Will we still buy PCs in 1998?
The `Internet Toaster' may render personal computers obsolete.
New technologies are appearing on the horizon that take aim at the desktop PC market. Some of these same technologies may even change the way we work with computers in the future. In the next two installments of this column we will examine this new frontier and weigh the Internet Toaster's debits and credits. Complete the attached, brief survey on your plans for the Internet Toaster. I'll share the results in next month's column. (2,300 words)
The question in the headline is certainly appearing in the minds of IT managers. The recent gusher of announcements promising compact, $500 devices sporting such names as Web PC, Net PCs, Internet Terminal, and (waggishly) Internet Toaster may herald no less than an authentic paradigm shift in office computing. (Then again, maybe the Internet Toaster is the next video telephone, video on demand, HDTV, Digital Audio Tape, or Newton.) Meanwhile, several companies are making steady, gains in emulation technologies that make virtual PCs alternatives for many users. Either with Toasters or emulation, the question remains "Will we buy PCs in two years?"
This month and next, we'll explore relevant technologies that have the possibility of changing your site from using PC-based desktops to something else. Just what that "something else" might be we hope to discover. This month, we will review developments on the technology side, followed by a product round-ups next month.
`Daddy, why do we have PCs?'
In 1977 Kenneth Olsen, founder and then CEO of Digital Equipment Corp., said no one would ever want a computer on their desk. (Of course, he also called Unix "snake oil.") Was he ever wrong -- or was he? As even non-technical people know, PCs are a huge business. Perhaps a 50 million PCs are in use this minute around the world. On this basis alone Olsen should hang his head in shame. Why did PCs take off? In a word, they are affordable, thanks to abundant software and ever-falling prices for peripherals.
As the PC phenomena matures, however, we're learning PCs make dandy analogs to vacation timeshares. While the average sticker price for a common Compaq or Gateway 2000 is $2,000 to $3,000, industry analyst Gartner Group (Stamford, CT) estimates the added costs of software and maintenance is $8,000 per year over a five-year period! Like vacation timeshares that look like good deals in the beginning, PCs pack a lot of hidden costs. So maybe Olsen was misquoted. Perhaps he really wondered why any IT manager or CFO would want to put a computer on every desk.
What's available now? What's on the horizon?
The technologies that may change the PC market include: the Internet Terminal/Web PC, the virtual PC Unix server, and PC-based software emulation. We can ignore hardware-based emulation technology (such as SunPC) for the desktop (except in demanding cases) in view of their minuscule market share and steep prices.
The Web PC
Although the Web PC, Internet Terminal, or Internet Toaster seem like a new idea, it's been tried before with limited success. The initial Sega Saturn was a cheap, cartridge-driven computer intended for small business use. For various reasons, Sega couldn't sell the Saturn as a business tool, and quickly repackage the Saturn as a home video game player. (Ironically, last month Sega said it would start selling Internet Saturn, a Web-browser, in Japan for the equivalent of US$190, and may sell a more expensive version in the United States later this year.)
In the past three or so years, another potential market in the cable TV industry, led by broadcasting and distribution giants such as Turner Broadcasting and Tele-Communications, Inc. (TCI) and consumer electronics leaders like Sony and NEC, has resulted in the development of set-top boxes which act as data receivers, transmitters and essentially small computer systems. It turned out that 1995 would not be the year of the TV set-top box boom. Although vendors are still pursuing this market actively, the jury is still out on when this technology will hit the common household. So a number of those vendors have also gone with the latest trend towards the Web PC.
Web PC hype follows the impressive draft left by Java, a clever programming language that's made James Gosling a household name (in Silicon Valley anyway). When a friend of mine told me about Java in December 1994, I was pretty skeptical about it as a functional distributed computing concept. By July 1995, I was raving about it to my friends, too. Java focuses on a method whereby users can write hardware- and OS-independent code. Imagine: non-proprietary cross-platform software development requiring no porting or recompilation.
The Toaster is conceived as a small computer with a low-powered processor, a smidgen of memory, little or no local storage. Its simple operating system would have just enough smarts to load a browser and handle the housekeeping chores of process scheduling, memory allocation and control, hardware resource control, and network interfacing. Imagine: appliance-like desktop computers.
Apps are stored on the server and delivered on demand to the Toaster, which discards them when the user moves on to other tasks. Imagine: paying only for applications used, and never having to worry about user-app version control again.
Keep in mind Java is not an operating system. As a move away from monolithic systems, the microkernel OS concept is almost ideal for a Toaster. The word microkernel however, does not mean the system requires a small number of bytes of memory to run. Windows NT, for example is a microkernel-based operating system that requires at least 16 megabytes. However, microkernels can be small. For example, QNX is a real-time microkernel Unix with a core size of 7K to 11K. In the Toaster, the microkernel must also be this small because of the intended pricing. A small core means more memory available for applications.
One level up from the Toaster OS is Java, which can interpret or run the Java applications. The Toaster runs compiled bytecode in the smaller models, and in larger models can compile or interpret them from source when necessary. The difference is in the system libraries that have to be included. The Java library of object classes is extensive but pedestrian from an experienced OO programmer's view. Libraries must be kept available somewhere, usually in ROM as compiled objects.
