In a previous blog, I identified an apparent disconnect between HPC organizations’ storage buying criteria and the operational issues they later encounter. While HPC storage gets purchased with little consideration for subsequent operational issues, the results from the Hyperion Research survey demonstrate the devastating impact these same operational issues can have on productivity and total cost of ownership (TCO).
The results suggest several depressing scenarios. One is that HPC organizations have grudgingly traded off any expectation of reliability, easy management and good support in order to get high performance at a good price. Or perhaps they see these criteria as having little impact on their HPC operations. Alternatively, HPC organizations may simply not believe there is an alternative because they’ve concluded that all HPC storage products are complex and unreliable.
Fortunately, there is an HPC storage system that is performant and reliable: Panasas. In the accompanying pie chart, you’ll find the most challenging storage infrastructure operational aspects.
Most Challenging Storage Infrastructure Operational Challenges
Here’s how Panasas uniquely mitigates those issues with a low-touch appliance solution that features the fastest parallel file system at any price point:
- Recruiting/Training Storage Experts – Lustre, GPFS and BeeGFS file systems require dedicated storage expertise that is expensive to find and retain. The Panasas PanFS parallel file system is a low-touch solution that requires no deep technical skills for ongoing administration.
- Storage System Installation Time and Cost – According to the survey, 56% of installations took weeks to install. Panasas systems are typically installed in a day.
- Tuning and Optimization Time and Cost – Requiring little need for labor and skill-intensive tuning/retuning to maintain peak performance, Panasas PanFS balances and optimizes every part of the storage system — CPU, NIC, DRAM and storage media — ensuring consistent, predictable performance across varying workloads, regardless of complexity, and with no need for manual tuning or reconfiguration.
- Administration Time and Cost – 60% of sites require more than one dedicated person to manage their storage. Only one part-time administrator is typically required to manage a Panasas storage solution, regardless of scale.
- System Downtime – 42% of survey respondents reported weekly or monthly downtimes. Panasas installations have demonstrated up to 8 years of no reported unplanned downtime.
Panasas Has the Price/Performance Required for HPC Storage
With PanFS delivered on the new industry standard hardware-based ActiveStor Ultra appliance, Panasas has the price/performance chops to meet the most demanding HPC storage requirements1.
Comparative Read Performance of PanFS v. Competitive Parallel File Systems in GB/s
A PanFS-based Panasas system is going to be approximately twice as fast as similarly configured Lustre, GPFS and BeeGFS systems – and often even many times faster. So, there is no need for HPC organizations to trade off performance and price for simplicity and uptime – Panasas delivers across the board.
In the world of HPC storage, only Panasas delivers the required performance in a cost-effective, low-touch appliance solution that offers the simplicity, reliability and competent support associated with enterprise-class storage systems.
Only Panasas Checks all the Boxes for HPC Storage