A3CUBE Shatters Storage Bottlenecks with PCIe-Inspired Networking Tech

Posted on February 25, 2014 By Pedro Hernandez

RssImageAltText

A3CUBE, a San Jose, Calif.-based tech startup, has taken the wraps off a new PCIe-based storage networking technology that promises to massively accelerate applications.

Called RONNIEE Express, the company's platform leverages PCIe to create a "brain inspired" network fabric that addresses the lingering bottlenecks between increasingly faster server processors and storage systems shackled to traditional networking paradigms. The technology is the result of a five-year research and development effort to extend PCIe's memory mapping features to create a latency-busting, in-memory network that eliminates the need for central switching, courtesy of 2D/3D torus and hypercube network topology support.

To narrow the widening I/O performance gap between compute and storage, the company enlists a combination of specialized PCIe network interface cards and software, namely the company's ByOS parallel storage operating system. RONNIEE Express then establishes a globally shared memory container across the entire network, accelerating high performance computing (HPC), Big Data and enterprise application workloads.

"Today’s data center architectures were never designed to handle the extreme I/O and data access demands of HPC, Hadoop and other Big Data applications," said A3CUBE's CTO and co-founder, Emilio Billi, in a statement. "The only way to accommodate the next generation of high performance data applications is with a radical new design that delivers disruptive performance gains to eradicate the network bottlenecks and unlock true application potential," he added.

By eliminating protocol stack overhead and implementing a direct memory-to-memory mapped socket that unites RONNIE Express nodes, data courses through the network practically unimpeded, according to the company. In tests, RONNIE Express scored nearly 400,000 messages per second in the Sockperf benchmark, well ahead of a 10 GBe cards from Chelsio and Intel with over 100,000 and 50,000 messages per second, respectively.

When implemented in a 5-node, storage network using a multipoint direct topology and 40 TB of low-cost SSDs, A3CUBE estimates that its platform can deliver over 4 million IOPS, besting competitive configurations from NetApp and IBM, which top out at just over 500,000 IOPS.

Moreover, any application can benefit from RONNIEE and the company's ByOS software, said Billi. Since the technology is based on the tried and true PCIe standard, "it is completely transparent to the application," he told InfoStor.

Further efficiencies are gained by the added performance A3CUBE was able to squeeze out of PCIe. "Our memory mapping is 10 times faster than [traditional PCIe] memory mapping," he boasted. Billi also hinted at massive cost savings by deploying the technology on commodity servers. Compared to storage infrastructures from top-tier vendors, a "system with eight times the performance is less than one-third the cost," he said. "The most expensive thing is the solid state drive."

In prepared remarks ESG senior analyst Bob Laliberte said, "A3CUBE's In-memory Network fabric leverages an innovative approach to transforming HPC, Big Data and data center environments in order to drive greater performance and efficiencies in the network and storage systems."

Pedro Hernandez is a contributing editor at InfoStor. Follow him on Twitter @ecoINSITE.

Photo courtesy of Shutterstock.


Comment and Contribute
(Maximum characters: 1200). You have
characters left.

InfoStor Article Categories:

SAN - Storage Area Network   Disk Arrays
NAS - Network Attached Storage   Storage Blogs
Storage Management   Archived Issues
Backup and Recovery   Data Storage Archives