News & Analysis
Comment
Manolito33
Nanoimprint says to get down to 5 nm. If HDD need this technology then ASML must ...
WSOCT
IMO, since a large chunk of customers in the HDD segment are individual users, ...
Decision time looms for hard drive makers
Rick Merritt
8/19/2010 6:29 PM EDT
SAN JOSE, Calif. – Currie Munce, vice president of worldwide research at Hitachi GST, sees a big change ahead in hard disk drives.
"We think it will be needed in the 2014 or 2015 timeframe, so we have to get ready for mass production," a process that can take three years, Munce said in an interview with EE Times.
"So in the next two years we have to make some significant decisions," Munce said. "The supply base needs a common direction from the industry, and that can't come from one company alone," he said.
That's why Hitachi GST, Seagate and Western Digital formed the tentatively named Storage Technology Alliance. The group will define a road map for hard drives and drive research to meet its milestones.
The three have initially kicked in a half million dollars each into a pool to form the group. They pitched Samsung on joining this week in a U.S. meeting and will talk to Samsung and Toshiba in Asia shortly.
But just what decision this still-evolving group will make is still unclear.
Executives from the three companies and Toshiba said this week they believe today's perpendicular recording can be extended using a new shingled magnetic recording (SMR) technique now in develop to pack as much as 1.5 Terabits of data on a square inch of disk space. SMR will take them out to about 2015 at best, then something radically different will be needed.
Seagate and Western Digital believe that is some form of recording that uses tiny but magnetically stable materials on a disk that to be read must be briefly heated with a laser diode or other device.
"We have been doing R&D on this technology for many years, feel that it is on a good path to commercialization and is less disruptive to the drive architecture and optimization," said Mark Re, senior vice president of recording media operations at Seagate in an email exchange about its approach called heat assisted magnetic recording (HAMR).
Hitachi GST and Toshiba have been researching new ways to pattern tiny bits on a disk that don't require heating to be read. Toshiba reported progress on the so-called bit patterned media approach this week.
But each company is doing some work in both areas and both fields include various derivative approaches. Indeed Seagate presented three papers on bit patterned media this week and Hitachi gave two on HAMR.
"I would say our efforts in bit patterned media have remained roughly constant, but we have been about doubling our work on HAMR over the last two years," said Munce.
HAMR's challenges include finding the right recording materials, then solving a range of engineering challenges such as how to integrate laser diodes and recording heads. Patterned media proponents have yet to demonstrate ways to cover a full disk with tiny magnetic dots in a way that can be mass produced and adds no more than two dollars to the cost of a disk.
The problems are huge and have implications all up and down the hard disk supply chain. Insiders say they have run the numbers a few times but decline to share estimates some say stretches into the billions of dollars for an industry with historically penny-pinching budgets.
"While R&D spending is likely flat to slightly up, we have become much more efficient in how we spend these dollars," said Re of Seagate.
Indeed, hard drive R&D budgets are not expected to expand substantially despite the looming and expensive technology transition. Drive makers' revenue growth hovers around ten percent a year at best and R&D budgets are typically at fixed percentages of sales.
"Doing more leading edge work in collaboration with our manufacturing sites allows capital leverage and the ability to transition technologies more quickly," Re said. "Some of the farther out research that may have all been done internally in the past is now done is collaboration with universities, or with industry/academic consortiums," he added.
The new STA group aims to kick that collaboration up a notch. Munce describes it as a magnifying glass to focus sunlight into a beam capable of burning a hole through a piece of paper.
Most drive makers are already members of the Storage Research Consortium, but that group has a Japan-specific charter, said Munce. STA is modeled on SRC and the two groups likely will collaborate, he added.
"We are looking for ways to share because the long-term goal overall industry convergence [on a road map], but it takes awhile for these things to congeal," said Munce.
