Server virtualization has already changed the economics of disaster recovery in the midmarket, making multisite failover accessible beyond large enterprises. Now, smaller businesses without the budget for a secondary site or storage area network-based replication have built out their disaster recovery (DR) infrastructure with virtualization.
Such products can be prepackaged "DR in a box" solutions, as in the case of VM6 Inc. But just as often now, IT professionals roll these systems themselves, assembling DR platforms out of commodity servers and hypervisors to address their specific DR needs at a relatively low cost.
While the market for low-end disaster recovery is growing, it's still in the early stages.
VM6 circumvents the expense of storage area network-based replication between virtual data centers by creating a Hyper-V server cluster with a logical pool of networked storage made up from a server's internal disk drives.
Christian Boivin, vice president of Montreal, Canada-based JLR Recherche Immobiliere, or Housing Search, uses 6 TB VM6 clusters for local and remote disaster recovery. Two physical servers running VM6, Microsoft Hyper-V and the company's real estate listing information processing applications reside at the company's main office on the island of Montreal. Two other servers reside at a remote site on Montreal's south shore as a mirror of the primary site.
Boivin said he initially looked at VMware Inc. and a storage area network (SAN) for two servers at the main site, but was quoted a price of $125,000 for the package. VM6 came in at about $40,000, including servers and software licensing. Boivin was so compelled by the pricing that he took a chance on a fledgling company. "I was their first client," he said.
There were some growing pains early on -- especially when he tried to transfer large files -- but no downtime. The two clusters at each end of the wire are synced over a private VLAN with redundant networking switches and go to separate Internet service providers (ISPs) to guard against multiple types of failures. Failover between the two sites is automatic.
IT managers have also taken DR matters into their own hands. For example, Citrix Systems Inc. XenServer user Brad Williamson, network administrator for a banking software company based in Florida, says his company has to be mindful of hurricane threats. Currently, Williamson runs an all-XenServer environment, having switched from VMware's vSphere because of budgetary constraints. These constraints also preclude advanced disaster recovery configurations with networked storage and distance replication.
In the interim, Williamson loaded Citrix's XenServer onto a small PC with plenty of local storage that he uses as a 'shuttle box.' Given a day or so notice, production servers can be shut down and exported to the shuttle server, which can then be unplugged from the network and taken to a hotel or other off-site location to keep the company running. The company hopes to turn that 'shuttle box' into a built-in DR feature of its software for its clients.
A 'shuttle server' doesn't protect against various kinds of failure, particularly those that can't be easily forecast. But it's better than tape, and takes "one box instead of a pallet of equipment," Williamson said.
Virtual DR in the cloud
Gregory Rosenberg, the CEO of Red Hat value-added reseller RICIS Inc., says his smaller clients use Novell's open source tool SUSE Studio, a kit for preconfiguring and packaging virtual machines. Rosenberg said SUSE Studio can create appliances that run on local VMware hosts, as well as Xen, XenServer, and even public cloud services like Amazon's Elastic Compute Cloud. "It allows them to have a standby virtual instance at Amazon, their own [VMs] at their own data center, and keep the image live in either case."
If users would prefer not to go with a public cloud offering, Tom Dugan, the director of technical services for Recovery Networks, says they can leverage virtualization to back up or replicate their workloads and use his company's private cloud service as a secondary site. Recovery Networks creates virtual appliances from backed up images of production servers created with Asigra data backup software, and also offers the option of using Double-Take's replication software for more continuous protection. This approach protects against a total disaster at the primary site with geographic redundancy, while the local backups can also be used to protect against data corruption during replication.
Chris Wolf, a research vice president at Gartner Inc., said that while this market for low-end DR is growing, it's still in its early stages. For example, approaches that require the restore of a full VM image may be cumbersome when users want to perform item-level operational restores. Many of these solutions, including do-it-yourself approaches, don't offer an index that makes it easy to locate a specific file or image, or require manual mounting of an image before specific files can be recovered.
"It's still a market that has some growing up to do in terms of the number of solutions and their completeness," Wolf said.
Beth Pariseau is a senior news writer for SearchServerVirtualization.com. Write to her at firstname.lastname@example.org.