| Chart Watch | Departments | MPRonline | More Research | MDRonline | Press | Sales | About Us | InStat.com | MPR Home  |

August 27, 2007

1TB

By Max Baron

Max

1TB equals 1,000GB equals 1,000,000MB. Very high capacity hard drives are becoming affordable. Like the introduction of very high performance processors that led to realistic games on the PC, the appearance of the oversized hard drive may usher in new applications.

It may change the desktop and notebook. It may change people's personal preferences in reading books and viewing movies.

New applications can be stimulated by the confluence of new and old products that make them feasible. Partnered with multicore desktops and high-quality monitors, the voluminous hard drive can foster change by following a path that has been traveled before.

It's interesting, (probably Psychology 101) that sometimes you are not ready to think about an application until you hold in your hand the tool that can make it happen. I bought a 1TB hard drive. An impulse buy. "What will I do with it?" I was asking myself as I made my way out of the local Fry's Electronics, reached my car, and strapped the 1TB external hard drive to the back seat of the vehicle. Earlier in the week, I could have purchased two separate hard drives made by AcomData, providing me with the same amount of storage for $218. However, after reading a few horror reviews on the Internet, I decided to go instead for the $299 sale-priced 1TB device offered by Maxtor (now Seagate). I drove back to my home slowly, forcing myself to focus my attention to the surrounding traffic. I was thinking of 150KB-sized magnetic drums employed by the supercomputer of 40–50 years ago, such as Ferranti built for the Atlas computer in Manchester, England. I was comparing, for the sake of fairness, only one of the two internal disks making up the Maxtor 1TB with one of the drums employed by the Atlas. In approximately 45 years, single-disk capacities have increased by 3.3x106! The increase would be 6.6x106 if one compared Ferranti's with the 1TB Hitachi five-platter single-drive Deskstar 7K1000. The Atlas's main memory size was approximately 100KB—smaller by 4x104 than the memory I had in my home computer. And, ignoring word width at 48 bits for the Atlas and 64 bits for my processor, I compared the 2GHz clock frequency of my slow computer with Atlas's 1MHz to find an improvement of only 2x103. (Atlas was, if not the first, at least one of the first computers employing transistors.) Faster computers than mine could claim a frequency improvement as high as 5x103, but it would fall short of matching the orders of magnitude shown by main memory and hard drive. "Has the frequency corollary1 to Moore's Law failed long ago?" I asked myself, avoiding a car that invaded my lane "Processors are trailing memory and bulk storage!"

The answer, I think, isn't a categorical "yes," nor is it a definite "no." It depends on the way we use the processor and its memory, be it main, hard-drive, networked, or connected by any other communication model we can envision. What we want or need doesn't dictate technology. We tend to find ways to use what is available and offered to us. In doing that, we bias research to improve the technologies we embrace. Not exactly an ideal process.

The frequency disparity is acceptable because we are beginning to use disk drives as libraries. These are becoming far larger than the programs, text, images, or music we can use at any one time. My purchase was justified: I had acquired my first library.

At home, I placed the drive reverently on my desk beside my computer. I marveled at the power of Maxtor's marketing people that probably decided the hard drive's profile should be shaped like an I-beam "to show strength," but, in my opinion, lose stacking capability. Marketing may also have decided that the indicator light should come from powerful white LEDs, a beautiful differentiator, except when the disk is connected to a computer—the intense light starts to pulse, distracting you from work. Relax Maxtor, I bought you because you were offered at a sale.

As large hard drives are entering commodity status, Maxtor and AcomData—and, for all I know, others—are offering more differentiating "benefits," such as programs to help you back up your data, encrypt it, or find it faster. Maxtor practically rams those programs down your throat—you have to install them to connect to its I-beam-shaped external hard drive. A program seems to be there to help format your drive, because it comes preformatted for the Mac! Worse, the label "Maxtor" appeared on my taskbar (PC XP), competing with useful programs that I need to run. Double worse: a backup program I don't need also occupied part of the computer memory (of course), slowing the system. But, contrary to the immortal phrase in the eternal play, I didn't come here to criticize the otherwise good Maxtor but to praise the appearance of these very large hard drives.

