Issue No.12 - December (2006 vol.39)
Published by the IEEE Computer Society
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MC.2006.449
Disk-drive manufacturers are transitioning to perpendicular storage techniques, which provide modestly improved capacity with relatively slight changes to current production processes.
Over the years, storage has become an increasingly valuable commodity as demand has increased. Businesses are expanding their storage capacity by 30 to 50 percent annually due to increasing volumes of customer, supply chain, and other important information; greater use of video and audio; and government data-storage mandates, noted Forrester Research analyst Andrew Reichman.
To meet this demand, hard-drive storage density and capacity have grown by about 30 to 40 percent per year. In the process, the cost has dropped.
However, many industry observers say current hard-drive approaches are reaching the technical limits of their ability to increase capacity.
Moreover, users want more performance from their disk drives, particularly those used in servers, to keep up with improvements in other parts of their systems.
Disk-drive makers face an additional challenge from NAND flash memory, which is playing a larger role in smart phones, PDAs, digital music players, and other devices that previously used minidrives.
To respond to these challenges, drive manufacturers are turning to new techniques.
Today, most are transitioning to perpendicular storage techniques, which provide modestly improved capacity with relatively slight changes to current production processes.
They are also looking at more complex advancements, such as heat-assisted magnetic recording (HAMR) and patterned media.
These approaches could increase hard-disk storage densities from today's 100 gigabits per square inch to perhaps 100 terabits per square inch, noted Mark Kryder, senior vice president of technology at leading drive maker Seagate Technology. New approaches could also increase data-transfer rates from the maximum 300 megabytes per second of today's hard drives.
And companies are pursuing more exotic storage technologies such as holographic media, which one vendor is about to release in a product; and nanostorage, which may be years away from commercialization.
All the new techniques face development and manufacturing challenges that will require years to overcome, noted Jim Porter, president of Disk/Trend, a hard-drive industry analyst firm.
Driving the New Drives
Several factors are driving storage-technology advances.
The superparamagnetic limit
Most hard drives consist of an aluminum or glass disk—coated with a thin ferromagnetic film—that spins quickly to permit high-speed data access.
The read/write head generally includes a component—also coated with a ferromagnetic film—that writes data to the disk longitudinally by inducing a magnetic field that produces magnetic bits whose positive or negative orientation corresponds to binary data's zeros and ones.
Inductive heads read data by converting the magnetic charge of the bits spinning by on the disk into an electrical current.
In 1995, Carnegie Mellon University professor Stanley Charap said that traditional hard-drive technology's capacity would be limited by the superparamagnetic effect.
A key way vendors have increased hard drives' capacity is to shrink the magnetic crystals they use to store data. However, if the crystals get too small, they become increasingly susceptible to thermal fluctuations, which can change their magnetic orientation, potentially causing data loss or corruption. It is also difficult to keep reducing the size of read/write heads to work with increasingly small crystals.
Over time, as technologies have improved, experts have changed their estimates of traditional longitudinal drives' superparamagnetic limit. The current estimate is about 200 gigabits per square inch.
The flash challenge
Flash memory is increasingly challenging hard drives in the handheld market. In the past, flash was used principally in devices that didn't require much storage, such as digital cameras and embedded controllers.
Now, though, NAND flash memory has more capacity and a lower cost per unit of storage than in the past (although still more than hard drives), making it suitable for new uses such as Apple's popular iPod nano and other digital-music players. Flash also offers the advantage of smaller size and no moving parts, which leads to lower power usage and greater durability.
Don Barnetson, Samsung's associate director of flash marketing, predicts that makers of ultraportable PCs could move entirely to flash by 2007 or 2008.
Disk vendors are currently focusing on achieving higher storage capacities via modifications to existing technology.
For example, Samsung and Seagate are working on flash/hard-drive hybrids, noted Frank Dickson, an analyst with market-research firm In-Stat. To minimize the use of energy-hungry hard drives, the hybrid approach stores data on a flash chip until it is full, dumps the contents to a disk, and then writes to the chip again.
The main markets for hybrid drives will be Windows-based PCs and handheld devices because the cost will be too much for price-sensitive smaller devices, said independent flash-market analyst Jim Handy.
Manufacturers are buying time for current hard-drive technology by building up their storage media vertically, as well as longitudinally.
In a perpendicular system, shown in Figure 1, highly magnetically responsive crystals stand vertically out of the surface of the disk, rather than lying horizontally along it. More crystals fit on a disc when they are vertical rather than horizontal. A permeable second layer makes the system more effective by focusing the magnetic field vertically.
In addition to using a new type of disk, perpendicular storage requires new heads to read and write vertically. Otherwise, the basic manufacturing process is the same as it is for traditional disks.
Nonetheless, noted John Rydning, an analyst with market-research firm IDC, "Perpendicular storage is a pretty major transition of the industry."
Toshiba is already using perpendicular technology to deliver 200-gigabyte microdrives with a storage density of 178.8 gigabits per square inch, said Maciek Brzeski, vice president of marketing for the company's Storage Device Division.
Seagate is transitioning all of its drives to this technology, which Kryder predicted could scale to 500 gigabits per square inch.
Perpendicular storage will let manufacturers increase drive capacity for five more years before the super-paramagnetic effect kicks in, he said.
Seagate is focusing on HAMR, which uses a finely focused laser in the write head to warm up spots on a disk—containing heat-resistant iron-platinum grains instead of the cobalt-platinum grains used in traditional hard drives—immediately before recording data on them. The use of the finely focused laser lets the drive record smaller bits, which can be packed closer together without affecting nearby bits.
