The effect of hypervisors on disaster recovery could be seen as a mixed bag. In this Storage Decisions video, disaster recovery expert Jon Toigo, of Toigo Partners International, discusses some of the pros and cons of the technology and how it impacts the disaster recovery process.
"Now the good news about hypervisors … is that it has increased the awareness of the value of building in certain recovery capabilities," said Toigo, who noted that applications weren't originally designed to be recovered, and part of DR was finding a retroactive way of adding that capability.
"At least some of the attention of the hypervisor stack has been to figure out some way to build in that capability, instead of bolting it on afterwards," he said.
Hypervisors also help encourage mapping where application data is stored, and free users from vendor lock-in of hardware, he said.
"The idea of flexible application rehosting -- going from physical to virtual, going virtual to virtual or going from virtual to physical -- I think this is wonderful. Anything that helps keep us off of a dependency for a particular hardware vendor is a good thing," said Toigo.
He pointed to the University of Texas, which fails over its Exchange server -- and its 40,000 or so mail accounts -- to an MSP using VMware and a 1U rack mount server during emergencies like hurricanes. This allows continued operation during an emergency, improves the DR process and avoids the cost of having to build a redundant email system in a second location, he said.
"The cost of the solution is minuscule, compared to building out a fully redundant capability," said Toigo, who later called VMware "a tool that we as disaster recovery planners can keep in our pockets, something that we can use judiciously."
But hypervisors can still have problems, such issues involving management of I/Os and hypervisor or cloud incompatibilities.
"Cloud service providers don't cooperate with each other. The incompatibilities between the software stacks that have been developed for server hypervisor computing continue to paralyze our ability to share resources and data efficiently in the cloud environment as well," he said.