As IBM Corp launches its Network Station derived from the Network Computing Devices Inc Explora, we could not help wondering just what Big Blue might have done if it decided to build rather than buy the box. Network Stations will be unable to perform some tasks for which personal computers are used unless they are tied to a server that gives them personal computer capabilities. If a Network Station has sufficient memory and processing power to run productivity applications, all that is required is a download taking several seconds. On the other hand, if the application cannot be executed by the Network Station, it will require support on a network server. A Windows NT server can support Network Computing Explora workstations using the extensions to NT, making each terminal look like an NT workstation and enabling users to run any NT application. Under those conditions, a Pentium server can support 10 intensive users on Exploras and a Pentium Pro 200 can support twice that many. In cases where the productivity application usage is light, 100 or more Network Stations can be supported by a four-processor Pentium Pro-based server. In addition, where Network Stations run off NT servers, each user of office applications requires 4M b to 5Mb of main memory. On a 100-user local network with one-third of the users getting server support, the quad P6 server would have to be configured with 256Mb of main memory. Again, if all the applications are available in a form that runs in th e Network Station, the server burden will be considerably less.

Green screen emulation

Thus, the data suggests that Network Stations may be less costly than new, fast personal computers, but, if they require NT servers to support them, not dramatically so. If the Network Stations are marketed primarily as boxes that can perform green screen emulation and net browsing using their internal processors, they look to be less costly to buy than personal computers and much less expensive to administer. But again, the concept raises a management issue. A personal computer equipped with only an operating system, 10Base-T local network (or coax or twinax), a terminal emulator program and a Web browser is not very complex or expensive to own or to manage. Moreover, it can be upgraded to support other applications if it is assigned to a different class of user or remarketed. Finally, the server support required of a personal computer can be minimal. Our guess is that the big market for Network Stations will be at AS/400 and mainframe sites that want to upgrade green screens into intranet terminals and also give them the same look and feel as personal computers. That looks like a niche market, not a mass market to us and one where the production economies of the personal computer business will keep the price differential be tween Network Stations and personal computers very small. IBM must also live with the possibility that its key supplier will be unable to survive as a specialized, independent company. Network Computing Devices, struggling to survive as the X-terminal market in which it was the leader dwindles, has been casting about for ways to apply its knowledge to new applications.

It was AT&T Corp’s partner in the ill-fated Mariner network appliance venture and it has more recently switched its emphasis from Unix to Windows NT. Network Computing’s revenue so far this year is off 18% to just under $60m and its losses exceed $6 m. The company lost $4m in 1995 and $10.8m in 1994. Network Computing had a net worth of $63.5m at mid-year and only about 40% of that was liquid. If IBM buys a half million Network Station boxes from Network Computing at an average price of $250, t he company’s revenue could double and, presumably, it would be back in the black. But even if that market materializes, it must do so quickly. If the Network Station business does not take off until mid-1997, when Phase II boxes reach customers, and if the other recent trends at Network Computing continue, IBM’s key supplier may find itself in great difficulty. If Network Computing pares itself down to gain staying power, it will become totally dependent on the success of the Explora family products and on IBM. This is not a particularly good basis for retaining the customer confidence and talent it needs to preserve the technical leadership it currently enjoys. Moreoever, if Network Computing Devices fails, IBM may have to bail it out, but we doubt it could buy it. The software licensing deals on which Network Computings products depend, including those involving Microsoft, almost certainly include shutdowns if Network Computing fails or is acquired by another entity. IBM’s other choice might have been a carefully tailored personal computer based on Intel Corp microprocessor technology, because it has in its possession an operating system for Intel chips, OS/2, plus Notes and other Lotus applications and access to Netscape C ommunications Corp’s browser. That is, it could have kept Microsoft Corp out of the equation. IBM also has (or is believed to have had) alternative foundation technologies, such as a partially completed version of OS/2 for the PowerPC and a variation of the PowerPC chip, the 615, capable of supporting an iAPX-86 code version of OS/2 while the native PowerPC version was completed. There is no question that even a minimal implementation of OS/2 would be far larger than the operating environment code inside an Explora. This suggests that a Network Station might need a significant amount of costly local storage. But this local storage does not have to be relatively expensive RAM: it could have been provided by some combination of Flash memor y, read-only memory and a high capacity removable disk.

All of these add-ons

Further, if the Network Station were made to run on a sufficiently fast local network – 100Base-T rather than the 10Base-T the IBM-Network Computing Devices product will at first support – the desktop devices could be made to run at acceptable speed even if they depended on server-based code. IBM could have then offered browsing, productivity and groupware software (and any necessary hardware) on a functional basis. And, because it would have control over most or all of these add-ons, IBM might have been able to provide end-user functionality under an appealing usage-based pricing scheme. This would make it easier for IBM to position the Network Station as an economical alternative to personal computers that still provided much of the fu nctionality end users require. And at the bottom end of the product line, we thought IBM might have been able to offer a box just big enough for a Java virtual machine that could accept a plug-in upgrade module to lift it to the first OS/2 environment level. But then IBM has been saying the Network Station is just one of the thin clients it has in gestation.

By Hesh Wiener

(C) 1996 Technology News of America Co Inc. All rights reserved