These days storage is in the HPC spotlight. How well an HPC application performs relies not only on total system memory bandwidth and sustained floating point operations per second, but also on a storage architecture that supports sufficient throughput to handle constantly increasing amounts of data.
The key to storage performance is not capacity – it’s efficiency. Top performance relies on a fully integrated storage architecture supplied by a single vendor who is able to optimize from the drive to the file system.
Low-efficiency storage systems by their very nature require IT to add hardware and storage capacity just to attain target performance levels. This increased complexity adds costs, enlarges the physical storage footprint, and degrades the overall reliability of the system, limiting the availability and operational performance of the entire HPC infrastructure.
Searching for a Solution: Start with the Disk Drive
When investigating HPC storage solutions, ask the vendor if the proposed hardware is necessary to achieve your desired performance. Be on guard to ensure your vendor isn’t adding capacity just to hit rack-scale performance numbers, thus leaving a lot of wasted capacity on the table.
Seagate, with its 30-year history of designing and manufacturing drives, knows how to combine performance with efficiency. Only Seagate designs and manufactures the disk drives and the enclosures. It conducts proprietary testing on the system and/or rack-level solution, and uses its software expertise and management-layer capabilities to deliver an optimized end-do-end solution no other company can match.
The result: an HPC storage solution that can provide twice the performance and twice the efficiency of competitive offerings.
ClusterStor Solutions
Consider the Seagate ClusterStor family. These systems leverage Lustre, an open source high-performance parallel file system, to deliver balanced scale-out storage with the greatest performance per disk.
Disk efficiency translates into performance increases and reduced costs – fewer devices, fewer failures, reduced management, and improved TCO. The savings at a small scale are impressive; at a large scale the savings are dramatic: sometimes hundreds of thousands of dollars in operational costs alone.
In Conclusion
Seagate achieves unprecedented HPC storage efficiency – with an associated reduction in costs and boost in performance to handle the most difficult technical computing and big data problems – by starting at the fundamental level: the disk drive combined with a unique enclosure design. Anyone can throw hardware at a storage problem to scale up and meet performance objectives. But not everyone can do it efficiently.
Traveling to New Orleans for SC’14? Please click here to join us at the Seagate HPC User Group.