Disaster Recovery in the Cloud Yields ROI

CIO — The promise of cost savings derived from cloud computing is attractive, but concrete financial returns are not always quickly achieved. Except, perhaps, when it comes to disaster recovery.

Offloading the expense of buying your own hardware, software and networking capabilities for use during a prolonged outage, as well as that of the ongoing maintenance of duplicate infrastructure, can produce measurable savings, says Eric Heidrich, director of IT at Help At Home. He expects to save $150,000 over three years, mainly by cost avoidance.

Infrastructure as a service is the original impetus for cloud computing, says Judith Hurwitz, CEO of the consultancy Hurwitz and Associates, and it can be cost-effective. Other companies also see cloud computing as a means for disaster recovery. TheStreet.com hired Iron Mountain (IRM) to back up and restore data and applications in the event of a prolonged outage, CIO Daniel Flax explained at the recent CIO Perspectives conference in New York.

Help At Home is a private company that assists the elderly and disabled with household chores—like cleaning, cooking and errands—and routine health care. The company employs 13,000 field workers in nine states.

Since Heidrich arrived at Help At Home in 2008, he has virtualized nearly all of the company’s servers on VMware (VMW). He had been doing backups with SonicWall’s continuous data-protection appliance. The vendor constantly searched files and databases for changes, and Help At Home paid a monthly fee, he says.

But SonicWall’s appliance wasn’t backing up operating systems or imaging any servers, because that required the added expense of installing SonicWall software on every Help At Home server. Yet restricting SonicWall’s use to only data meant that in a disaster—such as a long power outage caused by a storm—no server images or application configurations would be saved. This type of outage hasn’t happened yet, but the company must prepare, Heidrich says.

For full disaster recovery, he first considered making one of the company’s 85 offices the fail-over site. But duplicating and maintaining that infrastructure would have cost about $190,000 over three years, he says. That price tag includes a four-node storage area network, two switches, three servers and an uninterruptible power supply; a fast Internet connection between the main office and the branch; about 120 hours of set-up and configuration time (amounting to $15,000 in labor) and three hours per week of ongoing maintenance, at a cost of $20,000 over three years. Instead, Heidrich found Iland, a cloud infrastructure vendor, through a Google (GOOG) search for VMware-related disaster-recovery ideas.

Using Iland costs Help At Home only $48,000 over three years, Heidrich says. He counts the difference—about $140,000—as money saved. Iland taps into Help At Home via a VPN and uses its replication product to copy any changes in VMware images to Iland servers nightly. Some critical applications, such as client management and billing, are replicated more often. If a disaster occurs, Iland can restore all of Help At Home’s virtual machines on its own hardware, and branch offices will be redirected to Iland.

Restoring in the cloud is much quicker than other disaster-recovery scenarios, Heidrich says, “and there’s no hardware to buy.”