Fusion-io Blasts Through 1 Billion IOPS Barrier

By Thor Olavsrud

Memory specialist Fusion-io Thursday announced that it blasted through the 1 billion input and output operations per second (IOPS) barrier during a demonstration of its latency-reducing Auto Commit Memory (ACM) extension. It achieved that performance using software to optimize the performance of existing hardware.

IOPS is a performance measurement used to benchmark storage devices. Hard disk drives (HDDs) tend to achieve IOPS in the low hundreds.

Fusion-io broke the 1 million IOPS barrier on a single rack in 2008 and managed to get it down to a single PCIe card in late 2010. Its latest breakthrough used eight HP ProLiant DL370 servers, each using eight ioDrive 2 Duos, to exceed 1 billion IOPS.

"Rethinking how to provide modern CPUs with the data they need through sophisticated software architectures has enabled us to deliver the ultra-low latency performance needed to achieve 1 billion IOPS with existing hardware and Fusion ioMemory solutions," said David Flynn, chairman and chief executive officer of Fusion-io. "This breakthrough is not something that could be achieved with hardware alone. Intelligent software that optimizes NAND flash as a low-latency, high-capacity, non-volatile memory solution for enterprise servers can transform the way organizations process the immense amounts of data that powers our lives today."

The demonstration, held at "DEMO Enterprise: An Evening of Innovation," was a preview of an extension of the ioMemory architecture the company calls ACM, a part of the Fusion ioMemory subsystem intended to reduce the latency and system overhead involved in transferring data. Fusion-io said ACM delivers data at peak performance levels while assuring data integrity through the architecture's ability to flush all in-flight data, even if the power is cut and without relying on capacitors or batteries.

Fusion-io said ACM is a new memory type that uses the underlying flash to present persistent memory directly to applications. It effectively blurs the distinction between storage and memory by creating a tier of direct access persistent memory. Using ACM, applications can declare a region of virtual address space as persistent, and can then program the region using regular memory semantics, such as pointer dereferences. It can also access the memory directly via CPU load and store operations that bypass the operating system. In doing so, it significantly reduces latency, according to the company.

"Breaking the 1 billion IOPS barrier is certainly a powerful way to demonstrate our ioMemory architecture," said Steve Wozniak, Fusion-io co-founder and chief scientist (and co-founder of Apple). "As an engineer, what really excites me about our extensions to our core technology such as ACM are the possibilities introduced when flash is utilized as a new memory tier. Instead of treating flash like storage, where data passes through all of the OS kernel subsystems that were built and optimized for traditional storage, our core ioMemory technology offers a platform with new programming primitives that can provide system and application developers direct access to non-volatile memory."

Thor Olavsrud is a contributor to InternetNews.com, the news service of the IT Business Edge Network, the network for technology professionals.

Follow InfoStor on Twitter

This article was originally published on January 06, 2012