|
Magnetic recording has memorable future
Frustrated that callers couldn't leave a message when someone wasn't available to answer the phone, Poulsen began tinkering. He soon discovered that he could record sound by speaking into a microphone attached to an electromagnet that he moved slowly along a length of piano wire. The process left a magnetic trail of sound patterns. It was the birth of magnetic recording, which he patented 100 years ago this month, and it laid a foundation for the creation of the audio, video and computer storage industries. To celebrate this inventor and his pivotal discovery, nearly 200 of Poulsen's intellectual and economic descendants gathered at Santa Clara University last week to marvel at the long-term impact of Poulsen's magnetic tinkering. Without magnetic storage, the conference speakers noted, there would be no Internet, no personal computers, no databases and no gigabytes, terabytes and exabytes of data upon which computers now compute. ``In 1855, it cost five cents to send one word from Philadelphia to St. Louis. In 1998, it costs 4.5 cents to store 1.5 million bytes of information that can be sent anywhere in the world with the click of a mouse, virtually for free,'' said Jim Koch, director of Santa Clara University's Center for Science, Technology and Society. ``Without Poulsen's discovery, the advances we've enjoyed in electronic computing would not have been possible.'' As researchers continue to cram more and more bits of magnetic data into tighter and tighter spaces, magnetic storage will help quicken the pace of new product development and may help spawn new industries. Most experts believe magnetic storage will be a key technology for another century yet, although it is likely to require an assist from other innovations as researchers begin to bump up against its physical limits. Futurists now are talking about magnetic storage contributing to ``ubiquitous computing.'' They envision a world where the greatest computing power resides not on our desktops, but in devices surrounding us in our homes, offices, cars and even clothing. The origins of this magnetic storage really began with the theories of mechanical engineer Oberlin Smith, who in 1878 proposed an improvement on Thomas Edison's phonograph. He never tried to patent any devices based on his theories, however, and instead published his ideas for others to use, ``hoping that (his theories might provide) a germ of good from which something useful may grow.'' While Poulsen's Telegraphone answering machine thrilled inventors during its debut at the 1899 Paris world exhibition, the device's poor sound quality doomed it in the marketplace. It took more than three decades before other magnetic recorders achieved modest commercial success and five decades before Poulsen's innovation would become a foundation for the modern Information Age. The first magnetic storage device for a computer was a rapidly spinning magnetic drum developed in 1948 by a team of researchers at the University of California-Berkeley, said Al Hoagland, director of the Institute for Information Storage Technology in Santa Clara. An array of magnetic heads positioned just above the drum's face could feed in bits of data and read them out again at what was, at the time, considered lightning-fast speeds. The Berkeley drum -- two feet long and eight inches in diameter -- had a capacity of 10,000 ``words'' of 10 digits each, Hoagland said. A meager 800 bits of data fit on each square inch of the drum's surface. Initially, this drum technology was fast enough to feed data directly into the central processing units of the early mainframe computers. Soon, though, computers began to need data faster than a magnetic drum could provide it. In 1951, the Univac computer became the first to feature a magnetic tape storage system. Data could be fed quickly -- if only sequentially -- into the system. Different reels of tape could be dedicated to storing separate categories of information. But computers couldn't always handle records in a predetermined order. And it was time-consuming and inconvenient to have to record an entire reel of tape all over again every time a single section of it was modified. Computers would be more productive if they could read and change files at random, companies told computer scientists. American Airlines wanted a way to access and quickly change individual records in a reservations system. The U.S. Air Force wanted a better way to control its inventory. What they wanted was a hard disk drive. In 1957, IBM's RAMAC, or Random Access Method of Accounting and Control, used 50 24-inch spinning disks to store what was for the time a staggering amount of data -- 5 megabytes, or 5 million bytes of information. Its read/write heads, which moved between various tracks on the disks, used the first air bearing to maintain an even distance from the disk's surface. Pressurized air was forced out of tiny nozzles surrounding the magnetic head, forcing it to float above the disk's