The new EX10000E nodes will store up to 10 TB of physical storage per 4U device, up from 5 TB of physical capacity for the 3U EX5000. ExaGrid also updated its EX Series software to support up to 10 nodes per grid, up from six in previous versions. A 10-node cluster would support 100 TB in a cluster.
ExaGrid customers need extra capacity because it does its deduplication post-process, but vice president of product management Marc Crespi said customers can also set the dedupe processing window according to how they make secondary backups. For example, a customer who wants to replicate data or write it out to tape to send offsite can set the ExaGrid system to wait until those secondary backups are made before it starts deduping.
Data deduplication can also be set to run concurrent with the initial backup to the ExaGrid appliance, as soon as 400 GB into the initial backup process (this is similar to how Quantum Corp.'s DX Series can be configured). Thanks to new Intel Corp. Nehalem processors and new support for more cluster nodes to parallelize performance, ExaGrid is now claiming the EX10000E can perform at up to 500 megabytes per second (MBps).
"With a fully configured ExaGrid system, users should expect about a 14-hour window for all data backup, replication and data deduplication to take place," Crespi said.
ExaGrid nodes also share information across the grid and can be managed from a single console, though customers would still load-balance which backup streams go to which node in the grid initially. From then on, the redundancy mapping tables ExaGrid uses to do its data deduplication are stored in each node of the grid and synched. The grid then "spills and fills," allowing multiple nodes to take up storage or processing of data should the demand exceed the original node's capabilities.
"ExaGrid is like a mini-Sepaton," said data backup expert W. Curtis Preston, referring to the DeltaStor product from the enterprise virtual tape library (VTL) vendor. DeltaStor also does delta-differential data deduplications rather than running hash-based comparisons, and is also configured in a grid. Both systems also do forward referencing, keeping the latest backup whole within the dedupe system rather than the initial backup. Global data deduplication has been something of a sticking point for data deduplication vendors on the market so far. ExaGrid's "spill and fill" approach may not be the textbook definition of global dedupe, but Preston said, "They have something that's close enough to it."
Picking up where Data Domain left off?
When it first hit the market, competitor Data Domain was focused on small- and medium-sized businesses (SMBs), but moved to scale its DD Series inline data deduplication boxes up into enterprise-level shops. Data Domain, now part of EMC Corp., lacks the ability to perform global dedupe or pool storage among multiple DD Series nodes, though the vendor said the capability is on the roadmap. ExaGrid is looking to poach former EMC and Data Domain customers and resellers during the acquisition transition by focusing on the midmarket and the channel.
ExaGrid vice president of marketing Bill Hobbib said the vendor has more than 450 customers and is growing in the channel. He said ExaGrid has more than 150 value-added resellers (VARs) signed up, including several dozen that have signed up in the last quarter.
Bernard Westwood, CFO of Atlanta, Ga.-based VAR Syscom Technologies located near Atlanta, Ga., had been a Data Domain reseller but switched to partnering with ExaGrid last year. That was before the EMC acquisition, but Westwood said he felt that Data Domain was moving more toward direct sales than selling through the channel. "I didn't really see them as a channel company," Westwood said of Data Domain. "They had a rapid growth rate and started hiring more direct sales people, and relying on channel partners less."
Westwood predicted that the release of a bigger system from ExaGrid would allow it to compete more effectively with Data Domain, and that other VARs might evaluate new data deduplication partners following the EMC-Data Domain merger. "There are some resellers in the Southeast and Atlanta area that may not have a good reunion with EMC," he said.
Some EMC shops are also choosing ExaGrid's products. Jason O'Dell, IT manager for Green Bank, one of EMC's largest customers in eastern Tennessee, said he bought a combination of ExaGrid and CommVault data backup software over Data Domain and EMC's backup software last October based on ease of use and price.
"We operate with a small, lean staff," he said. "We don't have bodies that larger organizations have to manage multiple software products for snapshots, backup and replication." The ExaGrid product was also about 30% cheaper than a Data Domain appliance when O'Dell was making his evaluation.
Still, ExaGrid will have a long way to go to match Data Domain's track record so far in the market. EMC forecasts $200 million in revenue from Data Domain products in the second half of this year.