If time is money, then First National Bank of Pennsylvania was spending way too much to back up critical applications. A full weekly backup would take 90 hours, and nightly incremental backups were taking around 24 hours. The bank, which has over 220 branches in PA and Ohio, was also bogged down with a tangle of products: 13 different backup systems were employed to service 570 servers. The company had to devote an IT staff person full time just to maintain all these backup products. When the bank decided to migrate from direct-attached storage to a SAN, it also decided to streamline its backup operations and embrace deduplication.
In the summer of 2009, the company purchased a pair of EMC Clariion CX4-240 SANs, one at the bank's production data center and the second at a warm site. Around the same time, it purchased CommVault's Simpana 8 deduplication software. Brian Diegan, vice president of network services for First National Bank, says the company investigated several deduplication products, including EMC Avamar, but at the time Avamar didn't support Novell GroupWise, a critical component of the bank's software infrastructure.
Diegan says he's pleased with Simpana. "We did a week's worth of testing, and now we dedupe everything. We dedupe to the production site first, and then to the warm site." Weekly backups have dropped from 90 hours to 24. Nightly incrementals also happen more quickly. "Nightly backups used to take 24 hours. Now they're down to two hours," says Diegan. Deduplication has also reduced the amount of backup data he has to store on disk, from 40TB down to about 6TB.
First National Bank also sees other benefits in deduplication. One is around file restoration. The bank keeps 30 days worth of deduplicated files on its production site, and restoration takes minutes. Before the deduplication software, Diegan says they backed up to tape, so restoring files could take six or seven hours. In addition, the IT staffer only has to spend part of his time on backups. "He's working on progressive projects instead of day-to-day maintenance," says Diegan.
With deduplication of critical backups in place, the company is turning to other storage-related projects, including tiering of data, in which files and information that are accessed frequently will be stored on high-performance (and expensive) disk, while other data can be moved to less expensive disk. The company will have a mix of flash, Fibre Channel and SATA drives in its SANs. "We're trying to nail down what gets in the tiers," says Diegan. "The easy part is that if someone hasn't touched a file in a certain amount of time, that goes to a lower tier." The company is also adopting virtualization: more than 30 servers have been virtualized so far, and Diegan's IT staff is investigating backing up virtual images.