Bill Gates of Microsoft Inc reportedly told a meeting of the Tandy User Group in the US last spring not to rush out and buy the first releases of OS/2, because there’d be nothing to run on it. But recent predictions suggest that over one million 0S/2 systems will be in use by the end […]
Bill Gates of Microsoft Inc reportedly told a meeting of the Tandy User Group in the US last spring not to rush out and buy the first releases of OS/2, because there’d be nothing to run on it. But recent predictions suggest that over one million 0S/2 systems will be in use by the end of 1988. IBM says it has already shipped a million PS/2 systems to dealers and third parties (currently running PC-DOS, but many of them ripe for upgrade business), and the commitments of companies such as Compaq, Olivetti, Tandy and Zenith, amongst many others, to offer OS/2 on their products this month makes it a certainty that by December 1988 the new operating system will have affected the low-end computer market significantly. Single-user What form are those changes likely to take? Microsoft of course also has a large stake in the other two operating systems strong in this area: MS-DOS and Xenix, and is therefore concerned to show that each has its place in the future. Gates has been quoted as saying that MS-DOS will continue as the main workhorse for personal computers over the next few years, particularly for those only needing to run real-mode applications. And whilst OS/2 should provide the facilities required for a new generation of sophisticated applications for personal computers, Xenix offers a cheap, multi-user solution with a wide range of existing software available. But although Microsoft says that OS/2 is intended as a single user operating system, John Bondi of Kernel Technology in Leeds, said that this was more of a marketing decision than technical limitation. OS/2 is multi-tasking, and it would not be difficult to extend it for multi-user operation, he says. But even as a single-user operating system, OS/2 is far more suitable than MS-DOS as a vehicle for networking. Whereas MS-DOS always had to be bodged to make local networks operate, OS/2 will make competition between local area network systems and Unix viable for the first time. However, many hardware manufacturers, formulating their strategy over the next year, seem to believe that the three systems can co-exist, at least for the next few years. Apricot Computers has both Intel-based systems running both MS-DOS and Unix, and will be offering OS/2 from this month. Peter Horne, research and development director at the company’s Birmingham offices, says that although there will be no significant applications for OS/2 until at least 1989, Apricot will be implementing it on servers to provide increased functionality in 1988. Unix will continue to provide the most effective means of shared resources at a departmental level, communicating with workgroups of MS-DOS micros attached to their own OS/2 servers.
While seeing a gradual migration to OS/2 over a number of years, Horne says that in the meantime the company first out with effective network management software to bring them together will be dominant. Software devlopers, of course, have to decide when to begin porting applications over to an OS/2 base. John Bondi says that many will choose a halfway house solution. Monolithic applications will port simply on to OS/2, but to take full advantage you may have to change the underlying philosophy of the program, reorganising it as a suite of functions, in order to take advantage of all the OS/2 facilities. MS-DOS, however, will continue to be a mandatory requirement, so developers will need to provide both MS-DOS and OS/2 versions, a task not helped by the incompatibility of Microsoft Windows 2.0 and the forthcoming OS/2 Presentation Manager, due out in October 1988. Meanwhile, over the last year, IBM has made what many software developers interpreted as its most important endorsement of the Unix standard so far, by announcing the availability of its AIX version of Unix for the PS/2 Model 80 at Unix Expo and Comdex last October. This was followed by a statement that AIX was now regarded by IBM as its standard Unix platform, which would also be offered on the 370-type mainframes. AIX for the PS/2 Model 80 is a subset of the operating system as it appears on the
IBM 6150 workstation only in a very few areas according to Entry Systems Division vice-president Frank King, whose department took over responsibility for AIX in the summer. King said that the intention was to implement as full a version as hardware differences permit on the Model 80. He was not prepared to predict sales of AIX on the PS/2 range, but said that there were large numbers of customers for Unix that require multi-user applications, and that IBM was committed to helping developers port their applications to the PS/2. Merged RT/PS2? Gossip at Comdex had it that IBM was planning a merged version combining the 6150, which has been more successful in Europe than in the US, and the PS/Model 80, either by adding a RISC-based processor card to the Model 80, or by adding Micro Channel Architecture to the RT. Such a move would both expand IBM’s workstation range to compete more effectively with Sun Microsystems and Apollo Computer, and provide PS/2 users with a higher range system. Currently, however, it appears that there will be a considerable overlap between the 6150 and the PS/2 Model 80 running AIX when it becomes available in September. King would not be drawn on the rumours, but said that IBM would be using more common elements between the two systems, including the Micro Channel Architecture. The 6150 already includes a co processor board for PC-DOS applications, and that could be upgraded with faster processors. The frustrating piecemeal schedule for introduction of OS/2, culminating in IBM’s OS/2 Extended Edition Version 1.1, due out in November, will mean that even by the end of the year the full effects of OS/2 on the rest of the marketplace will not be fully apparent. As Bondi says, the key to OS/2 is that it is there to provide better applications to the end user. Those applications are unlikely to emerge in any number until 1989 at the earliest.