication strategies so that even virtualised environments will not be compromised. Other items to iron out for IT teams using virtualisation include figuring out how to handle mixed physical/virtual environments (a very frequent configuration), defining formal recovery objectives, and ensuring crash consistency if a virtual machine fails during backup.