20TB Hard Drives Will be Made of Glass
Over the past few years, the rate of hard drive density improvements has dropped significantly. While Seagate and Western Digital have both pushed ahead with larger hard drives (often thanks to more platters and the use of helium), the actual rate of areal density increase has slowed these last few years. As Seagate and Western Digital push towards 20TB, glass substrates could be a critical component of that effort.
Right now, only laptop drives use glass substrates, which have several advantages over aluminum. First, glass is more rigid than aluminum, which allows the platters to be thinner (and lighter). Glass substrates are smoother and flatter than aluminum, which allows them to be packed together more tightly, and they expand less than aluminum under heat, which makes it easier to cope with thermal expansion. The amount of energy required to spin a glass platter at a given spindle speed will also be at least slightly lower than if the same platter was made of aluminum, due to the latter’s higher weight.
A Japanese company, Hoya, believes that current HDD manufacturers will adopt its glass disks for 3.5-inch drives (glass is already used for 2.5-inch laptop drives). Hoya has prototyped glass platters that are 0.5mm and 0.381mm thick, compared with the 0.635mm platters currently in use. While each drive platter must have a gap between itself and the next platter, cutting the platter size by 40 percent means you can pack more platters into the same drive capacity. Glass substrates are also useful for HAMR (Heat-Assisted Magnetic Recording), which uses a laser to heat the part of the disk being written to. Seagate wants to introduce HAMR-equipped HDDs by 2018 and hit 20TB of available storage by 2020.
Banking on Capacity
One reason I suspect Nikkei is right about glass substrates being more widely adopted in 3.5-inch drives is because the hard drive industry really only has one card to play against SSDs — sheer storage capacity. The explosion in cloud services and the advent of so many comprehensive backup solutions means enterprise demand for back-end storage is going to continue to grow. Companies may use SSDs and NAND to cache data; we’ve even seen some interesting proposals for using NAND as a replacement for DRAM in cache servers. But while companies like Samsung sometimes make headlines for creating enormous NAND drives that dwarf anything hard drives can offer, they don’t bring those products to market for good reason: Nobody could afford them at the price they’d have to charge.
As the cost gap between hard drives and SSDs has fallen, we’ve seen SSDs muscling into what was traditionally the turf of higher capacity hard drives. The consumer market is clearly moving to NAND flash in the long run, with enthusiasts still buying some hard drives for local backups and archival purposes.
Long-term, however, HDDs are going to be seen as cold (or at least cool) storage options. That means driving up capacities is the most important way to expand their appeal, and the businesses that adopt them won’t be put off, even if glass platters require a higher cost. SSDs aren’t going to be cost-effective if you need 100 to 500TB of storage in the near-term. With the total amount of data created worldwide continuing to grow, there’s no sign of slowing the world demand for data any time soon.
By Joel Hruska on September 5, 2017