HP: Deduplication And The Future Of Tape

HP this week unveiled its StoreOnce software, which automates data deduplication across converged infrastructures. The company also talked strategically about the place of storage technologies including tape.

No matter how far deduplication goes in using disks more efficiently, users will still need a reliable place to archive stationary data, so tape remains (see page 2). However data deduplication can greatly reduce storage space requirements, by eliminating redundant data from a disk storage device.

This in turn reduces data centre power and cooling costs and lessens the amount of carbon dioxide produced in generating power to run the hardware.

According to HP, traditional data deduplication is inefficient, as often data has to be expanded and contracted every time it is moved from one platform to another. At each level, the basic data is the same, but has to be converted to work with different algorithms, different management, and operate on a different platform.

“Very complicated, very expensive, a lot of different management paradigms, very inefficient data movement, but this is how it’s done today,” said Dave Roberson, senior vice president and general manager of HP’s StorageWorks division, speaking at the HP Tech Forum in Las Vegas. “You can see why this type of solution creates complexity and sprawl, because you have a lot of extra costs and expense as you’re moving, shrinking and expanding the data over and over again.”

StoreOnce improves on that, he said, and it will be available in all HP StorageWorks D2D backup systems – from the D2D2500, aimed at small business and enterprise remote offices, through to the new HP StorageWorks D2D4312, designed for mid-size data centres.

Deduplication 2.0

With StoreOnce users do not have to repeatedly expand and contract their data, said Roberson (left), because it’s built for converged infrastructure, and can be deployed from end to end. “Think of this a data deduplication 2.0,” he said.

The software can reduce the volume of data by a ratio of 50:1 and, HP claims, is 20 percent faster than its competitors. It also has a 2x price performance advantage, which can be achieved by leveraging the supply chain of HP, as well as the R&D and the scale of its server business. “You can start at 18TB, it grows to 48TB today with this architectures and we’ll grow it further as we expand the architecture out over time,” said Roberson.

According to HP, the primary place where people are buying deduplication at the moment is in the backup market. However, the company plans to extend the reach of its StoreOnce technology to include virtual machines.

Speaking to eWEEK Europe at HP’s Tech Forum, Lee Johns, the company’s director of marketing in unified storage, explained that storage is no different from any other workload.

“When you think about the technologies we were announcing on the server side, with the high-end servers now at 91:1 consolidation, storage is essentially an application that runs on industry standard processors in almost everybody’s storage systems, so why not run it in a virtual machine so you’ve got so much power and performance,” he said.

Johns also explained that HP is planning to put its StoreOnce software on its scalable back-end file-based storage platform – the X9000. “Deduplication is really not very effective if you’ve got 50 different storage solutions, because it doesn’t understand how to go across those systems. So the fact that we have a scalable backend with one name space means that we can deduplicate across multiple different physical boxes in a way that we believe others are really going to struggle to achieve, because they don’t have that scalable file storage,” he said.