One clear example is the shift from tape to disk backup. DR providers have long made a living picking up and storing tapes. Now, almost all the vendors interviewed for our DR trends story said they were advising clients to move from tape to electronic vaulting for backup, both to save money and improve recovery service levels.
While tape certainly is less expensive than a replication infrastructure, providers said that CIOs need to factor in the "people and risk" costs associated with tape in a recovery scenario, including the time spent on managing the tape, the risk of lost tapes, and being unable to recover tapes.
Consultant Jon Toigo, CEO of Toigo Partners International, a DR expert and contrarian, argues that many of the problems with tape backup can be fixed by batching jobs according to volume of data and connecting interface speed to the network. "Set up the job as a string of sequential jobs," he said, by investing in a storage resource management tool.
Server virtualization is being used by both companies and their providers to drive down recovery costs.
"In the old days, if you had to recover 200 servers, that was 48 long hours, waiting for tape to spin, factoring in tape errors, system errors -- and there was always the question, 'Are we going to get the end state?'" said Bill Hughes, director of consulting services at SunGard Availability Services LP. A recent customer with a virtualized IT environment recovered 170 systems in 12 hours on the first test.
Bob Boyd, CEO of Charlotte, N.C.-based Agility Recovery Solutions Inc., said virtualization has changed the kind of technology that companies need for recovery. "Companies are going down to a vastly smaller number of servers in recovery mode but with much bigger processors and much bigger storage needs," he said.
Virtualization carries risks of its own, of course, Hughes hastened to add. "You are putting more eggs in one basket and there is always the chance that a single incident will take down more infrastructure," he said. The key is to design virtual systems with recoverability in mind from the start and to think through risks. When the recovery is at a different location, for example, IP addressing is an issue. "When you lose a virtualized image in your data center, it is easy to stand it up because you're in the same IP space," he said. "It's a little different when the recovery center is 6,000 miles away."
Toigo also sounded the alarm on the use of server and storage virtualization in DR at a recent Storage Decisions Disaster Recovery Seminar in Newton, Mass., cautioning users that virtualization can obscure visibility into the underlying infrastructure and complicate recovery.
IT disaster recovery in the cloud
The recession is squeezing IT disaster recovery budgets, but the big news in DR, according to Sam Gross, vice president of global outsourcing technology solutions at Unisys Corp., is not cutting costs -- "IT has been asked to do that since the creation of man," he said -- but how cloud computing is effecting a "tectonic shift" in DR. Companies can use cloud computing to create carbon copies of their applications that can be brought to life at a moment's notice.
"The economics are compelling," Gross said, noting that the big DR providers are heading to the cloud. Data security remains the chief concern, he said, touting his company's new Secure Cloud offering. Unisys has developed a set of technologies to cloak data in motion by splitting it into multiple packets, he said, and the company is working with customers to identify changes they need to make in the data center to take advantage of the "new era."
Of course technology is not the only way to improve the cost of DR. Many providers are also recommending changes to how the business assesses their needs, and to the underlying business processes, to further contain costs.
Let us know what you think about the story; email: Linda Tucci, Senior News Writer