Published: 10 Mar 2003
|Here comes holographic|
It's an interesting conundrum. The people who measure for SLA compliance are the same people being held accountable for them. Most experts say, however, that the system works if both the user side and the storage side are working towards the common goal of driving down costs. For example, if storage services operate on a chargeback basis, storage managers are probably working towards specific cost cutting goals. It's therefore in their best interest to meet the SLAs, says Mark Friedman, the vice president of storage technology at DataCore in Ft. Lauderdale, FL. "You enforce SLAs best at the level of the cash register," he says.
If the need for Blu-ray sometimes seems far fetched, holographic storage has risen to the level of science fiction. Since the 1970s, the notion of holographic storage has been tantalizing: Instead of having a single laser burn bits into a recording layer, two lasers could create a holographic 3-D image of data, which in theory could result in much higher capacity and faster read performance. But the stumbling blocks--particularly the search for a suitable storage medium--have been huge.
Bell Labs, one of the world's premier research facilities, worked on holographic storage for 10 years. In early 2001, Lucent--Bell Labs' new parent company--spun off InPhase Technologies, whose research team included scientists responsible for 20 patents on holographic storage. By that time, researchers had at their disposal new laser technology developed for DVD and improved CMOS sensors used in digital cameras--the latter vital for reading embedded holographic images. In April 2002, InPhase demonstrated Tapestry, a write-once holographic recording system targeting the professional video industry.
Just like the first holographic system developed in the 1960s, Tapestry relies on a single laser beam split in two: a reference beam and a signal beam. With Tapestry, the signal beam passes through a modulator that works like a transmissive LCD screen. The modulated signal beam becomes a projected laser image of data, so that when the two beams reconverge in the recording medium, the resulting interference pattern creates a 3-D image that traverses the entire thickness of the material. To read data, the reference beam passes through the recording medium at various angles, enabling a CMOS sensor on the other side of the disc to capture multiple data views of a single physical segment.
Special multiplexing techniques involved in writing and reading holographic images were only part of Tapestry's breakthrough technology. The largest single advance is the photopolymer recording medium, which displays high photosensitivity plus the highest dynamic range of any holographic material--not to mention the ability to be manufactured into a CD-sized disc that's durable and cheap.
The Tapestry prototype encodes its photopolymer discs with a succession of 1.3Mb holographic images, yielding a recording medium that can hold 100GB and be read at a high transfer rate of 20MB/s. Seek times, although not disclosed, are estimated to be quite slow--hence the initial target of video applications, where sequential streaming rather than random access is the name of the game. InPhase--which believes it will be shipping Tapestry in volume in 2004--also plans a rewritable version, although no timeframe has been set. InPhase will undoubtedly develop holographic systems for more generalized applications, but the company is wise to target a suitable vertical market first, particularly with so much R&D to amortize.
The first optical drive bearing Nakamura's blue laser technology--dubbed Blu-ray--should arrive in the summer of 2003, courtesy of the CO-based optical storage technology pioneer, Plasmon. The drive's ultra density optical (UDO) discs--measuring the same diameter and thickness as DVDs--promise to hold 30GB. But how deep into the enterprise will this dramatic advancement penetrate? That remains an open question.
The problem is simple: There's a general lack of interest in optical technology among IT managers. "Optical has just lost so much ground," says Gartner Group analyst, Mary Craig. "The optical jukebox market has been declining on the unit level for seven consecutive years." The reason is twofold: On one hand, IDE arrays deliver a much higher performance than optical at a slightly higher cost. On the other hand, tape delivers a much higher capacity at an extremely low price. Optical is getting squeezed between the two. And with the addition of relatively inexpensive serial ATA drives, the squeeze is getting even tighter.
The pressure is on for optical technologists to deliver new, cost-effective, high-performance products as soon as possible. For example, by 2007, Plasmon expects to ship a UDO unit that spins two-sided, 120GB discs with a cost per gigabyte of only 50 cents. After 30 years of promises, an alternative technology known as holographic storage may actually arrive as soon as 2004, serving up 100GB discs out of the gate.
These developments may be welcome, but very likely concomitant advances in RAID and tape will continue to relegate optical to archiving e-mail, checks images, storing transaction records, medical images and other items requiring long-term storage and occasional retrieval. In fact, Enron and WorldCom may be the best thing that's ever happened to optical in quite a long time, because corporations face new pressures to store critical information in formats that can't be tampered with. In the enterprise, the whizzy new Blu-ray and holographic technology will probably be consigned to the mundane--but necessary--task of helping companies meet regulatory requirements.
MO is dead--long live UDO
Currently, magnetic optical (MO) stands as the most successful optical technology in the enterprise. First introduced in the late 1980s, MO is the iron-clad storage medium of choice, with a 30-year shelf life, the ability to withstand a million rewrites and a casing that shields the CD-size disc from dust and scratches. Hewlett-Packard and Plasmon dominate the MO market, with HP looking at licensing Plasmon's UDO technology.
But MO has "hit the wall," according to Gartner's Craig. She says current MO technology can't get beyond the maximum capacity of 9.1GB per cartridge--some of the priciest gigabytes that you'll find anywhere.
With UDO, Plasmon will initially target the MO market, drastically lowering cost while maintaining reliability. Unlike MO, which uses a combination of laser and magnetic storage technology, UDO employs phase change technology similar to that found in ordinary CD-RW drives: A pulsing laser burns dull spots in a reflective crystalline layer, which can be restored to its original reflectivity (i.e., erased) with a beam of lesser intensity. An advantage of phase change over MO is that it supports true write-once capability, which is necessary to meet certain regulatory requirements.
