EMC repackages Avamar data deduplication

EMC announces new packaging for its Avamar data deduplication software, including a virtual edition and two new storage hardware products OEMed from Dell.

EMC has announced two new storage hardware products built to integrate with Avamar Technologies' data deduplication software, as well as new packaging of Avamar's software for deployment on VMware virtual machines.

Avamar Virtual Edition is a bundle of Avamar software and Red Hat Linux OS that has been revised to recognise virtual disks and virtual configuration files. It is available on a hard drive, DVD or FTP site from EMC. Avamar Virtual Edition assumes that the user is running VMware ESX and already has shared storage deployed. It can dedupe both within and across virtual machines.

"Being able to run dedupe inside a virtual machine, in addition to deduping virtual machines [themselves], makes Avamar unique in the industry at this point," said Lauren Whitehouse, analyst with the Enterprise Strategy Group (ESG).

More on data deduplication
Sun to resell Diligent VTL

Data Domain's CEO takes on nearline storage

Quantum first with data deduplication flexibility

Users dish on Symantec PureDisk
The Virtual Edition for now is a "small version" with support for up to 1 terabyte (TB) of physical storage, which Avamar claims can hold up to 35 TB of traditional backup data. According to co-founder and EMC vice president of product management Jed Yueh, there are still performance barriers when it comes to hosting full versions of certain applications on virtual machines, a concept known as virtualisation drag. It is a problem also encountered by FalconStor Software, which left storage virtualisation out of its version of IPStor for virtual machines and limited the number of host connections its virtual appliance will support.

Unlike FalconStor, Avamar Virtual Edition is not certified with VMware as an appliance due to technicalities in licensing regulations. EMC's licensing agreement with Red Hat requires it to pay for each instance of the OS running beneath Avamar, which means it can not bundle a copy of the OS with Virtual Edition. As a result, it misses the requirements for official certification with VMware. This means, that unlike other VMware-certified virtual appliances, it won't be available as a free download from VMware's Web site.

Technically speaking, EMC claims there's little distinction in terms of deployment between the Virtual Edition with a bundled-in OS from EMC and the licensed download from VMware, according to Yueh. "There's no significant distinction between the two; it's a matter of shades of definition," said Yueh.

Avamar Data Store: New hardware package from EMC and Dell

Avamar, acquired by EMC last year for $165 million, has already been integrated with the company's Celerra NAS array, and further integrations with EMC's other existing hardware, as well as with its backup software are planned, according to an EMC spokesperson. However, rather than offering the ability to front Clariion or Symmetrix with Avamar, as some in the market have been expecting, this announcement also includes two completely new storage hardware offerings from EMC based on products from Dell.

"Over time, we'll see more consolidation of platforms," Yueh said. "But as things stand right now, there wasn't a logical fit with any of EMC's existing storage systems," since the new Avamar Data Store also includes servers.

The Avamar Data Store comes in two varieties, single node or multinode. The single node is a Dell PowerEdge 2950 server with 4 GB of RAM, 3 GHz processors and 1.8 TB of raw storage. The multinode Data Store is a preracked, preswitched, preconfigured system of two to 17 nodes, plus one spare node. Nodes are added to the multinode systems in two node configurations.

Because the multinode system also uses Avamar's RAIN for parallel access to storage and includes dedupe, Yueh acknowledged it's very similar to the HydraStor product from NEC Corp. of America, released in March.

As a "drop in" system, it's also more similar now to the prepackaged appliances sold by Avamar nemesis Data Domain Inc., Whitehouse pointed out. "I think this product is somewhat in response to pressure coming from target vendors, like Data Domain and ExaGrid, which have been able to drop their systems into customer environments, while the complexity issue is the biggest objection Avamar has to overcome."

The fact that the system includes so many moving parts -- servers, cables, switches, HBAs, storage nodes and all their attendant connections -- suggests how difficult it can be to deploy a software-only system that relies on complex hardware configurations and has unusual performance requirements because of the data deduplication feature.

"Our customers said, 'we shouldn't have to do all this assembly when EMC has the technical expertise and supply chain to ship a system that's preconfigured and ready to roll," Yueh said.

"It's also going to make it easier for EMC to support, making it a more repeatable configuration, rather than having to qualify, certify and support all the possible combinations of devices you can use with this," Whitehouse said.

Joe Martins, managing director of the Data Mobility Group, said he agreed that the "handpicked" hardware will ease a lot of headaches, but said he still hopes to see EMC's existing arrays, like Clariion, become available for use as data stores with Avamar, as well. "The impression I get is that this is the [final] Avamar-focused product," Martins said. "But some customers might want to repurpose hardware they already have."

Avamar customer references did not respond to requests for comment as of press time.

Pricing and availability

Avamar Virtual Edition for VMware is based on Avamar v3.7.1, which is licensed and priced based on deduplicated backup capacity. Pricing will start at $17,000 for 1 TB of deduplicated backup capacity (equivalent to approximately 35 TB to 50 TB of traditional backup capacity). The product will not be generally available until November. For the Avamar Data Store, list pricing starts at $30,000. The product is available now.

Read more on Business applications