What’s Flash’s Role in Analytics? Watson Has a Leading Part

Flash is enabling a number of new things in the data center, even bringing transactions and analytics closer together.  Colin Parris, the general manager of IBM’s Power Systems Unit,  shared his unique perspective on solid-state storage and Big Data in a recent interview with Wikibon co-founder and chief analyst Dave Vellante, at an IBM Flash event held earlier this week.

After providing a few details about the Power 7+ servers that IBM introduced in October, Parris explains the reasons behind the rise of flash in the enterprise. He says the technology can support high-bandwidth workloads with minimal latency, an advantage that makes it far superior to traditional disk when processing Big Data.

Intelligence agencies and other data-driven organizations are looking for solutions that can handle both the data itself and the CPU-intensive algorithms that are applied to it, which is where IBM’s latest generation Power Systems come in. Parris boasts that Power 7+ leverages flash to deliver 60 to 90 percent utilization, above and beyond that 20-40 percent you might expect from the typical Intel box.

Vellante asks Parris if he sees transactions and analytics coming together in the future. He responds by saying that this is already happening.

“The entire thing is to get the data as close as possible to the transactions, so you see it already in this in-memory view of the world. Now the interesting thing about Power is that many people know it because of Watson…what most people don’t know is that was the largest in-memory database in history.  Because what we had to do is we had to ensure all the data used in the Jeopardy competition was on the Watson systems; it could not in any way connect to the internet. [It was the] largest in-memory database pulling the data closer because we needed that response time, we needed that close speed. Flash begins to get you even closer.”

Just like transactions and Big Data, storage, software and other components are also coming together in response to market demand.  Parris explains that the lines are blurring because it’s no longer possible to deploy cloud, analytics and other mission-critical functionality in isolated silo – it all has to be done in the same place.

See Parris’ entire segment below: