HPCwire

The Leading Source for Global News and Information Covering the Ecosystem of High Productivity Computing

HPCwire >> Features

In the Midst of a Recession, Panasas Comes Out Swinging


Most storage vendors were nervous about what lay ahead in 2009, and Panasas was no exception. In 2008, spending on IT infrastructure dipped as the recession forced many companies to scale back on capital expenditures. The good news for Panasas is that the customer base for the company's HPC storage products is heavily skewed toward sectors that rely on high-tech infrastructure to fulfill their mission -- aerospace, energy, manufacturing, semiconductors, life sciences, financial services, and government supercomputing. These customers are continuing to pour money into research and development. A recent report by the Wall Street Journal found that the most innovation-driven firms were holding the line on R&D spending, even as they slashed payroll and overall capital expenditures.

As a result, Panasas managed to grow revenues in the first quarter of 2009, and even expanded its customer base 25 percent year over year. Panasas was also fortunate in completing a $25 million round of funding last year before the recession put a damper on venture capital. According to Panasas VP of Marketing Larry Jones, the company is running "cash flow neutral" right now, and with money in the bank, they feel confident they will weather the current economic downturn.

With that as a backdrop, today Panasas introduced a revamped portfolio of its HPC storage products: the ActiveStor Series 7, 8 and 9. The three new products continue the Panasas model of offering high performance storage combined with enterprise-like availability and manageability. They're not likely to be the cheapest HPC storage out there, but the company believes the HPC market, especially in the commercial segment, is looking for ease-of-use on top of high performance. "That's kind of where we see the sweet spot for Panasas," explained Jones.

The new offerings replace the older ActiveStor products and span three price-performance levels: Series 7 is positioned as entry level HPC storage for decent performance at low cost per GB; Series 8 is the heart of the product line offering high performance and manageability; and Series 9 represents the highest performance, offering both high IOPS and low latency. The chart below summarizes the specs for each product line on a per storage module basis. For a rack, multiply the performance and capacity numbers by 10.
Panasas lineup
The Series 9 product is the company's top-of-the-line offering, and introduces a three-tier storage blade architecture. At the outermost level is the SATA tier, which consists of a 2 TB 7200 RPM SATA drive and is no different from the hardware in the Series 7 and 8 products. The middle tier is where the fun begins. It consists of a 32 GB SLC Solid State Disk (SSD) and is used to accelerate I/O by holding small files and metadata. The average latency on the SSD tier is 100 microseconds -- 100 times faster than the SATA drive. At the innermost level is 4 GB of RAM, which is used to aggregate writes for the SSD, store key metadata, and perform aggressive read-ahead. Latency for RAM is on the order of 100 nanoseconds.

The storage tiers are managed by the ActiveScale software, so internal I/O is transparent to end users and administrators. By integrating the different storage technologies into a blade, Panasas was able to achieve as much as a 10-fold performance increase over SATA alone. In particular, thanks to the SSD device, the Series 9 hardware can deliver nearly twice the I/O operations per second (IOPS) compared to Series 8, which lacks the flash technology.

The IOPS metric usually reflects how well a storage system will perform random I/O on small chunks of data -- an important capability for many HPC applications. Pumping random read/write performance takes a more sophisticated approach that just adding raw bandwidth for streaming I/O. Isilon recently introduced its S-series storage to go after more IOPS, but it did it more traditionally: with high performance SAS drives and lots of memory cache (16GB per module) instead of SSD flash.

"We looked at that approach -- in fact we benchmarked that approach," said Jones. According to him, they determined that the three-tiered design had two big advantages: it was faster and it was much less expensive. Even though the SSD flash is about 100 times as expensive as SATA storage on a per byte basis, by using a relatively small amount, the cost of the whole system will only be about 40 percent more than a SATA-only solution.

Panasas is betting that its Series 9 offering will open a few doors for the company in areas where random I/O is paramount. Credit risk analysis, genomics, EDA, satellite imaging, animation, and oil and gas all have applications that depend upon large numbers of relatively small files. For data-intensive apps, overall productivity often hinges on how fast the I/O performs, so delivering on IOPS can offer a lot of value to customers.

