De-Dupe Deep Dive

Vendors are cranking up their efforts around the emerging de-duplication market, with Data Domain expanding its product range today and EMC weighing up its de-dupe options. (See Data Domain Unveils DD580 and EMC Picks Up Avamar.)

De-dupe is gaining momentum amongst users looking to reduce the amount of data they back up. (See Users Look Ahead to 2007.) The idea behind the technology is that it compresses data that appears in more than one place, with vendors promising all sorts of new performance and capacity gains. (See De-Dupe Vendors Shake Hands, Quantum Delves Deeper Into De-Dupe, and Analysis: Data De-Duping.)

Data Domain, which recently joined the growing list of storage vendors moving towards IPO, today unveiled its disk-based DD580 device. (See Data Domain Files S1 and Storage Bubble Wrap.) With a throughput of 800 Gbytes per hour, the DD580 offers double the speed of its predecessor, the DD560, according to Brian Biles, Data Domain's vice president of product marketing. "Our previous version had two single-core [Intel Xeon] chips," he says, explaining that the DD580 uses dual-core Intel Woodcrest processors.

The vendor has also expanded the storage capacity of the DD580 compared to its predecessor, from 23.5 TBytes to 31.5 Tbytes. But the capacity hike comes at a premium; pricing for the DD580, which is available now, starts at $120,000 compared to $95,000 for the DD560.

Despite the price differential, at least one user is planning an upgrade. "It's fast [and] easy-to-use -- there is very little that I have to do with it," says Eric Eckman, IT manager at Santa Clara, Calif.-based manufacturing firm Atheros Communications, which already has four DD560s and is currently beta testing the DD580.Long term, Eckman's plan is to use the DD580 for disaster recovery between Atheros's headquarters and a disaster recovery site in Irvine, Calif. "If this building ever melts, all the data will be down there -- it's the peace of mind of knowing that the data can be quickly recovered."

Data Domain, like rivals Diligent and Quantum, de-duplicates data on the way to the backup target. In Data Domain's case, this is the DD580 itself, whereas Quantum and Diligent traditionally used VTL devices. (See Data Domain Adds VTL Option and Air Force Chooses Diligent.)

This is a different approach to that taken by vendors like Symantec, Asigra, and EMC, the latter of which acquired de-dupe specialist Avamar for $165 million late last year. (See EMC Acquires Avamar and Put Away the Check Book, Joe.) These vendors handle compression on the server before the backup. (See Data Domain Adds VTL Option and Quantum Leaps Into De-Duplication.) This is supposed to reduce the amount of data backed up, so less data has to be moved across the network, although the approach does consume server CPU cycles.

Another vendor recently entering the de-dupe fray is Sepaton, which added data de-duplication to its wares with the release of its DeltaStor Appliance in February 2007. (See Sepaton Adds De-Dupe to VTL.) Unlike Data Domain, Sepaton de-dupes data after it is written to a target disk.EMC weighs options

At this point, EMC does not offer target-based de-duplication on VTLs, although the vendor is now considering this possibility, according to spokesperson Todd Cadley. This could tie in with legacy backup software products such as Tivoli Storage Manager, Symantec's NetBackup, and EMC's own NetWorker. "Backup is a really sticky technology - it's not something that you can rip and replace easily," Cadley says.

Specific details of EMC's plans in this area are still hard to come by, and it's unclear to what extent the vendor will rely on Avamar technology or parts of other backup acquisitions such as Kashya. (See EMC Coughs Up for Kashya and EMC Takes CDP Downmarket .)Offering de-dupe on both target and source devices could potentially reduce firms' bandwidth requirements, although Atheros's Eckman tells Byte and Switch that he will not be leading the charge to deploy this type of technology. "Eventually, in a couple of years, that might be something interesting, but today, I wouldn't look at it," says the exec. "I don't want to be bleeding edge on the backup side of things."

— James Rogers, Senior Editor Byte and Switch