By Kevin Davies
June 20, 2011 | Panasas unveiled the latest version of its ActiveStor storage product line this week at the International Supercomputing Conference in Germany. The ActiveStor 11 product features 3-terabyte (TB) enterprise drives without a price premium. The California company hopes it will prove an attractive offering for life sciences customers in general, and next-generation sequencing (NGS) applications in particular.
Panasas was founded ten years ago and is in its fourth generation of its flagship storage line. The company’s background is in technical computing, and has five years of consecutive revenue growth (42% in 2010) as it pushes into new markets. Traditional verticals include energy, finance/risk analysis, universities, and government/defense, with a strong presence in Europe and Asia as well as North America.
“Our customers tend to start by buying 1-2 shelves of our storage, and then become loyal customers over time and expand that footprint. That’s one of the attributes that comes with having such a scalable system,” says Geoffrey Noer, Panasas’ senior director of product marketing. Noer drops the ubiquitous term “big data” as do many other storage vendors, but Panasas has a solid pedigree in this arena. But it faces a challenge to gain market share in the booming life sciences market.
The introduction of ActiveStor 11, which nestles between the top-of-the-line 12 (launched last year) and the more affordable 8 products, should appeal to life sciences organizations. Existing clients include NIH, Yale University, BGI in China, and Uppsala University in Sweden. “We left a gap so we could introduce 11 after 12,” Noer explains, adding that he expects ActiveStor 11 to represent the bulk of sales going forward.
The ActiveStor 12 delivers 80 megabytes/sec per SATA drive. “Where performance is the top factor, this is the solution,” says Noer. The ActiveStor 11 is some 20% less expensive. “Some markets need more of a balance between performance and capacity, that’s where ActiveStor 11 is available at a more attractive cost. We’re confident that will help us in life sciences,” says Noer. It represents “the lowest dollar/TB option for all three models.”
Noer says that the ActiveStor architecture is common to all of Panasas’ blade storage products. “Director blades are the traffic cop in the system -- they tell the Linux cluster nodes who has permission to read and write. The storage blades have two drives apiece—for a full shelf, you have 11 blades, 10 of which are usually storage blades.”
Until now, Panasas products carried up to 40 TB/chassis. But that has now expanded to 60 TB/chassis with the use of 3-TB drives, which has a substantial impact on scaling. “We now scale to 6 petabytes in a single file system.”
“Performance has been in Panasas’ DNA from the very beginning. That hasn’t been as much of a core need for a competitor like Isilon in their prior markets,” says Noer.
“Institutions are deploying more and more NGS machines, with faster run times. Workloads are becoming about large-file throughput rather than millions of small files. So NGS is a perfect application for ActiveStor storage. You have a blade design that allows you to grow as needed – a single shelf can stand on its own, but it takes less than ten minutes to grow capacity as you need it to. And you can maintain a single global namespaces.”
Noer says several genome institutes are using Panasas for high-performance needs, but also using storage from Isilon (which he admits offers a much lower dollar/TB footprint) for their bulk capacity requirement. “It can make sense to have both installed,” he says.
Panasas is actively looking at the private cloud to further its momentum, a move applauded by IDC analyst Earl Joseph, who said: “Increasing numbers of HPC users will require highly scalable parallel storage systems that are dependable, easy to manage, and deliver the high throughput required for a wide range of technical computing applications. The Panasas ActiveStor 11 appliance is well positioned to capitalize on this important [HPC] trend.”
But at least one top bio-IT industry consultant, BioTeam’s Chris Dagdigian, blasted private clouds as “empty hype” at Bio-IT World Expo a couple of months ago.
“With all due respect to Chris [Dagdigian], I see the public cloud as being more of an overhyped approach than private clouds,” responds Noer. “Our products are the best suited for big data workloads, typically hundreds of terabytes or petabytes of storage. That data is valuable and highly proprietary. Trying to leverage the public cloud fails on several reasons – the cost of the bandwidth is out of sight, and you have a lot of security and performance concerns having the data remote . . . We don’t see a lot of traction for big data workloads in public clouds.”
Noer acknowledges that ‘private clouds’ is a new marketing name for what was previously labeled utility computing or grid computing. “But the trend has been taking place for many years, before the term ‘private cloud’ was invented. The desire to centralize is a very real direction and gaining momentum,” he says.
Noer also admits that Panasas has been criticized in the past that its price/TB was unapproachable. “Now, with the 50% bump in capacity with the 3-TB drives in addition to cost reductions we’re announcing and ActiveStor 11, all those things make Panasas a very attractive option.”
Panasas says it is taking orders now for delivery in August 2011.