That convergence needs to happen in two years if hard drives—increasingly challenged by solid state flash memory—are to keep pace.
|
|
|
|
|
LarryM99
8/20/2010 1:22 PM EDT
You have to wonder if increasingly complex and precise systems to increase storage density might be a dead end. There will be a market for higher-capacity drives in data centers, but for most users they might just be overkill. The paradigm of one or more hard drives per computer may fall victim to a combination of fast SSD boot medium and fast network connections to shared storage. This could radically shrink the market for hard drives or at least constrain the growth to a fraction of that of the overall computer / device market.
Larry M.
Sign in to Reply
chuckfergus
8/21/2010 5:50 PM EDT
I have Windows Media PC rigged as DVR on my cable to record programs of interest. I started with 1/2 TB and filled it up way too fast. I upgraded to 1 TB and it is currently packed. It didn't take long to do this.
Demand for capacity will never end.
Sign in to Reply
CJS2
8/20/2010 1:56 PM EDT
I think what you said has merit. You have to wonder how long the current trend will continue. In the short run, I think the drive to increase storage capacity for consumers and industry will continue. Certainly, consumer appetite for storing digital information will likely continue to grow, especially as more and more digital information (i.e., video) becomes more widely available.
Sign in to Reply
SallyF
8/20/2010 2:48 PM EDT
Larry, much higher capacity is already needed. The slow down in storage capacity increases is already affecting the computer industry overall. The percentage of people owning HD cameras is increasing which means the amount of data to be stored is increasing. And we need more, much more. The storage industry, and Microsoft's buggy and excruciatingly slow file handling are choking a potentially vibrant market in video storage and editing. Shared capacity on a network won't solve the problems, in fact that would exacerbate the problems. Moving large files over a network is even slower than the already slow handling of hard drives. I hope people in the hard drive industry and competing operating systems know that the market is large and will grow. Wait till/if 3D becomes more prevalent!
Sign in to Reply
jaybus
8/20/2010 4:21 PM EDT
Higher density is certainly going to be needed very soon. Don't underestimate the non-pc market. Set top DVRs from cable/satellite providers is a huge hard drive market. 3D channels are already here, and more will soon be coming.. HD is already prevalent. Space and cost restraints prevent putting multiple drives in a DVR, and DRM prevents, or at least makes cumbersome, network storage of recorded video.
Sign in to Reply
docdivakar
8/20/2010 5:40 PM EDT
I think the hard drive industry needs to continue to provide reliable cost-effective solutions for many years to come. A comforting gap they have from their SSD counterparts is in reliability and cost that is not in any immediate danger of being overtaken.
HAMR technology sounds interesting (I am some what familiar with its SSD cousin called thermally -assisted TAS technology in MRAMs) but I think the jury is still out on whether that is the technology that will keep the pace of capacity and cost for the HDD industry. The growing trends in virtualization servers and storage will demand lower and lower cost of storage. This demand is even more acute in M2M versions of nodes where lately there have been discussions of putting more storage within servers for adhoc networks.
MP Divakar
Sign in to Reply
ProfChuck
8/20/2010 6:00 PM EDT
The rapid pace of development of non volatile solid state memory will make hard drives obsolete within 10 years, perhaps 5. Speed, Capacity, and size are the drivers for development. All of these are easier to achieve with solid state memories than mechanical systems.
Sign in to Reply
docdivakar
8/20/2010 6:29 PM EDT
@ProfChuck, I am in general agreement with your observation. I wouldn't trust mechanical clunkers over solid-state drives either. How ever, what I learnt at the FlashMemory summit this week is that there are lingering reliability issues of SSD's that many vendors are unwilling to publish or share. The sooner they make the FMEA results public, the faster the resolution of their reliability issues will be. Therefore HDD's still rule the datacenter applications.
While speed may be a primary justification for adopting SSD's, some experts opined that more is needed in making a case for SSD's to push them over the cost barrier. Perhaps smaller technology nodes and progress in 3D chip stacks will eventually bring the cost down but 15nm with EUV is still years away!