Likely Web PCs include the yet unnamed Java terminals from Sun, the Network-PC from Oracle, a set-top/Net PC from Microsoft, MIPS-based systems from Silicon Graphics and NEC, the Interpersonal Computer from IBM, and offerings from several Japanese computer makers.
As proposed, the basic Toaster uses a small processor, such as a lightweight PowerPC, SPARC, Intel, or R3000. Sun stands out amongst all in its choice of announcing a family of chips that execute Java bytecodes at the processor level directly. NEC, which developed the VR4300 as the brain behind the new Nintendo Ultra-64 (imagine, 64-bit processing for home video games), and LSI Logic, an ASIC maker for the video game industry, may become prime Toaster chip makers as well. Both vendors have created powerful (greater than 100 MIPS), yet cheap (from $30 to $60) processors and ASICs that may compete with the picoJAVA, microJAVA, and ultraJAVA processors from Sun that inhabit a similar price range ($25 to $100).
Here's the typical formula for a Web PC:
Web PC = Processor + 8 MB RAM + some ROM + modem/network ASIC + misc. control logic
With current costs for these components, it seems likely someone somewhere will find a way to manufacture these devices for $200 to $300 each, making a $500 computer (plus the price of a monitor) possible.
The question is: Do you get what you pay for? Or do these new technologies offer more for less? Proponents are counting on the "Field of Dreams" scenario: "If you build it, they will come." But there is more than faith in this electronic corn field. Real market trends include the growth of SOHO (small office, home office) Internet use, and Intranets in big organizations.
It is possible to write applications on the level of complex statistical analyses packages, databases, and so forth with Java. With the product itself is barely a year old, companies are lining up by the ark-load to develop new applications or convert current ones to this language.
Java has a perceptual (for now) competitor in Microsoft Visual BASIC (VB), which Microsoft is tarting-up with a pre-announced scripting language. Microsoft's Fear, Uncertainty, and Doubt machine predicts mayhem for Java when VB script is incorporated into the Microsoft Web PC. Of course, Microsoft is fighting to hang on to its developers and perceptual rule over desktop computing. Stay tuned, Act I will tell the hardware tale. But it's Act II, software, where the Internet Toaster adventure really begins.
The Virtual PC server
Then there are the emulators: SoftWindows and NTrigue from Insignia Solutions, WabiServer from SunSoft, Macintosh Application Environment (MAE) from Apple, and WinDD from Tektronix. Collectively, this bunch is pretty dull when compared to Java's drama and Microsoft's battle for survival. But then, these are proven products you can use today. (No wonder no one writes about them anymore.)
Emulators allow you to do away with your PC entirely and run your apps from a server system. These products provide a complete PC system at the software and sometimes even hardware level to bring near-perfect compatibility. Instead of requiring a PC at each desktop, all you need are X Window System-capable systems, and mid-range server to provide the raw processing power.
Emulators come in two flavors: those based on Unix servers and those based upon modified NT servers. SoftWindows, WabiServer, and MAE allow you to start a virtual PC anywhere on an X Window System network emanating from a Sun, Silicon Graphics, HP, or sometimes Digital Unix server. All processing is accomplished by either binary translation or emulation.
NTrigue and WinDD must installed on mid-range Intel-based server systems. Obviously, since the machines themselves are Intel-based, there is no need for hardware emulation. The converted NT 3.x Server operating system allows the creation of virtual NT desktops through X Window System or similar protocols.
Though served from different platforms the two approaches have much in common, Both require you have an X11 product installed on your desktop computer because they translate X Window System calls into MS-Windows or MacOS events that can be processed by the application locally or sent over the network to be processed by the remote server. Each virtual PC is usually a full desktop similar or exactly the same as what a normal PC would look like. They have either built-in license managers or use those available to them to control who has access. They have monitoring systems so that the network manager can keep track of which stations are using the products and sometimes even what applications they run. They use the networking and storage features of the server system and multiplex data across all the virtual PCs.
Each of these virtual PC servers eliminate the need for a multiple physical desktop computers at your station. Engineers or users who are based primarily on the Unix platform can save money and desktop real-estate by having only an X Window System terminal and all the functionality of both PCs and Unix systems.
Needless to say, major X terminal vendors like HP, NCD, HDS, and Tektronix, love this concept, since the focus of the X Window System desktop market has shifted slightly away from X terminals to PC-X server software, which is price competitive. There are real concerns on the performance issues but unless your shop is focused on a lot of PC activity, there is a way to reduce a lot of cost on desktop system management.
What else is there?
Next month we will complete the round-up of some of these new technological developments and see how they impact the current situation economically and technologically. We will also look at other non-technological factors, such as system maintenance, which may affect this industry. To balance both sides of the issue we will also talk about developments that suggest PCs are more likely to be around in the future.
About the author
Rawn Shah is vice president of RTD Systems & Networking, Inc., a Tucson, Arizona based network consultancy and integrator. Reach Rawn at firstname.lastname@example.org.
If you have technical problems with this magazine, contact email@example.com