One of the more lively discussions these days is the future of the hard disk drive, or HDD. HDD’s have been around for over a half-century, thanks to the work done by Rey Johnson and his IBM San Jose team in producing the world’s first disk drive in 1956. That very first drive, the IBM 350 Disk Storage Unit, was just like the ones being produced today; comprised of 3 basic elements: rotating aluminum platters, magnetically coated surfaces, and moveable recording heads.
Like all good technologies, the reason the HDD has survived so long is that nothing better has come along worthy of replacing it. During the 50+ year evolution of the HDD, components have become smaller while capacity has grown in incredible magnitudes. The IBM 350 literally required a forklift to move it, while today’s drives fit comfortably into the palm of your hand and store 1 million times more capacity than their ancestors.
While the HDD has had an amazing run, someday it will be replaced. Precision motors and actuators are subject to mechanical wear and will not perform indefinitely. Keeping drive platters spinning 24 hours a day but accessing data occasionally is an inefficient use of electrical energy. Finally, while HAMR (Heat Assisted Magnetic Recording) technology should take HDD capacities to at least 100TB, at some point a capacity, reliability, and economic ceiling will be reached.
Bring on the SSD’s
Currently, the only serious threat to the existence of HDD’s are the emerging class of Solid State Drives, or SSD’s. SSD’s were first built using DRAM but today largely depend upon NAND Flash modules. These devices emulate the look-and-feel of an HDD but since they are composed entirely of silicon, are not subject to the mechanical wear or the rotational latency delays suffered by HDD’s. SSD are growing in popularity but are still dwarfed in comparison to HDD shipments. There are two primary reasons why SSD sales are lagging – a) high costs and b) reliability concerns. On a cost per GB basis, SSD still commands about a 6X price premium over HDD. From a reliability standpoint, the “flash” in Flash means that memory cells are weakened each time an electrical erasure is performed – which is required each time data is stored in Flash.
A Flash in the Pan?
Flash vendors are tackling SSD reliability issues with sophisticated wear-leveling algorithms, and manufacturing economies of scale are driving costs down, but I don’t believe that Flash will be part of the long-term success of SSD’s. New technologies, such as Intel’s Phase-Change Memory, HP’s Memristor, and IBM’s Spin Torque Transfer should ultimately prove to be a more viable option for SSD design than Flash.
So, while SSD vendors continue to reduce costs and improve reliability, HDD vendors will continue to chug along with a proven formula and ever-increasing capacities. Last year, Seagate announced plans to produce an HAMR 60TB HDD within this decade. Capacities like this will make it awfully hard for SSD vendors to keep pace. As such, I don’t expect to see a major cross-over occurring between HDD and SSD for about 10 years, or perhaps 15 years at the most. In the meantime, we’ll continue to see Flash and SSD technologies implemented tactically, as an accelerant in hybrid storage arrays or in small capacity (purpose-built for high performance) all-Flash storage arrays.
Today we are seeing traditional HDD storage arrays transition into hybrid Flash/SSD/HDD storage arrays. Within 5 years we’ll see Flash being replaced by a new/better solid-state memory technology. Finally, HDD’s will be entirely replaced by new-gen SSD’s within 10-15 years – that’s the way I see it…how about you?
Data Storage Matters,