Moving virtual machines between sites could radically change disaster recovery, the two companies have said
Cisco Systems and VMware are planning to use VMware’s VMotion technology to move live virtual machines from one data centre to another.
The two companies showed off a proof-of-concept at the Cisco Live 2009 show in San Francisco, and demonstrated the capabilities during Cisco CTO Padmasree Warrior’s keynote address 2 July.
The project is still in the proof-of-concept stage, but VMware official Guy Brunsdon said in a recent blog post that moving live virtual servers to other locations over a WAN holds promise for businesses in a number of areas.
In particular, the capability would help enterprises in load balancing compute resources over multiple sites, Brunsdon said in his blog posted this week. Businesses also could save power and cooling costs by being able to dynamically consolidate VMs to fewer data centres, he said.
In addition, businesses could avoid downtime during maintenance procedures in data centres by migrating applications offsite, and they also could more easily avoid natural disasters by proactively migrating important application running on VMs to another facility.
VMotion has worked well in migrating live VMs from one host to another. In addition, VMware offers disaster recovery capabilities with its vCenter Site Recovery Manager, which enables businesses to improve their disaster recovery capabilities through automating recovery steps, testing recovery plans without interrupting the VMs, and providing steps for building and managing disaster recover plans.
However, there are particular challenges to the idea of moving live virtual servers from one site to another, Brunsdon said.
“This, of course, is a non-trivial thing to do,” he wrote. “There is the challenge of moving a VM over distance (which involves some degree of additional latency) without dropping sessions. To maintain sessions with existing technologies means stretching the L2 domain between the sites — not pretty from a network architecture standpoint. And then there is the storage piece. If you move the VM, it has to remotely access its disk in the other site until a Storage VMotion occurs.”
For example, both the data centre maintenance and disaster avoidance scenarios would require a Storage VMotion to move the disk image to the other data centre.
Cisco and VMware engineers last year began working on the idea of moving VMs over long distances between multiple data centres, Brunsdon said. The joint Cisco-VMware lab in San Jose, California, has run several tests of disparate distances, he said. The demonstration at Cisco Live covered a distance of about 50 miles, he said.
According to a diagram of the San Jose-to-San Francisco test, the San Jose site includes VMware ESX servers and Catalyst 6500 switches from Cisco. At the San Francisco site were ESX servers and Cisco’s Nexus 5000 and 7000 switches.
Linking the two sites was an 80-kilometer single-mode optical fibre.
Gordon Haff, an analyst with Illuminata, said disaster recovery is a benefit that VMware has been touting with virtualisation for several years.
A key benefit is the ability to create a disaster recovery plan that doesn’t entail spending the money to buy compute resources and having them sit idle in case of an emergency, Haff said. Virtualisation enables businesses to work with the systems they have and use VMs for disaster recovery needs.
“You can use most resources normally most of the time,” he said. “But in the case of a problem, you can shift resources, but you don’t have a lot of idle resources.”