Where will giant personal-storage resources take us consumers? Many business analysts will probably answer the question by forecasting large increases in sales. Some of the forecasts may be based on optimistic projections provided by hard-drive marketing managers. To play it safe, many forecasters will not describe new applications and the way they may affect sales of a product. I've seen an example in EE Times: projections concerning the rise of hard-drive utilization in automobiles—16 million units by 2013 per iSuppli's estimate. Better than CDs or DVDs for use in GPS, says the research company, and good for downloading multimedia for entertainment from the Internet via different means of communication, all obviously wireless. But don't we think that with terabyte resources, consumers will look for additional applications in their vehicles?

Not trying to provide business forecasts, or play it safe, we can attempt to put together a technically plausible vision of the near-term future. We can ask what applications would be possible, given multicore processors, gargantuan hard drives, security and digital rights management (DRM), the Internet, the cellphone, iPod, and, most important, human nature. The list of applications could be endless except for the concept making the large hard drive our library. Carried out successfully, it may change the way we read books and watch movies.

I can explain. Allow me to use as an example the white miracle: Apple's iPod. The iPod is not a revolution; it's the result of a rather long evolutionary process that started many years ago with the introduction of turntables and single-song records. The process took us through the creation and evolution of multiple technologies: the album; the cassette and living-room cassette player; Sony's Walkman—the first good way to take some of the songs with us; the CD and player that allowed us to bring along high-quality music; the MPEG-3 CD's short life (more songs but still large and heavy); the flash-memory-based player (pocketable, but too few songs); and today, but not at the end of the road, the little white box that can hold more than 50 days of songs.

The appearance of iPod player-amplifiers for the living room is closing the loop. It creates a new and more efficient user interface at home: selection of songs from a menu. It also defines a resource where all or most of our songs can be stored in one place.

I won't try to describe in detail the evolution that brought us to where we are today in books and movies, but if we start with prehistoric drawings on cave walls and stones, and return to our present books and DVDs, we find the same pattern we observed in songs. We are always looking for smaller and lighter devices. We want to take them with us on trips. We like instant and effortless access in the home. Proof: the remote control. Yet, despite our preferences, today's books and movies are frozen, still at the turntable and record stage. Some of us may have thousands of dollars' worth of books that we can't search through for technical information or a particularly beautiful paragraph we found in a novel. We buy movie DVDs because we want to watch them again sometime, but do we want to wade through hundreds of movies we own looking for the one we want to enjoy? And how do we access the scene that brought tears to our eyes or made us smile? The Internet is giving us click-access to the world, yet we have to go to the other room to get a book or load a movie into the player. The Gutenberg Project is an exception (http://www.gutenberg.org/wiki/Main_Page). The project's online book catalog offers, free of charge, more than 20,000 older books whose copyright has not been renewed. Boole, Riemann, and Einstein alongside Shakespeare, Voltaire, and Jules Verne. Today's contemporary personal library, however, is a throwback to the dark, solemn, brown-furnished room, book-lined from floor to ceiling, that we see in period movies. The library now also includes DVD cases, a bit slimmer than the almost forgotten video cassette.

The problem is not limited to the living room. With so much storage space available to us, why do we have to bring a basketful of books, CDs, and DVDs to the poolside, or to our annual vacation? And, if the player of our choosing accepts a particular compression format, where do we get the content to match it?

Of course, we have to remember that movies can be purchased as DVD (MPEG-2) for the home or special portable player; as MPEG-4 and H.264 (MPEG-4 part 10) in formats and sizes suitable for Sony's PlayStation Portable (PSP) and Apple's Video iPod; and as 3GPP (3rd Generation Partnership Project) for some cellphones. But who wants to pay $100 and up for the pleasure of owning a particular movie in various formats, suitable for different players? And what will happen to our movies upon the appearance of devices employing advanced codecs we don't have today, with DRM preventing us from transcoding? Or, if the DRM-protected movie original we have today "breaks" digitally or physically? How can we reliably store or obtain a copy of what is rightfully ours?

