A New Game Plan

With CEOs paying more attention to business-continuity planning, IT departments are looking for ways to provide higher levels of data availability. But real-time data replication is notoriously expensive, and technology budgets continue to be under pressure. That's got IT professionals seeking-and technology vendors promising-creative new ways to keep information flowing in emergencies.

Until recently, reliable disaster recovery typically required duplicate, or mirrored, computer systems located in another city, with up-to-date copies of a company's computer and storage hardware and software, and instantaneous updates to these standby systems. It's like keeping a spare Mercedes in a garage in the next town, to be used only if the family Mercedes breaks down. In addition to being expensive and inefficient, this approach can be inflexible, locking a company into a single vendor's product line and requiring a high level of staff expertise.

New products and services from well-known technology vendors such as Amdahl and EMC, as well as from smaller specialists such as CYA Technologies, FalconStor, and SteelEye Technologies, are giving IT managers options that are flexible and more affordable. They make it easier to mix and match hardware, from low-cost appliances to high-end Unix systems. These offerings don't require that entire IT environments be mirrored; instead, they replicate data in real time, or near real time, to storage systems and servers that can be simultaneously used for other tasks. And they can do it all over IP networks.

Rather than having a "hot" standby site that's used only in disasters, the new technologies can be used to build greater redundancy into everyday computing environments. This approach may require some rethinking of networks and IT architectures, but the payoff is that data replication becomes "a natural business process," says Jim Johnston, CEO of the Standish Group, an analyst firm.

Maharam Fabric Corp., a wholesale textiles distributor in Hauppauge, N.Y., was prompted by the Sept. 11 terrorist attacks in New York to re-evaluate its data-availability plans. The company is beefing up its IP network with new routers and switches and plans to add wide area network functionality-to be released next week by SteelEye-to SteelEye's high-availability software that Maharam currently uses to replicate servers over a local area network, says Sal Deaugustino, the distributor's network administration manager.

Maharam had already taken steps to bolster its data-recovery capabilities, but they weren't enough. After a few system outages in the last year interrupted business, the company considered mirroring its Hewlett-Packard K Class 9000, but management balked at the $110,000 price tag. So the company settled on two Compaq ProLiant servers running Linux and clustering software, at a savings of around 60%. But when the company lost a 45-Mbps telecommunications line for eight days after the terrorist attacks, its New York office was "knocked out," and data was unavailable to or from that office for days, Deaugustino says.

Stories like that have technology vendors scrambling to help. HP is expanding its business-continuity services unit, and IBM is about to deliver a new service based on technology developed by its research group (see story, p. 26). EDS CEO Dick Brown last week credited the company's strong quarterly results, in part, to interest in its business-continuity services. "Businesses are realizing that investing in contingency planning is a necessity, not a luxury," Brown says. "These requests are no longer coming from middle managers, they're coming from CXOs."

Amdahl this week will introduce a new A1 Business Continuity service to help companies back up, replicate, and store data across a variety of hardware platforms. Even though Amdahl sells storage systems from its parent company, Fujitsu Ltd., Amdahl will work with other storage systems, including Hitachi Data Systems' storage, more evidence that single-vendor solutions have fallen out of favor.

Startup CYA Technologies will unveil software next week that lets businesses replicate and move information, and applications, as software objects. In the event of a failure, that means users would be able to access not just data, but the applications needed to make the data useful. Officials at disaster-recovery company Comdisco Inc. say some customers spent hours re-creating applications in the days following the attacks on the World Trade Center.

FalconStor began shipping a new version of its IPStor management software this month that supports replication and what's known as virtualization, in which data can be replicated and stored on multiple systems that serve as backups to each other. For $50,000 to $100,000, a company can use the software to protect 20 to 30 servers.

EMC, the market leader in high-end storage, is moving toward a more flexible approach. The company will unveil software this week that makes high-end capabilities, including EMC ControlCenter/Replication Manager, available for competing systems from IBM, Hitachi, and others. The software includes WideSky, a product that lets companies manage from a single console a variety of databases, file systems, Fibre Channel switches, storage systems, tape libraries, and more.

