Greg noted that everybody with large-scale storage needs has bought into the fact that GPU can help them accelerate their workloads.
A major advantage of GPU-accelerated storage is that it enables customers to scale their storage quickly and efficiently.
He noted, “GPU-accelerated storage is a good fit for companies that aren’t always able to forecast their storage needs but know that those needs can grow rapidly.
“For example, a company may support five or six researchers with relatively stable, predictable storage needs. But then one receives a large research grant, so the storage needs quickly escalate to handle a much larger demand for storage. Software-defined and GPU-accelerated storage can handle this quick escalation much better than a RAID solution,” he said.
Greg explained this is of particular interest to university researchers due to their budget constraints. They don’t want to use up parts of their budgets on flash or other expensive storage options. They want to balance as much computing power as possible with fast storage. They need to carefully consider the percentage of any grant used for computing compared to that used for storage.
Greg added that Silicon Mechanics customers see that Nyriad’s new UltraIO storage system provides the benefits of a GPU-based system. “It enables them to configure the storage system based on their needs to provide the optimum storage solution to meet their budget and provide the best balance of risk and rewards,” he concluded.
UltraIO™ is a trademark of Nyriad, Inc.