SALT LAKE CITY — April 6, 2009 — Fusion-io, the leader in application-centric, solid-state architecture and high-performance I/O solutions, working with HP, the world's largest technology company, today announced that they exceeded an astonishing 1 million IOPS (I/O Operations Per Second) and eight gigabytes per second (GB/s) sustained throughput using a single HP ProLiant server.
Working together in HP's ProLiant engineering labs in Houston, technologists from HP and Fusion-io built a system using five 320GB ioDrive Duos and six 160GB ioDrives in a single HP ProLiant DL785 G5 server, running with four Quad-Core AMD Opteron™ processors. This standard configuration allowed the engineers to reach an unprecedented eight GB/s sustained throughput, making it possible to achieve 1,009,384 IOPS using 2KB random 70/30 read/write mix, as measured using the fio benchmark.
"These results show the true power of combining our PCI Express and NAND flash technology with HP's ProLiant architecture," said David Flynn, CTO of Fusion-io. "The ioDrive and ioDrive Duo are able to supply the extreme storage performance that data centers have only dreamed of, at a fraction of the power, cooling, and per unit-of-processing-power price compared to traditional solutions. This is especially valuable for accelerating I/O intensive applications and workloads such as database and data mining, virtual machine deployments, and financial transactions."
"Customers rely on HP to reduce their total technology infrastructure costs while also improving business results," said Lorraine Bartlett, director of server marketing, Business Critical Systems, HP. "Combining HP's leading ProLiant servers and Fusion-io's technology offers customers an ideal solution for significantly improving application performance and lowering operating costs."
"Quad-Core AMD Opteron processors help drive efficiencies and reduce complexities with innovations that enable superior performance," said John Fruehe, director of business development, Server Workstation, AMD. "The capabilities of Quad-Core AMD Opteron™ processors combined with the high I/O bandwidth of HyperTransport™ technology compliment the outstanding performance and power efficiencies enabled by Fusion-io."
Fusion-io's solid-state storage technologies use NAND flash in a PCI Express form factor, which places storage as close to the application as possible and provides benefits over disk-based storage solutions for performance-intensive applications in the enterprise. It allows IT organizations to gain storage performance improvements on the order of 1,000 times, while consuming less than 1/100th of the power, and at as much as 1/10th the cost of traditional storage area networks.
HP provides its BladeSystem customers this same performance capability with their new HP StorageWorks IO Accelerator, a NAND flash-based storage adapter based on Fusion's ioMemory technology. Exclusively designed and built to complement the industry leading HP BladeSystem, each IO Accelerator card combines enhanced reliability with uncompromising data integrity to achieve more than 100,000 IOPS (4KB random read or write) and 800MB/sec bandwidth (read or write), with access latencies of less than 50µs and requiring as few as six watts. A single HP BladeSystem server can accommodate 2 or 3 IO Accelerator cards.
To find out more about how Fusion-io's enterprise solid-state storage products can benefit your organization, please visit www.fusionio.com.
Please visit: www.hp.com to learn more about HP. For more about the DL785 G5 server, please visit: www.hp.com/servers/dl785. Details about HP StorageWorks IO Accelerator can be found at: www.hp.com/go/io-accelerator.
Details about the fio benchmark can be found at: http://freshmeat.net/projects/fio.
Fusion-io is a leading provider of enterprise solid-state technology and high-performance I/O solutions. The company's solid-state storage technology closes the gap between processing power and storage needs delivering breakthrough performance at a fraction of the cost of traditional disk-based storage systems. The result is a world of possibilities for performance-starved applications.