DR Units Respond to Hurricane Georges

When Hurricane Georges roared ashore and struck the town of Pascagoula, Miss., one of the casualties was the roof above the data center of the Ingalls Employees Credit Union (IECU). The damage left the company little choice but to make a disaster declaration to the company’s disaster recovery provider, setting into motion a sequence of events company managers had hoped to never use.

• 10/21/1998

At the time, IECU data processing manager Chris Crabtree recalls saying to himself, "Here we go again." This is not the first time IECU has been struck by severe weather. "We lost half the roof about 10 years ago, but it spared the computer room," explains Crabtree. "This time it pealed all the way back and spilled into our backyard." Crabtree says access to his company's data center was crucial. "When one of these [hurricanes] comes through, people want to leave and they want access to their cash," he says.

IECU called on SunGard Recovery Services (Wayne, Pa., http://recovery.sunguard.com) to help them through the storm. "I told them we needed everything, send it, and they did," Crabtree explains. SunGard immediately sent out their 18-wheel mobile data center and reached Pascagoula in 8 hours.

"People pay us to be prepared for those things," says Doug Clauson, spokesperson for SunGard. He says that each company's recovery window -- the maximum time systems can be offline before there is major financial loss -- varies based on the industry and business requirements the IT infrastructure supports. "Some companies can handle much larger windows, while others not only require a shorter window, but they need to meet regulatory guidelines," Clauson explains. SunGard was put on alert by 75 of its customers, 8 of which eventually made disaster declarations.

"For us, it was the largest simultaneous recovery effort ever taken," says IBM Business Recovery Services North America manager John Nevola. "We recovered 31 customers with a total of 43 configurations from Puerto Rico to Florida and the Gulf Coast." As it turned out, the Puerto Rico situation was the worst, Nevola says, where business damage was widespread. One of the most difficult problems was widespread loss of power.

IBM’s recovery process starts with an alert from a customer, when IBM prepares for a possible declaration. "In a situation such as this, they call our command center to make them aware they'll need service," says Nevola. We set up a site that matches their configuration. Within 12 hours, a customer can be relocated to a destination and begin the process of restoration."

Burned before by Hurricane Andrew and other tropical weather threats, Florida-based companies took the threat seriously. "They didn't wait until the hurricane cut off access. They took action days before," observes Nevola.

Rosemont, Ill.-based disaster recovery company Comdisco Inc. (www.comdisco.com) also found itself supporting 17 recoveries. Gerald Butler, a Comdisco recovery specialist, rebuilt a Florida retail company's Windows NT-based data center in Comdisco's Atlanta recovery center. The first step was to build an image of the company's environment and then burn that onto bootable CDs. Once the workstations were booted up, Comdisco connected to the company's server using a hot node. A hot node is a constant connection from the customer's home site to the Comdisco recovery site. The node is kept active on a daily basis.

Butler ran into some complications in reconfiguring an NT environment that would exactly match the original site, with registry settings causing some incompatibilities. Ultimately, the company chose to rebuild the registry from scratch.

IBM's Nevola says it's important to remember that while a hurricane inflicts the most geographically widespread damage to data centers, it's far from the only reason a company might declare a disaster. "It only takes one failure to put a company out of business," says Nevola. "There are so many different risks other than hurricanes."