The Missing Network Link

by Lloyd Borrett
Today's Computers, Opinion, March 1985

Last August, IBM announced both the PC Network and the IBM PC-AT. On the surface these announcements seem to mean little — yet another local area network and a faster PC — but let's look a little deeper.

With the PC-AT came the direction from IBM many have been waiting for: the multi-user operating. system for the PC range will be Xenix. There had been many predictions on what IBM would do to provide a growth path for single-user PCs to the next generation of multi-user systems. Well, now we know: IBM has picked Xenix, an implementation by Microsoft of the UNIX operating system.

What is the significance of that decision? For some time, most system houses have been waiting for IBM to indicate which multi-user system to use. These system houses write software for small and intermediate businesses such as accounting packages, sales analysis systems, and stock management systems. Although they could initially get away with selling single-user PCs to this market, they knew that many of their customers would eventually demand a multi-user solution.

The choices available were minicomputers, or microcomputers linked together in local area networks (LANs). The problem was that most minicomputers were considered too expensive, and there were no standards for LANs. The IBM PC-AT running Xenix is a "supermicro" that can fill the gap.

So the system houses rejoiced at the news, and are now all very busy preparing multi-user Xenix versions of their software. The potential users of the software should also be pleased, but what about the rest of us?

Any small multi-user single-processor system has a lot of inherent problems. The potential for conflicting interests on such a system should be obvious. If one user is making heavy use of a shared fixed disk or printer, other users can be severely affected. The smaller the processor, the less opportunity there is to duplicate resources. The less powerful and distributed the resources available, the greater is the chance of conflict.

But there is another way. IBM's announcement of the PC Network, with support from PC DOS 3.1, offers an attractive alternative. Unlike most other LANs announced by small hardware developers, this has the necessary backing for it to become a standard. So now we can solve some of the inherent problems of multi-user single-processor systems with multi-user multi-processor LAN-based systems.

For many PC users, LANs are a far better alternative. Each user has their own processor, memory-mapped screen, and local disk storage. Expensive resources such as high-speed printers, plotters, and communications links can be shared, as can key data files. The pool of resources available can easily be increased, making it possible for users with heavy demand for a resource to have it duplicated locally.

Unfortunately, the system houses responsible for providing most of our application software are heading in the other direction - multi-user single-processor Xenix. IBM did not announce support for PC Network from Xenix. It will come, but it may be too late if the software houses do not plan for it now.

In the computer industry eight years ago, minicomputers highlighted the problems of too many shared, but unique, resources. The fastest, most powerful minicomputers available had the same problems as today's supermicros. The peripherals available, such as disks and printers, were not powerful enough and the computers lacked the capacity to support the necessary number of less-powerful peripherals.

The first solution was to look for more powerful computers and peripherals, but eventually it became common to see minicomputers linked together, sharing a greater pool of resources. Today the supermicro exhibits all the potential of the early minicomputers. If used properly for the right applications, multi-user microcomputer-based systems can be very powerful business tools. But we should heed the lessons of the past, and plan to use local area networks to link systems and create a greater pool of resources.

IBM-Microsoft Relationship

All is not rosy in the IBM-Microsoft relationship. The dominant partner, IBM, is creating its own software, and much of it is in direct conflict with Microsoft's, offerings.

In the middle of the excitement over the IBM PC-AT, IBM slipped in a little goody called Topview. Simply put, Topview is a windowing system that will allow programs to be executed and integrated similar to those on the Apple Macintosh. But, unlike the graphics-oriented technology of Visi-On, Lisa, Macintosh and Microsoft's much promised Windows, Topview can run on the two-thirds of PCs that use the character-only monochrome monitor.

This has given the marketing people responsible for Microsoft's Windows and Digital Research's GEM a few sleepless nights. IBM has already put Topview in the hands of hundreds of software developers, which means that applications software using Topview should be seen very soon after Topview becomes available.

Topview lets you fire up several programs and switch freely between them. It handles work that can run unattended as background tasks, and provides any program that is written to use it with the ability to handle panels, sub-tasks, and communications between programs. Topview keeps all programs in memory instead of paging programs to disk, but that may be removed before the product is available.

Topview has the potential to vastly upgrade the quality of programs and, unlike most of the other windowing systems, it is easily accessible to both big and small software developers. But forget about the technical details and concentrate on the impact of this product, and some of the other moves IBM is making.

Microsoft has been promising Windows since November 1983 and now says it will be available "sometime" this year, but IBM has produced its own product. Although Microsoft has a local area network (LAN) interface, IBM is selling their own. IBM has announced Xenix from Microsoft, but certain key elements are missing. There is no PC Network interface or applications software, and lately IBM has been actively announcing DOS and Topview applications software.

But it seems likely that IBM has a "Topview 2" that is a true multi-user, multi-tasking operating system with full PC Network, 3274 terminal emulation, and SNA support. To make it available too soon would provide unnecessary competition for the existing System 34 and System 36 range.

Although there are no real Topview programs yet, we should have to wait only until the first half of next year. By then all of the major programs such as Lotus 1-2-3, dBase III, and Multimate should have been rewritten to work as Topview modules. If they are not, someone will come up with products that replace them (remember that IBM has a big lead, and has already been announcing Topview integrated products).

You can all but forget Xenix and UNIX. There will still be a market for applications that use "supermicros" to replace minicomputers, but the mainstream will be IBM's Topview and "Topview 2". Why would IBM want to encourage people to write software that can be easily ported to other UNIX machines?

IBM led the way into 16-bit micros with the PC, and set the hardware standard. Now we are being shown the software standard. The future does not seem rosy for Microsoft; IBM looks like giving them the flick. No wonder Microsoft is putting so much effort into developing Macintosh software.

Digital Research, with a Topview-compatible concurrent DOS, could become the alternative for IBM-compatible manufacturers wanting to stay in the market. But to elaborate on that would require another session with my crystal ball.