The trends suggest that 1996 will be the pivotal year in the battle for control of systems software on small and mid-sized servers. The contenders include Windows NT, OS/400 and Unix. The trends we have studied show that OS/400 will be a very popular choice and that its user base is likely to grow during the coming year. But OS/400 is not going to be number one in terms of popularity. That leaves Windows NT and Unix as the two top contenders. The main difference between them is that Windows NT is a single operating system that runs on several chips. Unix is a collection of similar but not identical operating systems… and with very few exceptions, each variation on the Unix theme runs on only one family of processors. For instance, AIX runs on computers of various sizes, but all of them are based on the IBM Power processor architecture. While Windows NT is most popular on servers using iAPX-86 chips, it is also available in an identical form for a number of other kinds of computers. (NT applications, which include Windows95 applications, do have to be recompiled for each hardware platform, but they do not have to be rewritten.) The roster of NT platforms include the DEC Alpha servers, various machines using Silicon Graphics MIPS chips, IBM and Motorola PowerPCs, to name a few. During 1996, NT’s domain will be extended to include any Macintosh machines that might be built to the common hardware specification. After that, Apple can kiss its sweet OS goodbye. Too late now Windows NT wouldn’t seem so threatening if the Unix crowd had ever fulfilled the promise of standardisation so often talked up in the past. Porting Unix applications to new platforms involves not just recompilation but actual changes in source code, raising the cost of a software vendor’s initial debugging and after sales support. But that’s academic; it’s too late now. Sure, there’s yet another Unix unification campaign under way, but we think it’s a lot of malarkey… as usual. Unix won’t go away, of course. But as standardised Windows NT applications increase in variety, scope and quality, there is only one likely outcome. There may be resentment of Microsoft within the computer industry and even among users. But practical considerations will in the end determine the course businesses must take. If Unix software is substantially more costly than equivalent NT software, Unix software will not be as popular. The main reason Unix became successful at all was that it was cheaper for a software developer to reach a large market with a Unix program than with a program for a proprietary environment, such as MVS or OS/400. Windows NT will be to Unix as Unix was to MVS. The Unix situation is nothing less than tragic. The hubris that led to AIX being different from HP-UX is now going to lead to the demise of the Unix hero. copyright (C) 1996 Technology News Ltd, London.
