The Academia Sinica Grid Computing Center (ASGC) hosts a Tier-1 center in Taiwan where performance is critical for day-to-day operations. ASGC demands low-latency reliable storage devices, yet has limited rack space to expand. By using Promise data storage, ASGC scaled up to approximately 2.3 petabytes (PB) of hard-disk storage in its data center. By tuning the computing grid, ASGC enabled the grid to respond dynamically, efficiently and responsively to user demands. Additionally, the center enhanced the overall grid system computing performance as well as data storage reliability.
The Academia Sinica Grid Computing Center (ASGC) hosts a Tier-1 center in Taiwan where performance is critical for day-to-day operations. ASGC demands low-latency reliable storage devices, yet has limited rack space to expand. By using Promise data storage, ASGC scaled up to approximately 2.3 petabytes (PB) of hard-disk storage in its data center. By tuning the computing grid, ASGC enabled the grid to respond dynamically, efficiently and responsively to user demands. Additionally, the center enhanced the overall grid system computing performance as well as data storage reliability.
Improved performance and data storage formed the solid foundation for the entire grid computing system. Promise RAID subsystem drives up to 2.3 GB/s sequential write and 5.0 GB/s sequential read throughput. In grid computing, the data request size varies from a few kilobytes up to multiple gigabytes. Promise storage can accommodate simultaneous requests for data in various sizes while offering performance. With redundant system, Promise storage offers no single point of failure and exceptional reliability.
"After using Promise storage, we were able to achieve significant savings in rack space, and save 18,100 kilowatt of power," said Shi-chang Lee, CEO of ASGC.
According to Dr. Lee, speaking for data center operation management, the important factor is rack space, followed by power consumption and cooling. By using Promise storage, ASGC was able to continuously increase the storage capacity within the limited rack space, as well as improve performance and reliability of the overall infrastructure, therefore increasing overall grid data center efficiency.
"In the past, four storage subsystems were connected to four blade servers. Now, one Promise storage subsystem can provide four times the storage capacity as well as the computing power. It only requires one blade server to connect to it, and the capacity adds up to 200TB per server. In other words, it achieved significant savings in rack space," Lee said. In the meantime, with fewer blade servers, fewer storage enclosures and energy saving design, ASGC not only reduces its initial acquisition cost, but also achieved savings on power and cooling by 18,100 kilowatt.