The ongoing spread of artificial intelligence and machine learning are among the major drivers pushing the explosive growth of data across the globe.  As organizations generate and consume ever-growing amounts of data, storage becomes an overriding issue from the C-suite all the way through the enterprise.  

While the reasons for capturing and using data vary from company to company, most would say that a primary reason is to derive insights, through advanced analytics, in order to better run and manage their organizations. After all, if your company sells something, has a supply chain, or wants to innovate products or processes, you’re more than likely already using big data analytics to push forward.

Obviously, storage of all of this data is becoming increasingly important. If you can’t save it, it can’t be used. Traditional on-premise storage is now splitting mindshare with cloud-based workloads. Computing power is also moving to the cloud, as companies are beginning to recognize that it’s oftentimes more cost effective for cloud service providers to manage those resources as well other associated expenses (such as power, hardware upgrades, up- and downtime, etc.).

THE PROMISE OF QLC

At the recent Micron Insight 2018 event in San Francisco, corporate vice president and general manager of Micron’s Storage Business Unit Derek Dicker discussed how his company is enabling enterprises to manage all that data. Micron is one of the world’s largest storage companies, with deep roots in DRAM—dynamic random access memory chips. Those chips provide memory to all kinds of devices, including laptops, servers, and more.  

Derek Dicker Micron

Derek Dicker of Micron

One of the more intriguing technologies Dicker talked about is QLC, or quad-level cell NAND technology. In QLC, 4 bits of information are stored per memory element, making this technology efficient from both space and power consumption perspectives. And, according to Dicker, this allows for the possibility of data storage to be physically close to the compute capacity, where it’s acted on. 

“To be able to architect systems where you can have processing take place closer to the media is a very interesting area,” Dicker remarked. “It’s ripe with a ton of research going on right now.”

CHALLENGING HARD DISK DRIVES

By crunching that travel time from the storage to a computer, data-intensive processes can be speeded up.  And, QLC is being applied to solid state drives (SSDs) so there are efficiencies to be gained over spinning hard drive technologies.  

HDD spinning disk

As Dicker pointed out, QLCs are very well tuned to ‘read-intensive’ environments. Data is normally generated once, and all the new analytic technologies, like deep learning, AI, and machine learning then read that data regularly to perform calculations and analysis, sometimes as many as a million reads of single data points.  “The drive writes per day that are required in a machine learning infrastructure, we believe we can address with QLC”, added Dicker.

Experts insist on the role QLC can play in accelerating the uptake of NAND technology for higher capacity storage needs. Speaking in May, Jeff Janukowicz, research vice president at IDC, said QLC SSDs had the potential to expand the footprint of flash in the enterprise data center. “For read-intensive and performance sensitive workloads, QLC enterprise SATA SSDs provide an affordable way to move enterprise applications to flash and have the opportunity to increase the addressable market for flash in the enterprise,” he commented.

The migration from spinning drives to SSDs won’t happen overnight though, as Dicker pointed out. The ecosystem includes APIs, interfaces, and algorithms, and none of those are easy to turn on a dime. Hyperscalers, and those members of the OCP,  have existing infrastructure that will require time and effort to change over if they decide that this new technology is viable and effective.

For read-intensive and performance sensitive workloads, QLC enterprise SATA SSDs provide an affordable way to move enterprise applications to flash and have the opportunity to increase the addressable market for flash in the enterprise. – Jeff Janukowicz, IDC