In a case of stating the obvious UCSD & Microsoft have decided that come 2024 Flash is a doomed storage technology. That won’t matter to too many of us as we’ll have started the move onto the next big thing by the end of the decade.
I’ve stuck my neck out on that as far as it’s willing to go without having to face the executioner but there’s more reasoning behind it than just the issues UCSD-Microsoft discuss in their paper. However, just how it is stated every year that Moore’s Law is three years away from hitting a wall the authors don’t take into account semi-conductor manufacturers coming up with smarter ways to work around Flash’s well known issues.
The bottom line is the higher the density of Flash the lower the performance and if it’s SLC or MLC will continue to matter now and well into the future.
As people are hoping to close the pricing gap between Flash and Disk Drives and putting aside the fact we’d need to spend about $100B building new Fabs just to meet that capacity requirement, the fact remains that with current technology the closer you get to a disk drive capacity price point the less of a random read benefit you’re getting from the Flash itself so you might as well just use a disk anyway.
Fast low capacity Flash devices will kill off fast low capacity disk devices. Expect the disk drive industry to turn towards dense and ultra dense technologies with improved reliability as they exit the high speed disk segment.
