It has been too long, two-and-a-half years, since the last of Tom Coughlin's Storage Valley Supper Club
events. But he just organized one to coincide with the Flash Memory Summit
. It featured an extremely interesting talk by Karim Kaddeche, CEO of L2 Drive
, a company whose technology seems likely to have a big impact on the hard disk market. Follow me below the fold for the explanation. I didn't take notes, so what follows is from memory. I apologize for any errors.
Ever since the IBM 350 RAMAC
was introduced in 1956, disk heads have used an air-bearing
In a hard drive, the heads 'fly' above the disk surface with clearance of as little as 3 nanometres. The "flying height" is constantly decreasing to enable higher areal density. The flying height of the head is controlled by the design of an air-bearing etched onto the disk-facing surface of the slider.
The role of the air bearing is to maintain the flying height constant
as the head moves over the surface of the disk. If the head hits the
disk's surface, a catastrophic head crash can result.
The RAMAC's heads were about an inch in diameter. Current technology heads are much smaller, and are made of ceramic via one of three fabs. If you could halve the flying height, you could quadruple the areal density, but one of the problems in doing so is that the disk arm and the head it carries exist in a 100-mph tornado. The aerodynamic forces make the arm vibrate at a high frequency (flutter), making it more and more difficult to prevent head crashes as the flying height decreases.
In the past decade there hasn't been much improvement in head technology, R&D has been focused on the long- and still-awaited HAMR and MAMR head and platter technologies. Increases in areal density (and thus decreases in $/GB) have been almost completely due to three factors:
- Shingling. Moving the tracks so close together that they interfere with each other increases areal density but either requires major operating system changes (host-managed shingling) or causes significant performance issues (drive-managed shingling). State-of-the-art shingling increases capacity by about 14%.
- Extra platters. Adding platters increases drive capacity without increasing areal density, but it increases cost, and there's a limit to how many platters can be squeezed into the 3.5" form factor. I believe the current limit is 9, so even if an extra one could be shoe-horned in it would increase capacity only 11%.
- Helium-filled drives. Because Helium has lower density than air, the aerodynamic forces in the arm are lower and the head can safely fly lower, increasing the areal density and allowing space for an extra platter. And it reduces the power needed to spin the platters.
Here is L2 Drive's image of the flying height in a state-of-the-art hard drive. Note that despite the gap maintained by the air-bearing being 1-2nm, the magnetics-to-magnetics gap is 7-10nm.
Reducing The Gap
Apparently, some years ago Seagate asked themselves "if Helium is good because it reduces flying height and power demand, could we go one better and remove the gas entirely?" They were stymied by the lack of a way to manage the flying height without an air-bearing (or in the case of Helium, a gas-bearing).
But Kaddeche is a car guy, and remembered that in the early 90s, until it was banned, Formula 1 cars had "active suspension"
Active suspension first found its way into Formula one racing in 1981 when teams were searching for a means of using "skirts" fitted to the sides of the cars to increase aerodynamic downforce and improve handling. Crucial to the success of the design was a controlled ride height: Enter Lotus’ first ever active suspension.
These early systems used hydraulics to control the car’s attitude in response to bumps in the road, rather than proactively preparing for the event ahead of time. Although they were effective in controlling the car’s height, they were slow to react to inputs.
In 1983 Lotus refined the system by fitting an onboard computer to actively control the ride height, at which point the technology was deemed effective enough to be transferred to road cars. ...
Having achieved considerable success in F1 racing, active suspension’s development also came to an abrupt halt in 1994, when it was banned because of safety concerns over the high speeds attained whilst cornering.
L2 Drive's idea is to evacuate the drive and:
- use a capacitance sensor to measure the head-platter separation at 60KHz,
- use a piezo-electric actuator to move the head up and down at 15KHz to maintain the desired spacing.
They have evidence that, in a vacuum drive, flutter is almost completely eliminated, with no frequencies about 2.5KHz having amplitude above 1nm. Thus a 15KHz actuator should be able to maintain extremely accurate head spacing.
Their approach has a synergistic set of benefits:
- Because the head-platter separation is actively managed, the head and platter are not going to come into contact. Thus there is no need for lubricant on the platter or head. This allows the effective magnetic-to-magnetic separation to be decreased, increasing the areal density.
- In a vacuum there is no corrosion, so the protective layers on the head and platter can be greatly reduced, further decreasing the effective magnetic-to-magnetic separation, increasing the aereal density.
- Because there is no aerodynamic drag, the power needed to spin the platters is reduced, requiring a smaller motor and reducing the lifetime cost of drive ownership.
- Because the head doesn't touch the platter, the drive reliability is increased, reducing the lifetime cost of drive ownership.
- Because the head doesn't touch the platter, the head can be made on silicon wafers instead of on ceramic. This is not merely cheaper, but will allow conventional silicon fabs to compete in the head manufacturing space for the first time, further reducing the cost. In the future this will allow more functions to be integrated onto the head.
- Because the heads will be much cheaper, the additional cost of increasing IOPs by adding a second actuator will be significantly reduced.
- Active management of the head-platter separation will increase the drive's shock tolerance, a parameter which has been steadily eroding.
Here is L2 Drive's image of the flying height they expect. Note that the magnetic-to-magnetic separation is 2-4nm, or more than halved.
Why Does This Matter?
L2 Drive asserts
By our calculations, existing HDDs using PMR would see an increase of ~40% in their capacity.
and they estimate the technology might ship in 2022, adding a small amount to the parts cost.
These days the development of SSDs has confined hard disks almost completely to the nearline market, storing vast amounts of infrequently used data in data centers. I used this graph from Chris Mellor's How long before SSDs replace nearline disk drives?
in SSD vs. HDD
. It shows that the rate of improvement in the $/GB for nearline HDD since 2013 (in fact since the recovery from the floods in Thailand) has been marginal. This is the basis for the optimistic projections
that SSDs will replace HDDs in their last remaining large market.
When HAMR and MAMR finally ship in volume they will initially be around 20% lower $/GB. L2 Drive promises a cost decrease twice as big as the cost decrease the industry has been struggling to deliver for a decade. What is more, their technology is orthogonal to HAMR and MAMR; drives could use both vacuum and
HAMR or MAMR in the 2022-3 timeframe, leading to drives with capacities in the 25-28TB range and $/GB perhaps half the current value.
The desperate search for ways to delay shipping HAMR or MAMR drives continues. Chris Mellor reports that Seagate, WD mull 10-platter HDDs as pitstop before HAMR, MAMR time:
"With 10-platter conventionally recorded disk drives touting capacities of up to 20TB by 2021, the arrival of HAMR and MAMR drives could slip back to 2022."
"Wells Fargo analyst Aaron Rakers said SMR nearline volumes remained low. Vendors must sort out their host-managed schemes to better meet performance requirements before large-scale adoption takes place, Trendfocus added."
Post a Comment