Chicago Firm Improves Failover With Virtualized BC/DR Services

Last summer, Mark Zwartz, manager of IT for the privately held JMB Companies, had a difficult choice: Continue managing a pool of physical servers at the business-continuity/disaster-recovery site for the Chicago-based insurance and realty company, or switch the company to a fully virtualized BC/DR site.

The company's data center was running out of space at its production site, and because the servers were starting to age, maintenance costs were escalating. One of the problems he was having with managing physical servers for fail-over was that a fail-over rarely ran as smoothly in action as it did on paper.

"We started to rethink what new technologies are available, and we seriously began considering the possibility of virtualization," says Zwartz.

For about three years, JMB had been replicating to its BC/DR site with physical servers using Sungard Availability Services. The servers were running Microsoft Exchange for email, SharePoint for content management, and BlackBerry Enterprise Server. Double-Take Software was used for data replication across both sites. Unfortunately, the hassles associated with hosting physical servers at the BC/DR site were many. First, certain applications such as SharePoint didn't replicate well, he says, and the secondary site was located about 35 miles away from the primary data center.

When something went wrong, it was time consuming and costly to have someone drive out to the BC/DR site. "There are all sorts of services you can purchase for remote management, but at the end of the day you still have to go out. Hard drives go bad and basic hardware issues arise, even if it just requires a power cycle. The reality was that it is going to cost two hours of productivity."After evaluating a number of options, Zwartz decided to stick with SunGard and use its Virtual Server Replication and AdvancedRecovery services. The transition went swiftly. "We managed to replicate 10 servers inside of two weeks," he says.

The move quickly proved itself. Not only did virtualization cut down hardware costs, but it also cut lost work hours devoted to maintaining the remote site -- and failover improved dramatically. When a failover is initiated, the SunGard service moves virtual disk files to production mode at one of its highly secure data centers, within a six-hour recovery time.

All aspects of replication and failover are monitored and managed by SunGard. "A few weeks after we virtualized, we had a SharePoint disruption, and the Virtual Server Replication service responded with a flawless system failover," says Zwartz. "It was perfect, came up like a charm."

While the switch quickly proved itself, it wasn't easy initially to sell the approach to business managers, Zwartz acknowledges. "It was very hard to explain virtualization to people who don't understand the concept. When they buy 10 servers, they see 10 servers come through the door. With virtualization, there's nothing to see. Early on, management didn't like the idea, and it was hard to even get approval for a beta evaluation."

But Zwartz didn't give up. "For a month, I would send them articles about virtualization. I would show them Webcasts, I would share with them anything I could to make the case." What changed management's mind was a clever demonstration Zwartz put together. "I created a virtual PC on a flash drive. And then I explained to them that these big PCs on their desks that weigh 25 pounds with the monitor are in this tiny flash drive. When the virtual PC booted on the flash drive, they started to catch on to what virtualization was."0