As for the book: it's further back in the dark ages. Take five or six books on a trip, and some airlines in Europe will not allow you the added weight to your carry-on. Why can't book publishers at least offer books on the same type of protected media movie creators use?

A 1TB hard drive will hold the equivalent of 250 movies on MPEG-2 DVDs. Its cost today will add only one dollar and change to the price of a DVD. The 80GB Video iPod will hold about 140 full-length movies in MPEG-4 format; a 200GB DRM-protected external hard drive for notebooks could double that quantity, with capacity left over for a few programs. Or it could accommodate 50 MPEG-2 DVD-quality movies. Books have a far smaller footprint: about 5MB would be required for 800 pages of text compressed via Adobe Acrobat. Adding grayscale block diagrams to the book may demand an additional 5MB.

If you think that streaming movies over the Internet is an alternative, consider, aside from low quality, the availability of content defined daily by a vendor's tastes or business interests and the recurring cost. Go to www.netflix.com and see how many movies you like from the 4,000 available from the website. And check out the requirement to place in your computer an ID number for DRM purposes. A recent survey from researcher Parks Associates speaks of consumers' dissatisfaction with videos available over the Internet. There seem to be only 16% or fewer respondents who are satisfied by the available selection and the prices charged for the service—more than 84% are waiting for better solutions.

I believe the high-capacity hard drive can do for books and movies at home what the iPod has done for music. I'm thinking of a server at home or a secure area in a computer that can implement the book or movie physical-object model: the object you purchased can be played locally or taken on a trip via copying or transcoding—but it can exist in only one instantiation, irrespective of format. If you take it with you, it can't be played at home. The movie can be stored on the server disk in the quality and format in which it was originally purchased. Disk-speed-limited but very fast transcoding, executed on multicores, can "remove" the movie by negating access to it at home, and, for example, placing it temporarily in your DRM-compatible iPod. Erasing the movie from the iPod will restore access to the original copy in your server. We can create a digital implementation of removing a movie or book from the library and returning it. Sooner or later, a device of this kind will have to appear to continue our evolution toward small, fast means of delivery and the use of quality content. The library and computer can be smaller than your subwoofer.

Daydreaming or soon to become reality? Time will tell. But the multicore-based systems are no longer program independent. The "serial" high-performance program execution that we used to scale with frequency is gone, irrespective of what we are told. Today's combinations of architecture and software will only speed up portions of programs that can offer fairly long threads. The threads may be few, and the speedup will "feel different." Program sections that can't be multithreaded may be slower—expect a core that's part of a dual or quad configuration to be clocked at a lower frequency than a single-core chip.

Data-intensive processing, such as video and audio, is the new winner. It thrives on parallel configurations. The multicore processor that scales its performance by adding cores has found an apt partner in the hard drive that scales its capacity by adding tracks and disks. Together they can offer unparalleled performance on these workloads that can be distributed across many engines.

And the fate of books and movies will continue to be at the mercy of content creators, unless we design the high-performance secure systems they can't refuse.

MB_Signature

References

1. Silicon semiconductor technology has been common for microprocessors and memory for 25 years. A 30% reduction in linear dimensions results in approximately 50% area, doubling the number of transistors that can be designed into the reference area. The reference area (high volume MPU: 280mm2 in 2007) is no longer a constant (see http://www.itrs.net/Links/2006Update/FinalToPost/00_ExecSum2006Update.pdf). A 30% reduction in device delay provides a 50% increase in speed.

Editor's note: The author gratefully acknowledges contributions by Dr. Don Alpert, Professor Mike Flynn, and Dr. Don Gaubatz.

Most Recent Articles

  | MPR Home  | Chart Watch | Departments | Events | More Research | MDRonline | Press | Sales | About Us | In-Stat.com |
Copyright © 2007 In-Stat
A Unit of Reed Business Information, A Division of Reed Elsevier, Inc.
Read our Privacy Statement.