Data Recovery: The Time Ripe For the Midmarket
Midsize enterprises in growing numbers are realizing that backing up data isn't enough these days. Gleaning lessons post-Sept. 11, customers now realize that information has to be recovered in a reasonable (meaning, fast) amount of time to avoid bringing their businesses to a grinding halt in the event of an emergency or natural disaster.
Dimension Data is a solution provider at the forefront of the backup arena. The company, which was founded in 1983, is headquartered in South Africa and now spans 30 countries, recently finished up a project for the manufacturer of electronics to government agencies. The 1,000-employee manufacturer, which asked to remain anonymous, was concerned it would take days to recover its tape-archived data to four U.S. locations if an emergency occurred, according to Kris Domrich, a senior solutions architect at Dimension Data. The customer's key information assets include CRM data and inventory records located on a variety of Oracle and legacy IBM UniData databases on Unix-based HP/UX servers.
"Stuff was getting backed up and tapes were being moved off-site, but it wasn't being done in a regimented manner, and it wasn't providing them the ability to have a system that they could essentially bring up in near real-time," Domrich says.
Suffice to say, Domrich notes, a prolonged outage could have impaired the customer's ability to conduct business. "It would have impacted their ability to provide service to customers, to be able to field inquiry and general customer-service inquiries because the applications would not be able to access the databases," he explains.
The customer gave Dimension Data a recovery-time objective of five hours and, like many mid-size customers, a limited budget. That was no small order given the off-site backup locations were hundreds of miles apart. Data Dimension recommended a solution that uses storage-based replication to mirror volumes of data from one array to another. The specific solution consisted of EMC Clariion arrays with Fibre Channel disks, EMC's MirrorView Asynchronous Replication software and Cisco's MDS 9000 family of SAN switches connected over a dedicated fractional T-3 circuit.
Because of the distance involved, it was decided that data from the primary site would be backed up to a secondary location approximately 150 miles away. That afforded the opportunity to be far away from the headquarters, though closer than the other offices.
The Cisco switch in combination with the Clariion arrays linked well because the MDS 9000 supports both native Fibre Channel, iSCSI and Fibre Channel over IP. The latter was the only suitable connectivity protocol because of the distance, Domrich says.
The solution used storage-based replication as opposed to host-based replication. The latter is preferred for volume-based replication because it doesn't impact CPU cycles.
Overall, Domrich notes, there's a growing demand for array-based replication among midsize customers. "Replication used to be so reserved for the enterprise," he says. "But we're seeing a lot more demand for it from the SMB."
SMB-focused storage solutions are also in demand among companies that have both a headquarters and multiple branch offices. For example, Cisco's new Wide Area File Services, or WAFS, is becoming a popular option for storage consolidation because it doesn't require infrastructure at each location. Such a solution consists of the Cisco Edge File Engine, a file-caching appliance for branch offices that replaces file-and-print servers at the branch location rather than caching the data and replicating it with a server or NAS appliance at the headquarters. There is also a management module that allows IT administrators to monitor and configure the remote file engines from the headquarters, thereby eliminating the need for on-site support at the remote office.
Domrich says he is seeing a large amount of interest from small regional banks that may have one central office and several dozen branches. This solution eliminates the need for storage arrays and even Microsoft Exchange servers.
"There's a lot fewer trips that IT staffs have to make out to these branches," Domrich says. Also, he notes, if a branch loses connectivity to the headquarters, the data is now cached locally, meaning it will be able to continue to conduct transactions.