With the next release of Windows Server operating system, Microsoft has conquered one of the thorniest problems in virtualization: moving an operational VM (virtual machine) across a WAN (wide area network).
Microsoft unveiled this new feature at the Microsoft Build conference, being held this week in Anaheim, California. It will be available in Microsoft's Hyper-V version 3 hypervisor, included in Microsoft's next server operating system, Windows Server 8.
[ Doing server virtualization right is not so simple. InfoWorld's expert contributors show you how to get it right in this 24-page "Server Virtualization Deep Dive" PDF guide. | Use server virtualization to get highly reliable failover at a fraction of the usual cost. Find out how in InfoWorld's High Availability Virtualization Deep Dive PDF special report. ]
The capability, while viable in only a handful of use cases, nonetheless shows Microsoft's growing expertise in virtualization, analysts said.
With live migration, a working VM is moved from one computer to another without any disruption of the services offered by applications within the VM.
Most server virtualization software providers, including Microsoft, have long offered live migration within a local network. A far harder problem has been to offer the ability to move a live VM across different subnets, or separate WAN sub-networks. Network latencies and network addressing complexities have made this task a challenging one.
At the BUILD conference, Microsoft program manager Ross Ortega explained how the new technology worked. Essentially, each virtual machine gets two IP (Internet Protocol) addresses.
One address is the home address for the VM itself, which it uses to communicate with the rest of the network or, if accessed publicly, the Internet. The second IP address will be available for the local data center.
"The trick is to keep these mappings between the two address spaces," said Bill Laing, Microsoft corporate vice president for the server and cloud division, in a subsequent interview.
Microsoft is not alone in offering this capability. VMware, in its VXLAN offering, also offers the ability. This solution, however, requires Cisco networking gear using the OTV (Overlay Transport Virtualization) technology.
Long-distance live migration has been a research topic in the virtualization community for several years, though it is useful only in a limited number of situations, said Chris Wolf, a Gartner research vice president covering virtualization.
Replicating a VM memory state over long distances can take a considerable amount of time, Wolf noted. It could be handy for emergency situations, such as when a data center is threatened by an upcoming flood or hurricane, and its contents must be moved to another facility. But, given the sizes of most data center-run systems, it is likely that only a handful of VMs would be moved in time.
During another presentation at Build, Brian Dewey, Microsoft group program manager for Hyper-V noted one way around this problem. Administrators could also save VMs on thumb drives or optical disks, and then ship them to the new location, where the VMs could then be incrementally updated with the latest changes.
Another possible application of the technology would be in permanently migrating systems from one data center to the next without incurring any downtime, Wolf said. Such work would be necessary in data center consolidation projects, for instance.
Beyond the practical uses, the new capability carries symbolic importance for the company as well, Wolf said. It shows Microsoft is knowledgeable enough with this emerging technology to stay on the frontier of its developments.
The technology "marks a point where Microsoft will ship the first real formidable challenger to VMware vSphere," Wolf said.
At BUILD, Microsoft touted a number of other new features to make Hyper-V and Windows Server more competitive with VMware as well. The company debuted the ability to do live migration of the VM's virtual hard disk. Scalability was also touted: the new software will able to support up to 32 virtual processors for each VM, and each VM can now appropriate 512GB of memory. A new virtual file system, VHDX, can be used to create virtual storage disks up to 16TB in size.