EMC Offers Avamar De-Dupe as Virtual Machine

Casting its weight behind the latest data protection trends, EMC has unveiled a version of its de-duplication product that runs as a virtual machine within VMware. The vendor has also introduced a turnkey version of Avamar's software running on EMC hardware. (See EMC Intros VMware-Based Avamar.)

First, the virtual machine: Dubbed the EMC Avamar Virtual Edition for VMware Infrastructure, this version of Avamar's software resides on an ESX server, to which it appears as another virtual machine. The product won't be available until November, but it's priced at $17,000 per 1 Tbyte of de-duplicated data. EMC claims data reduction ratios of 35:1 to 50:1, so that 1 Tybte represents 35 Tbytes to 50 Tbytes of traditional backup capacity, in EMC's view.

EMC's new approach with Avamar is in keeping with a trend several data protection vendors have followed lately. Earlier this week, for instance, FalconStor launched a CDP virtual appliance for VMware. And Arkeia has built a VMware virtual appliance for its Network Backup software. (See Arkeia, VMware Partner.)

One thing: Avamar Virtual Edition is not officially a VMware virtual appliance. According to Jed Yueh, VP of product management for EMC's Avamar software business, that's intentional. To register as a virtual appliance in VMware's program, a product must not contain any software licensed from third parties, he says. Virtual Edition packs the Red Hat operating system.

At least one analyst says the distinction is moot. "For all practical purposes, [Avamar Virtual Edition] is a virtual appliance," says analyst Lauren Whitehouse of the Enterprise Strategy Group. And she expects to see a lot more data protection solutions appear on VMware, whether as appliances or simply applications. "It's a very popular packaging and deployment option," she says.EMC claims that running as a virtual machine decreases the hardware cost of de-duplication. It also serves to reduce the sheer bulk of data that occurs when virtual machines are used for backup, replication, and disaster recovery. "Traditional backup has OCD," quips Yueh.

Avamar Virtual Edition interacts with physical machines as well as with other virtual machines. EMC already offers virtual machine versions of its database and operating system host agents, which can work with Virtual Edition as a target.

As with other Avamar software, Virtual Edition uses integral backup software to send the de-duplicated data to an external Avamar appliance or to Virtual Edition, where de-duplication is repeated.

Avamar's technique of de-duplicating data before it's sent to backup puts it in the "in line" category of de-duplication product, along with Data Domain, Diligent, and Quantum. In some quarters, the in-line approach is seen as leading to faster RTO than a post-processing approach that de-duplicates data after it is backed up. (See Experts Share De-Dupe Insights and De-Duplication Rumors Highlight Controversy.)

Virtual Edition packs Avamar's own backup software, though it can interact with VMware Consolidated Backup. For instance, data from virtual machines can be sent from Avamar Virtual Edition to VMware Consolidated Backup acting as a proxy server, then forwarded to a centralized Avamar appliance -- or to a new Avamar Data Store.EMC is also offering a one- and two-node version of Avamar's de-duplication, backup, and recovery solution in a turnkey version packaged with Dell servers. This product is shipping now and starts at $30,000 for one node, which again covers 1 Tbyte of de-duplicated data.

EMC says Data Store will make life easier for customers who've wanted a turnkey solution. "We have over 700 customers," says Yueh, a co-founder of Avamar before its acquisition by EMC in November 2006. "Most want an entire solution consisting of server, software, and storage... This simplifies setup, service, and support."

One analyst thinks Data Store has one potential downside. "The implication is that you would replace what you've got," says Joe Martens of the Data Mobility Group. While Avamar's software is available for other platforms, there's a chance it's optimized for EMC's hardware, he says. "Data Store suggests there's room for a combination [of platforms], but that remains to be seen."