Now that I’ve got your attention, “the death of the SSD” can be explained in two different ways. The first, and least important, is the fact that the term SSD or solid state drive should never have been used and adopted. Drive? What drive? The term itself is an oxymoron. Solid state by definition: designating or pertaining to electronic devices, as transistors or crystals, that can control current without the use of moving parts, heated filaments, or vacuum gaps. Unlike a hard disk drive (HDD) that has a motor and moving parts, our solid state solutions have no moving parts. The only things being “driven” here at Micron are the engineers trying to keep up with the customer demands for bigger, faster, and cheaper memory solutions.
The second and most important reason for “the death of the SSD” is the fact that treating solid state storage like an HDD has simply outlived its usefulness. Its purpose was to get customers who have used hard disk drives for the past three decades to use solid state storage. There’s nothing like backward compatibility when you’re trying to introduce something new to the industry. So emulating HDDs was the quickest and easiest way to get customers to use solid state storage.
So what can we expect going forward?
Because I spend most of my time thinking about the market in 3–5 years, I enjoy meeting with customers who have been using solid state storage for 5+ years—the so-called early adopters. Now that they believe our total cost of ownership models, which they now repeat back to us, they are asking us why we ever put flash memory behind an interface as slow as SATA? We, of course, have to remind them that at the time it was the only way to get them to use the technology. As they shrug it off, we talk about ways to architect solutions that don’t simply read and write 4KB at a time, like an HDD, and ways to add significant value via CPU offload functions.
We are just beginning to scratch the surface with respect to how solid state storage can reduce power, increase performance, and be the change agent for significant cost reduction at the system level.
What do you think? Where are you seeing flash forge a new storage path that is different from HDD?