Blue lasers operate at a wavelength of 405 nanometers (nm), enabling them to produce a smaller focal point than the 650 nm red lasers that burn DVDs. That means more data can be packed into tracks and the tracks themselves can be closer together.
Damon Ujvarosy, VP of engineering for Plasmon, faces several challenges in rolling out the first UDO drive. To keep the price of the drive reasonable--and to hit Plasmon's cost per GB target of $2--he still relies on existing technology. For example, for the laser, he chose a lens with a numerical aperture (NA) of .7, rather than the .85 NA stipulated by consumer electronics companies that developed the Blu-ray spec. The lower NA results in a slightly larger focal point, but eliminates the need for spherical aberration correction. This is a lens technology demanded for consumer Blu-ray that has "not really been developed yet," says Ujvarosy.
Another problem lies in the detectors that read laser discs. "Silicon detectors classically are very good at detecting red," says Ujvarosy, "but their responsivity falls off a cliff as you get down around the blue violet region." With such low signal levels, a preamp with extremely low noise characteristics is essential. In addition, the disc surface needs to be free of smudges and dust--which is why Plasmon will be opting for an MO-style protective casing.
Plasmon will offer both read-write and WORM versions of its first UDO drive, which will deliver transfer rates as high as 8MB/s along with a 25 ms access time. The company conservatively estimates that its WORM discs will have a 50-year shelf life.
Stumbling over DVD
Plasmon hopes UDO's relatively low cost will help the technology sell beyond MO customers, extending perhaps as far as near-line enterprise storage. But UDO isn't intended for consumers. That distinction belongs to the forthcoming Blu-ray drives planned by such consumer electronics giants as Phillips, Pioneer, JVC, Toshiba, Sony, Matsushita, Hitachi, Sharp, and others. The first consumer Blu-ray devices should hit the market by 2004, offering a 27GB capacity per disc and a 36Mb/s transfer rate.
With CD and DVD technology, mass production for consumers drove down the cost of components first and enterprise products debuted later. But most observers agree Blu-ray will take three to four years--perhaps longer--to gain popularity among consumers. Why? For one thing, neither recordable nor rewritable DVD has caught on yet among consumers or enterprise buyers, posting a lackluster $169.6 million in sales for 2001. In the consumer space, DVD+RW (DVD plus rewritable) appears to have the best shot, thanks to support for low-cost 4.7GB DVD+RW and DVD-R discs that work in most standard DVD players. DVD-RW (DVD-rewritable) drives cost more than DVD+RW drives and burn discs compatible with fewer DVD players.
But neither DVD+RW nor DVD-RW is on the horizon for enterprise optical. In the enterprise data center, the next most popular optical device after the MO drive is still the 12-inch WORM drive. CD-R has gained some traction but the insurgent technology today is DVD-R, which provides a low-cost alternative to the 12-inch WORM. Pioneer originally developed DVD-R for the desktop, and has been selling its 720 disc, $16,000 DVM-7000 jukebox to enterprises since 1999.
Paul Meyhoefer, the director of sales and marketing for Pioneer USA, Long Beach, CA, doesn't have Blu-ray on his enterprise roadmap yet. "The reason DVD-R is where it is today--both in media costs and in drive pricing and capabilities--is because it was driven by the consumer market," says Meyhoefer. "And if I look at Blu-ray, it's not going to be really driven by the consumer market." Without those economies of scale to drive down the cost of components, Meyhoefer believes the price of Blu-ray will stay high for some time.
Meyhoefer's assertion contradicts common wisdom, which suggests Blu-ray's 27GB capacity is being driven by demand among high-end consumers to fit an entire HDTV movie on a single disc. Blu-ray discs will have the room for that, says Meyhoefer, but don't expect to see a Blu-ray HDTV player in Circuit City anytime soon.
|How ultra-dense optical compares|
Ultra-dense optical disks can store more data by creating smaller pits. That's done in two ways: by using shorter wavelength light and focusing it in a smaller beam than either CD or DVD allows. The numerical aperture number is related to the size of the spot the beam can create--the larger the number, the smaller the spot--and therefore, the more densely data can be packed.
Seeing the light
The resistance to optical in the enterprise seems to be growing, rather than diminishing. JVC national sales manager Doug Arnold, who sells both DVD-R and DVD-RAM jukeboxes, puts it bluntly: "Unfortunately, tape and RAID really have a stranglehold on the enterprise space."
The pace of change in magnetic solutions, even tape, makes optical seem pale by comparison. "Tape is going to 400GB in a year or so," notes Gartner's Mary Craig. "To speed up and have a better appearance of performance, they'll use a RAID front end." In addition, sophisticated tape management systems offered by vendors as StorageTek are improving not only the efficiency of backup, but also the reliability of data recovery.
Bryan Bruton, CTO and founder of online archiving service Scan-Direct Inc., Carrollton, TX, employs optical primarily for clients bound by regulatory requirements. But Bruton uses optical only when he has to, partly because the cost of IDE RAID has already fallen close to that of his DVD-R storage--$7,200 per TB vs. $5,800 per TB. Figure in the added labor to maintain DVD-R, says Bruton, and the delta is more like $500. In a year, he expects it to be a wash. Moreover, his Pioneer DVM-7000 jukeboxes take up roughly three times the physical space per TB as the latest RAID systems.
Optical is a prime example of fascinating technology that doesn't necessarily yield the best solution. As Bruton says, "If you've got to have something non-modifiable, optical is the only choice out there." But for almost any other application, magnetic trumps optical every time--and given ever-increasing hard disk and tape capacities, that dominance isn't likely to change anytime soon, despite advances in optical technology. When asked what would make optical more attractive to him, Bruton quickly replies: "They need to bring the price down."