All of the new storage blades are cross-compatible with each other and even older Panasas gear. They can be mixed into Panasas storage racks according to price-performance preferences and deployment strategies. List prices range from about $2,000 to $3,600 per terabyte for Series 7 and 8, respectively. Both products are available immediately, either through Panasas or its resellers. Series 9 is expected to be available in Q3, and because SSD pricing is in a state of flux, final pricing will be determined at launch time.

Complete software support of all three products will be folded into ActiveScale 3.4. The new version adds a disk migration utility that can transparently move data from older storage blades to new blades. This takes place in the background, so there's no down time for production work. In addition, an asynchronous replication capability was included to improve disaster recovery, backup consolidation and content distribution. Panasas also slipped in some performance tweaks as well as a number of other features. ActiveScale 3.4 is expected to be released in Q3.


HPCwire on Twitter

Article Tools

  • Print This Page
  • Bookmark This Article

Share Options

(Digg, Technorati, more)


Subscribe

Discussion

There are 0 discussion items posted.  

HPC in the Cloud Part 2
People to Watch 2010


Top Headlines

IBM, Microsoft Help Create Montana Supercomputer

Feb 08 | eWeek | A new kind of Rocky Mountain high. Read more...

AMD Aims for GPUs in Mainstream Servers Starting 2012

Feb 08 | Computerworld | Chip maker hopes to bring CPU-GPU processors to servers in two years. Read more...

Graphene Transistors That Work at Blistering Speeds

Feb 05 | Technology Review | IBM has created graphene transistors that leave silicon ones in the dust. Read more...

Intel Sneak Peeks Westmere EP Server Silicon

Feb 04 | The Register | Intel will preview first 32nm Xeon chips on Monday. Read more...

Cheap Stuff: Trends in Commodity HPC

Feb 03 | Linux Magazine | A couple of relatively new commodity solutions could make a huge impact in HPC. Read more...

Featured Whitepapers

Virtualization for Aggregation And The vSMP Architecture™

Jan 12 | | In-depth look at vSMP Foundation server virtualization technology, technical implementation, use cases and capabilities. The technical whitepaper provides an architectural overview and details on the three vSMP Foundation products: vSMP Foundation for SMP, vSMP Foundation for Cluster and vSMP Foundation for Cloud.

Copper Cable Technologies for High Performance Computing

Jan 18 | | This white paper discusses Gore’s copper cable assemblies, and how they continue to exceed the standards for providing reliable, cost-effective solutions for high-performance computer applications.

Appro Assists LLNL with Cluster Designed for Extreme Scale Visualization

Jan 11 | | LLNL is home to some of the fastest computers in the world. In 2012, LLNL expects to have the Sequoia supercomputing cluster operational with a projected performance of over 20 PFLOP/s. These systems will focus on strengthening the foundations of predictive simulation through running large suites of complex simulations and then comparing model predictions with experimental data. To visualize this project’s large amount of data, LLNL requested an Appro Supercomputing Cluster specifically designed to support interactive data analysis.

Multimedia

Webcast: Virtualized Data Center Roundtable

Join this online panel discussion for live Q&A with leading industry experts, analysts, and end-users to discuss the latest innovations, best practices, barriers to implementation, and measurable benefits of server virtualization with a particular focus on today's real world solutions.

Webcast: Watch SC09 Birds of a Feather Video: Scalable Fault-Tolerant HPC Supercomputers

Learn about scalable fault-tolerant architectures and examples of energy efficient and scalable supercomputing clusters using dual QDR InfiniBand to combine capacity computing with network failover capabilities with the help of programming languages such as MPI and a robust Linux cluster management package.

Webcast: High Performance Computing for a Smarter Planet

LIVE@SCO9: The IBM team discusses new innovations in hardware, software and services that help clients better understand their workloads and get insight from their R&D efforts. Technology demonstrations include the soon-to-be-released Power7 HPC processor, the DCS990 system with 2.4 petabytes of storage, the xCAT management tool, secure HPC cloud computing and more. Winners of two HPCwire Readers' and Editors’ Choice Awards! Take the IBM virtual tour at SC09 or more information go online to: http://www-03.ibm.com/systems/deepcomputing/sc09.html

SC09 HPC in the Cloud

Newsletters

Stay informed! Subscribe to HPCwire email Newsletters.






HPC Job Bank


Featured Events

BrightTALK
HPCC
HPC User Forum DICE
Cloud Slam
Cloud Computing Expo
DEISA PRACE Symposium