The technology could produce drives with densities of up to 10 terabits per square inch, said Kryder, adding that HAMR also offers great promise because it uses traditional magnetic-media manufacturing processes.
However, it could be challenging to integrate a laser that can heat up the desired bits without affecting nearby ones into the recording head, noted Toshiba's Brzeski.
In patterned media, manufacturers use electron-beam or nanoimprint lithography to etch well-defined and physically separated magnetic nano-structures onto a disk's surface. The heads would write data to each nanostructure.
This technique, which Hitachi and Toshiba are pursuing, would let each crystal pack a larger amount of data more closely together than in the past because the physically separated bits are less likely to affect nearby ones.
However, this approach requires the use of lithographic techniques like those used to etch circuitry patterns on chips, an expensive process not used in making hard drives today, explained Seagate's Kryder.
"The challenge is how to do this in an assembly line really inexpensively," said Hitachi chief technologist John Best. "If we can do that, we can expect one to 10 terabits per square inch of [density]."
"This is one of the issues we have to work out over the next two to three years," said Brzeski.
Hitachi and Toshiba said they still have considerable work to do to develop the technology and thus did not want to predict what its data capacity or transfer rate will be when released commercially.
Researchers are looking into two novel storage technologies, including one that is about to hit the marketplace after years of research and discussion.
Holographic storage, the subject of global research for decades, aims to store data in holograms—laser-generated images that yield 3D effects on 2D surfaces—in multiple layers throughout an optical medium. It could lead to storage systems that contain exabytes (a million terabytes) of data, said Michael Thomas, president of vendor Colossal Storage.
A holographic system encodes a data set's binary zeros and ones as an electronic bitstream. Lasers store the bitstream as a holographic image by projecting the data along various planes within a medium such as a photopolymer block. A single storage device could hold multiple data holograms, yielding high storage volumes.
The systems read data by shooting a laser at a specific angle into the medium and projecting an entire hologram onto a camera-like, sensor-based imaging device. Accessing an entire hologram at one time makes the read process very fast.
InPhase Technologies plans to produce the first commercial holographic-storage product, slated for beta testing this month and commercial release early next year.
Improvements in the media, lasers, and sensors have made it possible to release holographic storage commercially after years of study, said Kevin Curtis, InPhase's chief technology officer.
The company's Tapestry HDS5000 product will use a removable 300-gigabyte photopolymer block with a storage density of 350 gigabits per square inch and a data transfer rate of 20 megabytes per second, he noted.
The device would be useful to companies that must archive and transport massive amounts of data. The system will sell for about $15,000 initially, but InPhase expects the price to drop over time.
By 2008, InPhase plans to introduce rewritable versions of the technology. And by the following year, the company expects to sell small, 25-gigabyte holographic-storage products for use in consumer electronics, according to Curtis.
During the next few years, he noted, InPhase plans to release 800-gigabyte and 1.6-terabyte products with a transfer rate of 120 megabytes per second.
According to Curtis, his company has already demonstrated systems with storage densities of 515 gigabits per square inch, which could increase to 2 terabits per square inch during the next decade.
Nonetheless, holographic storage is still unproven in terms of economies of scale and has not been shown to be useful for regular rewriting, as opposed to write-once archival storage.
IBM and Seagate are working on similar approaches to nanostorage technology, which uses micro-electromechanical systems (MEMS) to store data.
At the heart of IBM's Millipede nanostorage technology is an array of V-shaped silicon cantilevers on a 10 mm × 10 mm chip. Each cantilever is 70 micrometers long, and at its tip are both a micrometer-sized sensor for reading and a heating resistor for writing. A recent design consisted of 4,096 (64 × 64) cantilevers.
The storage medium is a thin polymer film on a silicon substrate, which moves across the cantilever tips during the read or write process.
The system writes bits by heating the cantilever tip, which softens the polymer and makes a tiny indentation in it. To read data, the system heats the reading sensor slightly. The sensor then moves into indentations and cools somewhat, creating a measurable change in its electrical conductivity, which is read as data.
Generally, Kryder noted, the data writing and reading process is relatively slow, which is why systems use multiple heads that run in parallel. Neither IBM nor Seagate is considering using nanostorage to replace hard drives. IBM has expressed interest in using nanostorage as a flash-memory replacement. Seagate is talking about using nanostorage to create low-cost storage for handheld devices, Kryder said.
IBM and Seagate are still researching the technology and said it is thus too early to guess at data capacities, densities, and transfer rates or when the companies might release products.
IBM says it has a prototype that has already achieved a storage density of about 1 terabit per square inch.
Despite the challenges, most experts agree that hard drives will continue to dominate the computer storage market for at least the next five to 10 years. They say hard drives—unlike the more exotic technologies—have proven economical and effective in mass-produced, heavily used, durable commercial applications.
In the immediate future, the major drive manufacturers will continue to focus on squeezing improvements out of perpendicular storage, which will avoid the need to make major changes to their production lines, noted Disk/Trend's Porter.
IDC's Rydning expects to see commercialization of patterned media by 2010 and HAMR drives by 2012, in devices such as PCs, servers, and digital video recorders.
Eventually, said Seagate's Kryder, vendors anticipate they will use both HAMR and patterned media to create drives capable of storing up to 100 terabits per square inch, which is hundreds of times more dense than today's high-end hard drives.
And, Hitachi's Best said, exotic technologies such as holographic storage and nanostorage will more likely be used in niche markets rather than compete with hard drives for general computer usage.
George Lawton is a freelance technology writer based in Brisbane, California. Contact him at firstname.lastname@example.org.