So, out of curiosity when researching the Raspberry Pi 4, I decided to take a look at what prices terabyte SSDs are at today. Now they are at about $100 per terabyte? Wow. The technology has come a long way. I remember the days (early 2010s era) when SSDs like that would cost over a thousand dollars. Meanwhile, hard disk drives have been stagnant for too long, awaiting the maturization of technology in research such as HAMR and MAMR. Until then, you’re seeing market list prices around $200 per 6/8 terabytes. That is no longer a compelling differentiator from solid-state drives. Seriously, the capacity-to-price factor is only a difference of a factor of 3, 4, or maybe 5. No longer do hard disk drives have the factor of 12 advantage in storage economy.
You know what that means? As far as the near-term market projections are concerned, hard drives have failed. With the current momentum behind SSDs, everyone is prepared to see SSDs pass up hard drives in storage economy, and when that happens, despite the disadvantage of lower rewrite durability than hard drives, the vast majority of the market is going to quickly switch over to using SSDs exclusively, and suddenly there will be a crash in revenue available for research and development of hard disk drive technology. You can argue that it’s already happened today, that because there is so much more money going into the SSD market, they could fund more research, move faster, and pass up hard disk drives. Soon enough, we may end up viewing hard disk drives just as we view CRT monitors: a bygone legacy of computing times past.
Please note that there is a method to solve the problem of limited rewrite durability in solid-state drives, often called “self-healing.” The technique uses heat to assist in the process of erasure, which not only eliminates the wearing of the drive due to erasures, but also speeds up performance. So remember, it’s definitely possible to eliminate this disadvantage of solid-state drives, but no companies are currently pursuing putting this function in commercial drives at the moment. It’s just a matter of time until it works its way into commercial drives.
The prime innovation that you may be seeing in commercial drives, of course, is to use multi-level cells (MLC) and 3D NAND to increase storage density. Hindsight is 20/20, but you can see the obvious advantage of increasing storage density as a means to improve storage economy to the point that write-once-read-many (WORM) use is more economic for SSDs than for HDDs. The limited rewrite durability only concerns as small set of rewrite-intensive applications such as swap drives and caches.
What about long-term archival in cold storage? That is a quickly waning application area, especially thanks to cheap live servers like Raspberry Pi. Ultimately, that’s a shrinking market that hard drives once served well for the low-budget niches. But without hard drives, magnetic tape and tape drives still serve that market well.
I’ve discussed this with other folks who had useful insight questions to ask. “Who’s making it?” “South Korea. Samsung, SanDisk, and Micron are some of the big companies. Yeah, pretty much.” “They’re just a tiny little country, but they sure have a lot in electronics coming out of them.”