Visit additional Tabor Communication Publications
June 14, 2012
Today storage maker Xyratex introduced ClusterStor 6000, a Lustre-based storage platform which doubles the throughput of the company’s first generation product. HPCWire spoke with Eric Lomascolo, director of solutions marketing and Mike Stolz, VP of marketing at Xyratex to get the particulars about the new system.
At last year’s International Supercomputing Conference, Xyratex introduced the ClusterStor 3000, a highly scalable storage system and aimed squarely at the Lustre-friendly HPC market. At its core, ClusterStor was designed to provide higher density and throughput per square foot compared to traditional storage offerings. The technology was acquired in 2010 when Xyratex bought ClusterStor, which at the time was an independent storage maker specializing in Lustre-based systems.
Using an appliance approach, the hardware, operating system, file system and management software reside under one roof. The system is brought together in Scalable Storage Unit (SSU) building blocks. “Everything is embedded in the scalable storage unit in terms of the object storage server for Lustre,” explained Lomascolo.
Each storage unit is 5RU and contains 84 3.5” drives. 80 drives provide storage with two global hot spares and a pair of 100GB SSDs. Two embedded server modules enable data throughput. The SSUs are currently offered with 3TB SAS drives, amounting to 252TB of raw storage with 192TB usable. In the not-too-distant future, SSUs based on 4TB drives will ship, and they will support 250TB of usable storage from 336TB of raw capacity.
The use of the higher reliability SAS drives (versus the less expensive SATA hardware) gives the platform a more enterprise-like profile than the typical storage solution in the HPC space. “Xyratex is rooted in reliability, so we felt that having dual port and a reliable drive factor was going to be a key in this system,” said Stolz. “By using SAS-based technology in terms of the disk drives, we get that extra enterprise quality associated with that.”
Lomascolo expanded on that, noting the extensive QA processes carried out on each unit before it’s shipped. Prior to SSU installation, all disks undergo accelerated life testing. After the units are put together, they go through multi-week load testing. “At the end of the day, DOAs on large installations can have a great impact on how quickly they can get up and running,” he explained.
Initial deployment of the ClusterStor system starts with a Cluster Management Unit, which contains the Lustre metadata server. It is seen as the building block, which enables storage and throughput scalability. Once the management unit is setup, additional SSUs can increase capacity and performance. A base, single-rack system contains the management unit and seven SSUs.
The main difference between the 3000 and 6000 is the SSU internals. The 6000 platform still uses the same drive chassis as the 3000 series, but their modules have been upgraded with faster compute, more memory, FDR InfiniBand and an updated application framework. The result is that each SSU has twice the throughput (6GB/second) of its 3000 series sibling.
File system performance also scales relative to the quantity of SSUs deployed. This is the primary advantage of the 6000 series. HPC users that required 360GB/second of throughput would have to install 120 ClusterStor 3000 SSUs, taking up roughly 17 racks of floor space. With the 6000 series, that same throughput can be achieved in half the footprint with 60 SSUs. Xyratex says a 6000 setup can deliver up to 1TB/second of throughput, requiring 165 to 170 SSUs.
The ClusterStor 3000 series will remain in production and support the same storage capacities as 6000 series SSUs. This gives users the option to prioritize their investment towards storage capacity or throughput. Furthermore, current 3000 series SSUs can be updated with 6000 series server modules, enabling upgrades without disrupting chassis and storage configurations.
“[Customers] can move up to FDR InfiniBand and 40GbE, moving from 2.5 to 3GB/second up to 6 GB/second per SSU. All of the investment in the chassis and the storage remains,” said Lomascolo
At the moment, Xyratex’ only OEM partner for ClusterStor is Cray, who sells it under its Sonexion-branded storage line. But according to Lomascolo, more OEM partnerships are in the works.
One notable sale of the platform is at the National Center for Supercomputing Applications (NCSA), where a ClusterStor 3000 system is currently providing storage for the Blue Waters supercomputer. The build out is not complete, but as of early March, the Sonexion-branded system was providing two petabytes of storage for the not-yet-finished supercomputer.
There’s a lot more to come. NCSA’s spec sheet for Blue Waters calls for more than 25 petabytes of storage, with throughput surpassing the 1TB/second mark. Given the ClusterStor 6000 series’ performance, the center may opt for the upgraded server modules.
Pricing for the systems are ultimately left up to the OEMs, which for now is Cray, but Xyratex said ballpark configurations for a base ClusterStor 6000 system rack is approximately $300,000. Base ClusterStor 3000 systems are approximately $250,000. This includes a single rack with a cluster management unit, seven scalable storage units and an InfiniBand switch.
Systems will be available to ship in the third quarter of 2012.
Jun 19, 2013 |
Supercomputer architectures have evolved considerably over the last 20 years, particularly in the number of processors that are linked together. One aspect of HPC architecture that hasn't changed is the MPI programming model.
Jun 18, 2013 |
The world's largest supercomputers, like Tianhe-2, are great at traditional, compute-intensive HPC workloads, such as simulating atomic decay or modeling tornados. But data-intensive applications--such as mining big data sets for connections--is a different sort of workload, and runs best on a different sort of computer.
Jun 18, 2013 |
Researchers are finding innovative uses for Gordon, the 285 teraflop supercomputer housed at the San Diego Supercomputer Center (SDSC) that has a unique Flash-based storage system. Since going online, researchers have put the incredibly fast I/O to use on a wide variety of workloads, ranging from chemistry to political science.
Jun 17, 2013 |
The advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
Jun 14, 2013 |
For all the progress we've made in IT over the last 50 years, there's one area of life that has steadfastly eluded the grasp of computers: understanding human language. Now, researchers at the Texas Advanced Computing Center (TACC) are utilizing a Hadoop cluster on its Longhorn supercomputer to move the state of the art of language processing a little bit further.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?
Join our webinar to learn how IT managers can migrate to a more resilient, flexible and scalable solution that grows with the data center. Mellanox VMS is future-proof, efficient and brings significant CAPEX and OPEX savings. The VMS is available today.