Everyone from the chip makers Intel and AMD, to the virtualization software vendors VMware and Microsoft are pretty sure the future of the data center (and some say mainstream computing) is a virtualized one.

So sure are the processor makers that they’ve promised an indefinite roadmap of future chips with technology that aids server virtualization--the running of multiple OS instances on a single machine that’s not bound to the machine’s physical hardware.

Speaking this morning in Singapore with VMware’s executive vice-president of worldwide field operations, Carl Eschenbach, who said in a very well rehearsed manner: “Virtualization is now mainstream. It is not just used in test and dev, but in live production servers.”

And concluded with, “VMware’s vision is one where whole data centers are virtualized.”

So the question at hand is “If my live servers are now virtual, how ready is virtualization to my handle my disaster recovery (DR) needs?”

Virtually Possible

This morning, in a crowded conference room with only standing space available, VMware’s technical trainer for professional services, Brian Rice spoke to an audience of almost 300 about a customer of theirs who has had VMware’s ESX Server running in a data center for over 700 hours without down time.

“The virtual machines running on it have crashed but not the virtualization kernel,” he said.

The specially created and proprietary kernel is designed for nothing else than to host virtual machines, he explains, citing that point as robustness that would allow high availability and business continuity.

He goes on to show how VMotion, a tool comes with VirtualCenter and works on top of the ESX Server, can seamlessly without downtime migrate a virtual machine from one physical machine to another. A God-send for anyone who’s ever done it the old-fashioned clone-the-harddisk way.

But there’s a catch. For that to happen so flawlessly, the virtual machine has to exist on shared storage such as a SAN or even something as simple as an iSCSI NAS.

Rice goes on to explain that ESX Server, VMware’s flagship data center grade product, has some built-in redundancy and intelligence such as the ability to automatically detect failed hardware and reboot the downed servers on another physical x86 machine, even one from a different vendor.

That’s really good in terms of high availability of live servers, but can it manage real-time DR?

According to Eschenbach, the answer is “no”.

“There is no instantaneous DR,” he said. “The advantage here is that the technology is very portable, each virtual machine is stored as a flat file and replication on a different physical server is not an issue.”

Echoed by Intel, one of VMware’s biggest fans who has been touting their hardware virtualization technology, the next-generation of its VT technology will be ideal for real-time failovers and disaster recovery.

Said Ravi Ravichandran, regional marketing manager, server platform group at Intel Technology Asia, “The eco-system is not ready with I/O virtualization including players in networking and storage.”