As 2015 was in its home stretch, DataDirect Networks (DDN) refreshed its high performance SFA block storage line with the launch of SFA14K and SFA14KE, formerly codenamed “Wolfcreek.” DDN also took the wraps off its Infinite Memory Engine (IME14K), which leverages solid state and nonvolatile memory technologies to create a data caching tier between processor and parallel file system, effectively cutting latency and speeding applications.
In an interview with HPCwire, DDN CEO Alex Bouzari discusses how this product set can be used in a broad range of applications and use cases across domains. As we’ve covered previously, DDN has been ramping up its strategy to serve two distinct markets, traditional HPC, where it has long been a dominant player (the company supports about two-thirds of the top 100 systems), and the burgeoning advanced scale segment. Bouzari was one of the first HPC leaders to articulate a vision around the HPC-big data coherence. It’s a vision that is gaining real-world traction.
Having spread from its roots in academia and government, HPC is “permeating the fabric of our lives,” says Bouzari — creating new opportunities for tech companies and the markets they serve, such as medical science, financial services, manufacturing and more. Climate and weather research is also benefiting. Thanks to big data synergies, DDN’s SFA14K hyper-converged storage platform (in tandem with an SGI ICE XA supercomputer) will enable climatologists at the National Center for Atmospheric Research (NCAR) to more accurately model and predict long-range climate trends.
DDN’s approach centers on a scalable and flexible framework, one that Bouzari believes will enable exascale computing for DDN’s customers within a five-to-six year time frame. This goal will be achieved via “application acceleration with IME, using a framework which combines processors, interconnects, new advances in silicon and new advances in non-volatile memory technology in order to greatly accelerate the performance of an IT infrastructure — in essence beating Moore’s law in very significant ways.” The CEO posits that these same technologies will enable something akin to exascale for its emerging customer base: “As we set out to solve the problem of exascale for high-performance computing, we also ended up solving it for the enterprise and Internet world.”