Live Migration Will Make Virtualization Mainstream
May 12, 2008 Timothy Prickett Morgan
The ability to teleport running applications around the network–what most vendors are calling live migration these days, but which comes in various names and server virtualization flavors–is probably one of the coolest technologies to come along in the data center in quite some time. And it may turn out to be the technology that pushes server virtualization through its tipping point, finally making such virtualization truly mainstream.
The ability to virtualize servers with a hypervisor and then get multiple (and sometimes incompatible) operating systems to share a single box was useful for server consolidation (getting rid of some boxes) and for driving up utilization on expensive resources inside those boxes at the same time. But making applications or whole operating system instances mobile has huge implications for improving the availability of applications and their underlying operating systems and for making a new kind of high availability–and a cheaper and perhaps less complex alternative to system clustering–available to a whole new breed of customers who pray every day that their systems stay up because prayer is the only option they can afford.
According to a report released by IDC last week, called The Future of Virtualization: Leveraging Mobility to Move Beyond Consolidation, server downtime cost companies, governments, and other organizations a total of $140 billion in lost worker productivity and revenues in 2007. To put that number in perspective, that’s roughly one-eighth of the total IT spending globally (including telecom and people costs) and almost three times the amount of money that companies shell out each year for servers. Those are some big numbers for server downtime costs, and that explains why the most sophisticated and forward-thinking companies have spent so much money on high availability clustering over the past decade and a half.
While many people are looking for desktop virtualization to take off next, with companies putting virtual PC slices on servers back in the data centers and plunking thin clients onto their corporate desktops instead of full-blown PCs, John Humphreys, program vice president in IDC’s enterprise platform group, thinks that server virtualization will be used to cope with system downtime before it will take off on the desktop. The ability to move live workloads around machinery is a key reason why Humphreys believes this to be true.
“By directly addressing the need for cost effective business continuity, virtualization will alter the economics of IT a second time,” explains Humphreys. “More importantly, mobility will be the defining feature that will move virtualization beyond just a tool for consolidation. The embrace of mobility will allow customers to use virtualization for business continuity, capacity planning, and eventually as a solution in delivering service-oriented computing.”
Because everybody has to relate any and every IT effort to SOA and cloud computing these days, Humphreys does so in the report, saying that the nirvana for IT would be to have “policy-based automation,” whereby SOA-enabled applications, themselves abstractions sometimes of underlying legacy applications more tightly tied to their underlying iron, are hooked into management systems and server virtualization hypervisors to make one giant whirligig of flexible software–including moving applications out onto the “cloud” in a utility-style computing environment.
Good luck debugging that, fellas.