Flash Memory Summit 2019: What’s Next and What’s Missing in the Future of Flash (Video)

This week, I was in Santa Clara, California for my annual pilgrimage to Flash Memory Summit (FMS). While I was there, I got a few moments during a very busy week to stand in front of a camera with my colleague, Mark Peters, and discuss this year’s summit, as well as the latest news.

Before I get to the new tech, I feel I must address the big question: “Is Flash Memory Summit still as important as it once was?” With flash memory becoming the de facto standard for data storage now, do we need a separate event for flash? Isn’t everything storage-related about flash now?

If you look at the video, you will see a busy show floor packed with attendees. While the show certainly has evolved to more of an internal industry discussion and collaboration on flash, rather than an end-user conference, the conversations at FMS are no less valuable.

 

 

Business’s demand for data performance is insatiable. Modern businesses are built on data, and every innovation that reduces the latency involved with data access, whether PCIe 4.0, NVMe, NVMe over fabrics, or persistent memory, offers a valuable edge for IT.

And few workload trends illustrate this incredible demand for data performance more than the rise of machine learning. Nearly half of the keynote presentations at FMS 2019 had some reference to artificial intelligence (AI) or machine learning in the title. FMS is not just about flash storage; it is about taking flash to the next level to transform your business.

With all this new technology, there is something missing, though, and not just from FMS but from IT as well. It is appropriate that artificial intelligence was a dominant topic at FMS this year, because it is another type of intelligence that needs to increase its presence in modern IT. What is lacking in modern IT is a sufficient level of detailed workload intelligence at the infrastructure level.

Will the innovations shown at FMS accelerate your application environments? Yes, probably. But by how much, and which investments offer the greatest returns? Should you leverage an NVMe-based storage system? This is probably a no-brainer. But what about persistent memory? What about both? Which will give you the biggest improvement?

Now, flash storage vendors have invested considerable time understanding how their technology will impact different workload environments. The problem, however, is that modern IT organizations often lack a detailed understanding of their own specific workload environments. This creates an intelligence gap. The scale of modern IT is becoming so large and the technology is evolving so rapidly that IT’s lack of the tools and the time necessary to understand the details of their specific workload requirements will likely become a major hurdle to new technology adoption.

There is a wealth of new flash-based technologies poised to transform the data center. There is also a wealth of demand. Workloads, such as machine learning, are fueling a need for the low latency performance that these technologies offer. Seems like a perfect match. It is, just not quite. IT needs better tools to understand their specific workload ecosystems to maximize their return on the flash innovations just over the horizon. Solving this gap will be the difference between achieving an evolutionary, incremental performance improvement and capturing a transformational advantage.