I have good news and bad news. Recordable DVDs are about to double their capacity, with a further sixfold increase around the corner. Is that the good news or the bad? In my opinion, its both.
Ill start with the good news. Current DVD technology is underexploited by first-generation recorders, which use only one of the two recording layers that current DVDs provide. Internal dual-layer burners, already priced below $300 and falling quickly, store 8.5GB at a media cost thats heading down toward $10 per disk.
Dual-layers time in the mainstream will be brief, though, because the next step in DVD technology is already here in the form of the multivendor Blu-ray standard. Using a shorter laser wavelength to pack bits more closely together, Blu-ray Discs store up to 50GB each at prices that will soon be comparable to DVDs.
Sony led the way last year with a 23GB Blu-ray recorder. Matsushita, under the Panasonic brand, followed late last month with a 50GB recorder whose capacity Sony will soon match, and other vendors have their competing units in the works. Blu-ray recorders currently sell for well above $2,000, but their prices will drop quickly in the next two years as integrated, multiwavelength optical heads reduce device complexity.
Most discussion of recordable DVDs revolves around HDTV (high-definition television)-quality movies, for which I am not the target market. No one in the TV business wants to admit this, but tests show that people find conventional TV pictures—when fed with equal signal quality—subjectively sharper than HDTV pictures on any screen smaller than 24 inches.
If you have a big-screen home theater that needs HDTV to look good, thank you for helping to drive the volume and cut the cost of what I actually do want: an overdue means of affordable backup for tens or hundreds of gigabytes of mass storage. My current backup discipline, using double sets of CDs stored at separate locations, is cheap and reliable but far from convenient when I need to find a particular file.
This brings us to the bad news: When were finally able to store tens or hundreds of gigabytes on low-cost removable media, this will merely put the equation of data availability out of balance in a different and more difficult direction than it is today. Well find that we lack the indexing concepts to find things based on anything but primitive criteria, despite Apples and Microsofts respective “Tiger” and “Longhorn” visions of automatic background classification of our files.
The indexing problem that first comes to mind is images. Its trivial to sort pictures by criteria such as date, and it will soon be easy to tag them with locations using Enhanced 911 or GPS (Global Positioning System) technologies, but indexing the content of the images themselves is still very much a research topic.
The general-case image recognition problem, though, isnt all that important in most enterprise applications. Like speech recognition, current image recognition methods work pretty well when a limited number of things need to be recognized in predictable situations.
Its more difficult to index into a video stream and query it for details such as the times when a particular face appears. Ive seen one IBM project, however, that was highly successful in recognizing voices in the soundtrack and using this ability to locate the associated video frames. Perhaps would-be airplane passengers should be asked to look into a camera and speak a standard phrase. IBMs results suggest that this would be far more difficult to fool than face recognition alone.
But this just scratches the surface of nontext data, which also encompasses anything from music to fingerprints. Enterprise systems are getting amazingly good at capturing such things, and there are many applications for multimedia databases in enterprise sectors as diverse as retail, public safety and health care. The problem is that were much, much better at accumulating data than using it.
Enterprise IT buyers must get ahead of the curve in working with application designers and researchers to identify and address future data-retrieval needs. Without that effort, continued reduction of raw storage cost per byte will prove to be a bargain that we cant afford.
Technology Editor Peter Coffee can be reached at [email protected].