Based on what's reported in the mainstream computing press, you might think that the single most important issue facing humankind is the "last mile" problem: the challenge of finding the right infrastructure to deliver high-bandwidth data access (usually exemplified by the Internet) to individuals and organizations. Usually, I find the battles between the cable companies and the "telcos" on this issue tedious and better left to the heads-in-the-clouds staff at Wired. The case of Teledesic, however, is much more interesting if for no other reason than it threatens to cleave the Gordian knot of ISDN, T1, ADSL, cable, and VSAT like a hot knife through so much room-temperature butter.
Teledesic, which was founded by Bill Gates and cellular pioneer Craig McCaw, each of whom have made a personal investment to the tune of $10 million, is developing a global, geodesic constellation of low Earth orbiting (LEO) satellites exclusively targeted at computer networks. Under manufacturing and launch contract to Boeing, Teledesic is designed to deliver data communications at downlink speeds of 64Mbps--2,000 times faster than analog modems--to 95 percent of the Earth's surface and nearly 100 percent of the world's population. Talk about universal service: The more I hear about Teledesic, the more I think that Microsoft's investments in cable, asynchronous transfer mode, and ADSL technologies are really just something to keep the company busy in the media business while Teledesic is being built. The first satellite launch is scheduled for 2001, with the network going online in 2002.
Teledesic's ostensible mission is to deliver high-bandwidth data services--video on demand, high-resolution image transmission, and so on--to places where there is no high-bandwidth infrastructure (and even where there is), and to do so for an affordable price. (You may have heard of Motorola's Iridium, an LEO system projected for operation in 1999. However, unlike Teledesic, Iridium focuses exclusively on single-user voice and fax communications.) But Gates and McCaw's ultimate goal is to own the plumbing for global telemedicine, teleconferencing, direct banking, e-commerce, and many other applications no one's even thought of yet.
Furthermore, there's more to Teledesic than multimedia content. Because conventional geosynchronous Earth orbiting (GEO) satellites are stationed so far up in the sky, any applications sensitive to latency--such as banking transactions--have been out of the question. LEO solves that problem by moving the satellites much closer to the Earth, reducing latency to milliseconds and making even a satellite-based OLTP system much more realistic.
The IT angle here is that if Teledesic is successful, the word "distributed" will imply much larger distances. Theoretically, organizations will be able to locate their employees and facilities with complete disregard for terrestrial communications infrastructure. Should this scenario come to pass, the impact on database architectures will be profound. With the network no longer being the bottleneck and the problem of latency solved, we may see the massive centralized (and remotely accessed) database become dominant and distributed architectures fade in appeal. Conversely, we could see a proliferation of specialized repositories organized as a distributed database with strong parallel (and one would expect, object/relational) functionality. Or perhaps by 2002, an entirely new architecture will emerge that offers the best of both worlds. In any event, Teledesic has the potential to change the nature of communication and information management completely.
Perhaps the most interesting thing about Teledesic, however, is how it serves as a textbook example on a grand scale of Gates's consistent pursuit of what George Gilder calls the Law of the Microcosm: the decentralization of information architectures. For example, in Teledesic, system responsibilities will be distributed across the entire network to reduce operating requirements for each satellite. (Gilder calls Teledesic "one of the world's largest and most expensive massively parallel computers.") Furthermore, Boeing will build the network's 288 satellites using off-the-shelf components. Sounds suspiciously like an NT cluster in the sky, doesn't it?
Before we look forward, however, we have to carefully consider where we are
now. To that end, this
issue of Database Programming & Design provides some "signposts" about technologies such as object/
relational databases (Seth Grimes's "Object/Relational Reality Check,"),
data warehousing (John Giles's "Is Data Warehousing Only First Aid?"), and cluster computing
(Stephen Brobst's "Safety in Numbers"). I hope you'll finish this issue with a better, more
objective understanding of these issues than what's typically offered by vendors and the rest of the trade
press. Enjoy!