surface, Hoagland said. Production of the RAMAC was ``not exactly high tech,'' said Al Shugart, founder of Seagate Technology Inc. and a member of the RAMAC development team. Disks were made by pouring iron oxide paint from a Dixie cup onto a spinning platter. It never occurred to anyone to use a clean room to keep dust off the device. The RAMAC was enormous -- it stood nearly six feet tall and covered the wall of a small room -- and expensive. IBM preferred not to sell it -- the company wanted to introduce an improved system quickly -- instead agreeing to lease it to customers for $750 per month. Customers had to pay $150 more each month for the air compressor. The industry has advanced rapidly in the five decades since Berkeley's researchers manipulated 800 bits in a square inch. Today's most-advanced hard disk drives can cram more than 5 billion bits of data -- five gigabits -- into that same square inch. Scientists in the laboratory have pushed disk capacity to more than 12 gigabits per square inch; they expect that hard drives at that capacity should be available to consumers within two years. By 2001, hard drives should have more than 20 gigabits of data in each square inch of space, a density once thought impossible because of a physical phenomenon called the superparamagnetic limit. Eventually, researchers say, they are going to start cramming so much data so closely together on hard drives that the minuscule bits of magnetized metal will start to ``forget.'' A bit that was supposed to flip one way will flip the other. A bit nearby will flop the wrong way, too. A whole chain of data bits will flip and flop and finally fall into utter magnetic chaos. This moment of physical truth -- this superparamagnetic limit -- can be dodged for several more years, but not forever. Researchers had thought they would be bumping up against it somewhere between 20 gigabits and 40 gigabits per square inch. But researchers have continued to pull physical tricks from their hats to push the limit well above that. The first dodge they are trying involves a fundamental change in the way bits are stored on a hard disk. Right now, each bit consists of a collection of metallic grains magnetized as a group to flip one way or another. As researchers have shrunk each bit to fit more in a given space, they have had to shrink the metallic grains, too. But if the grains are made too small -- perhaps only a few atoms wide -- they become unable to hold the magnetic position required of them. So researchers are working on ways to make a bit from a discrete particle of material, instead of a group of particles. That allows them to use larger grains of material, but it requires some tough engineering to etch microscopic patterns in the disk. The approach has been demonstrated in the lab, said John Best, director of IBM's Almaden Research Center, and should push the superparamagnetic limit up around 100 gigabits. What would a consumer do with so much data in such a tiny space? Not to be flippant, says IBM Fellow Dave Thompson, but the only limit is the imagination. ``One hundred years is 3 gigaseconds, or 3 billion seconds,'' he said. ``Today's 30 gigabyte disk drive would allow you to store 10 bytes a second for 100 years. If we got (an increase in storage capacity by) a factor of a million -- 10 megabytes per second -- that's more than you need for high-definition TV. You could have a diary that stores everything you could see in your entire life.'' Thompson believes that tomorrow's business person will carry just such a ``memory prosthesis'' or ``prompting diary'' that will record everything that person sees, hears and does over the course of a day. At night, the user would plug the device into a central computer that would analyze and compress images, create text transcripts of conversations and generate keywords to make the day's data easily accessible. ``In the future, when you see a guy muttering on the other side of the room, it's not because he's senile. He's talking to his prompting diary and it's prompting back,'' Thompson said. The user could ask his diary ``Who is that person who just walked in?'' and the diary would tell him both who it is and when and where they met. It could even play back audio snippets of their last conversation, Thompson said. While such a technological feat would require tremendous computing power, there would be nothing for the computer to crunch without vast amounts of magnetic storage. Any initial resistance to such a device -- think about the privacy ramifications of recording everything that happens to you in a day -- should dissolve as early adopters begin to demonstrate the advantages of carrying a photographic memory, he predicts. ``So many people focus on accessing and searching the public database, the Web. But your own private database is more important than the public one,'' Thompson said. ``There won't be any reason ever to forget anything anymore. Ultimately, how can you afford to be the only guy in a meeting who forgets things?''
|
|||
|
|||