MP Divakar
Sign in to Reply
ProfChuck
8/20/2010 6:50 PM EDT
I think that HDD's will continue to dominate the mass storage market for another 5 years, maybe longer. However, in the long run the SSM's will rule the day. There are many factors that will bring this about. I like Arthur C. Clark's admonishment, "No machine may have any moving parts." There are so many factors that favor SSM's that it is difficult to see any other technology, especially mechanical ones, being victorious in the long haul. Multi terabyte drives that you can hold in the palm of your hand are amazing but multi gigabyte drives that are smaller than your little finger nail are astonishing. Terabyte systems of this size are being developed now and should be on the market within 2-3 years. I have been involved with computers since the late 50's and have seen many orders of magnitude advances in performance, size and cost. My two year old Iphone 3G has more computing power than the supercomputers of a few years ago and other than 4 buttons it has no moving parts. It boggles the mind what the next 10 years will hold.
Sign in to Reply
TanjB
8/20/2010 8:33 PM EDT
Actually, there are moving parts in circuits these days. They may be in the form of electromigration of wires, or electrons trapped in defects, or dopant profiles that shift, things which at human scale appear to involve no motion, but when your device is nanoscale and the effect is not reversible, it is in effect just as serious as the wear we see in mechanical movements. Flash cells in particular are near their wear limits. Disks appear to have at least as long a horizon of improvement as Flash. Other forms of NVM hold promise but as yet are nowhere near the price and maturity they need to be contenders.
Sign in to Reply
prabhakar_deosthali
8/21/2010 7:24 AM EDT
In my opinion the popularity of cloud computing in the coming years and the availability of faster networks may make local HD storage a thing of past, paving way for massive storage capacity requirements in the Cloud servers. This will require a totally new way thinking by the HD manufacturers. Thus though the total capacity requiremnt may increase many fold the type of storage devices used may see a sea-change
Sign in to Reply
resistion
8/21/2010 12:37 PM EDT
I agree the cloud computing paradigm is coming and it will reduce the use of HDD's or SSD's to only servers. The client devices (notebooks, PC's) would be replaced by smaller or some would say 'dumber' devices like netbooks or tablets or even smartphones. Eventually HDD's will have to be effectively built to supercomputer requirements.
Sign in to Reply
shukla
8/21/2010 4:37 PM EDT
In my opinion there will be huge demand of high capacity storage in video surveillance applications.As video surveillance ownership cost is going down day by day and development of high resolution camera,generates demand of high capacity storage.Please don't forget that video data storage requirement is quiet different from data center and general computing storage requirement.for example 60 GB hard drive required for 1 day recording of 1 mega pixel camera.It is very common to keep video recording safe for 60 to 90 days.It mean that for 60 days recording 3.6 TB HDD required.It is not alone sufficient for data security then mirror and backup storage will require.So high capacity HDD demand is unlimited and exponentially rising day by day.
So I don't think High capacity HDD market will shrink till SSD will become cost effective and mature.
Sign in to Reply
goafrit
8/21/2010 5:29 PM EDT
The future of HDD remains bright since the world is moving to the digital domain. In all cases, we will need to store more. That said, the underlying technology must have to evolve and that is where innovation will matter.
Sign in to Reply
WSOCT
8/23/2010 10:41 AM EDT
IMO, since a large chunk of customers in the HDD segment are individual users, cloud computing isn’t going to cause any major changes in the storage landscape. While I’m all for enhancing HDD technology to reduce the footprint and improve access speed, I’m not sure if an alliance having handful member companies is the best way to move forward. I’d rather pitch for an open alliance with a larger participation base in order to avoid interoperability fiascos.
Be it individual users or companies using cloud-based services, storage needs are increasing for everyone. I’m excited by the new developments in HDD space, especially HAMR looks very promising. I understand that Toshiba and others are under severe pricing pressure as they spend a large proportion of their revenues into R&D activities such as HAMR and therefore there’s definitely a need for low-cost mass production.
As for HDD losing out in the race against solid state flash memory, I’d like to remind that the faster you run, the harder you fall.
- Keith Schaub
Sign in to Reply
Manolito33
8/24/2010 7:22 PM EDT
Nanoimprint says to get down to 5 nm. If HDD need this technology then ASML must buy Obducat or Molecular Imprints other wise ASML will be out of business soon.
EUVL would have no chance against nanoimprint then.
Me
Sign in to Reply