Disaster preparedness has always been a get-what-you-pay-for proposition, with elaborate real-time backup sites sometimes costing millions of dollars a year. But not every company needs, wants, or can afford that level of protection. "In most businesses, instant recoverability is a luxury not worth the money," says Edwin Floyd, VP of IT with Blue Cross and Blue Shield of Georgia, in Columbus, Ga. A thorough business-continuity plan must be flexible, but it doesn't have to mirror every IT system in a company, Floyd says.

USA Today has implemented network-attached storage from Network Appliance Inc. as a way of replicating data over an IP network to a secure site in Maryland. The newspaper considered implementing EMC's Symmetrix storage system, but that far exceeded the paper's data needs and required private data lines. Darrell Griffith, an IT analyst at the newspaper, says USA Today is paying about half of what it would have for the high-end approach.

The Tysons Corner, Va., newspaper's contingency plans were almost put to the test recently when an employee discovered powder in a piece of mail and some feared it was anthrax. Authorities took a couple of days to determine the substance wasn't dangerous. For a while, "we thought the county would shut down our facility," Griffith says. If that had happened, Griffith says, USA Today's IT systems would have stayed online because its economical backup data center is equipped to handle such a scenario.

North Shore-Long Island Jewish Health System is also testing a new approach that relies on network appliances rather than high-end storage. Richard Jerothe, director of the New York health-care company's enterprise infrastructure, says North Shore-LIJ considered using a disaster-recovery service provider such as Comdisco or IBM, or purchasing a storage area network based on Fibre Channel. Instead, Jerothe is testing a new and more-affordable system that includes Microsoft Windows-based storage over an IP network, network-attached storage, and a suite of software from startup FalconStor.

Jerothe isn't yet convinced that the FalconStor system can do it all. For backup and replication among the company's three data centers, all on Long Island, N.Y., North Shore-LIJ may call on a service provider. "From a purely financial viewpoint, FalconStor has potential, but we have to hammer out our strategy before we jump in," he says. He wants to assess how scalable, reliable, and fast FalconStor's solution is compared to a service like Comdisco's.

Either way, Jerothe knows that when he approaches top management with a business-continuity proposal, they'll be inclined to listen: "It used to be a harder sell," he says.

Bayer Pharmaceuticals in Osaka, Japan, an affiliate of the German company that makes the anthrax-fighting antibiotic Cipro, is installing backup, restore, and standby software from CYA Technologies as part of a project to replicate documents and front-end applications. CYA's Hot Backup synchronizes metadata and content, "making sure whatever happens, documents won't get lost," says Mithat Mardin, head of IS coordination. Given all that's happened, Mardin says, to buy IT systems without robust recovery capabilities is "simply foolish."

Ventive Health subsidiary Health Products Research Inc. is taking things even further. Abraham Jacinto, network engineering manager for the Whitehouse, N.J., company, is negotiating with FalconStor to design a system that uses virtualization to let the pharmaceutical sales research company replace a network of storage appliances with an inexpensive hard disk attached to a server.

FalconStor should help the company's IT department resolve data outages more quickly, Jacinto says. Further, if Health Products Research loses any of its EMC Clariion storage systems, FalconStor will automatically replicate the data and move it to less expensive hard disks. The setup won't perform quite as fast as Clariion for standard operations, but at least the company will have information available to users with shorter interruptions. Jacinto says that beats waiting for EMC to get Clariion back to work-a service that EMC is contracted to do within four to six hours.

In addition to disaster-recovery service providers such as Comdisco and SunGard Data Systems Inc., IT executives can opt to work with a new breed of service providers that focus on keeping data available at all times. Digex, Storage Access, and Storage Networks offer complete managed-storage services. These vendors, like SunGard, will house a company's storage systems and take responsibility for its data availability. They charge monthly fees based on capacity and levels of service. One thing customers should check: Does the storage service provider itself have fully redundant infrastructure and business-continuity processes in place?

Some storage providers also will let a customer store data at the customer's own facilities, and send support staff in to perform tasks such as replication and to manage the whole environment. That appeals to companies that don't want their most critical data residing off-site. But the slumping economy has created problems for storage service providers that had projected rapid growth and that, in anticipation of increased demand, had purchased or leased huge amounts of capacity that's now underused.

For all their promise, the new products and services have to be carefully evaluated. Some, for example, don't support mainframes, which are still used by many companies. John Webster, an analyst at Illuminata, acknowledges that there's money to be saved, but says IT managers need to gauge the time it takes to recover information, as well as "how much of it they actually get back."