LAS VEGAS – Although data protection and disaster recovery have been hot topics for administrators attending VM world, the sad truth is that in most organizations, backup and larger storage issues are still catching up to the server virtualization wave.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Until recently, according to Ron Oglesky, service director for virtualization at GlassHouse Technologies, storage has been treated as an afterthought to storage virtualization projects, although it's often the most expensive element of VMware rollouts. "Virtualization is a tremendous foundation for data protection reinvention, but few consider the storage ramifications up front," he said.
In many companies, management orders virtualization rollouts to cut costs without considering storage. As a result, Oglesky said, snapshot space, VMDK space and storage requirements all skyrocket. And then, he notes, "management says, 'Now let's talk about DR,' and meanwhile, backups haven't even changed yet."
Furthermore, 57% of the respondents said they also require two backups for single file and full image backups, and nearly half were not using data deduplication in a virtual environment. And 10% of those polled said they were not aware that data deduplication is an option within a virtual environment.
Storage admins shares virtualization lessons
At a VMworld session on Tuesday, storage administrators divulged the lessons they had learned from designing new backup and disaster recovery environments for virtual server data. Like many backup scenarios, these lessons often include reducing reliance on tape with data deduplication.
Daniel Lewis, manager of network services for the University of Southern California's Marshall School of Business, said that at first his organization stuck to agent-based tape backups of individual virtual machines after deploying VMware. Although Marshall pared its onsite retention time for backups to a week, its IT staff was running out of power, space and man-hours for changing tapes.
"There were still concerns over tape loss," Lewis said. "We had to get out of the tape management business, and SAN replication was too expensive."
Bringing in data deduplication appliances from Data Domain boosted onsite retention to six weeks, while shrinking disk backup space from a rack and a half to 3U, Lewis said. The business school has one Data Domain DD565 unit at the main campus and another at a disaster recovery site in Arizona. The primary box stores 50 TB of accumulated backups in 5 TB of disk space, while the Arizona site stores about 125 logical terabytes in 5 TB of actual space.
Lewis found the bandwidth throttling feature in Data Domain's replication software necessary for multisite replication with limited bandwidth. "I've tuned it as far down as 700 KBps," he said. "There was a couple hours' lag, but it got across." Lewis has already discontinued sending tapes offsite with Iron Mountain and is in the process of phasing out tape entirely.
Despite consolidating and simplifying their backup and disaster recovery hardware environments, Lewis and Gonzales said they use Vizioncore vRanger for snapshots on top of their base backup application.
Lewis runs Symantec Backup Exec 11d and is considering upgrading to 12.5, which can do single-pass snapshot backups of both system information and file-level user data. He's struggling between updating vRanger, which will increase his support costs, or updating Backup Exec, which he said has been "painful [to upgrade]."