Return of the OS Wars?

By Julie Jones and Rolf Jester


Reprinted with permission from UniForum NZ News, the official publication of UniForum New Zealand, the New Zealand National Association of Open Systems Professionals, Vol 7, No 2, March 1997.

Just when you thought it was safe ...

If a modern-day technological Rip Van Winkle had been asleep for the past five years (time moving faster now than in the nineteenth century) he might wake to think he hadn't slept at all. Why, here we are fighting battles over operating systems again! I thought we'd left all that behind in the bad old days. Just when we thought it was safe, we see an article in the Australian "Computer Week" of February 21 headed "Battle-lines drawn for a new OS war."And that leads to this month's observations on the view from across the Tasman.

The "Computer Week" article was talking about the desktop, specifically the ineptly named "thin client". Because Java runtime environments are readily portable, we face multiple operating systems once again, at least for the Network Computers. The contenders, in the red, green, blue, spotted and turquoise corners of the polygonal marketing ring are said by "Computer Week" to be Acorn RISCOS, OS/2 Warp, JavaOS, JavaOS for DOS, NT Workstation, Windows CE, Oracle NC OS and Toshiba JVOC. The bloody battles of System V vs. BSD and OSF vs. UI may have been a picnic by comparison.

The "thin clients" (conjuring up visions of svelte purchasers in an upmarket shop) are a return to a more centralised model, with more stress placed on the server software. That should be good for UNIX, the predominant operating environment for servers today, although Microsoft's Windows NT Server is gaining some ground, at this stage mainly at the low end. On that subject, we saw the regular AUUG column in "The Australian" (Feb. 25) being devoted to an examination of the shortcomings of Microsoft's product in comparison to UNIX. As factual as the article may have been, we are saddened by its defensive tone, because we've seen that before. When we first started promoting UNIX in the mid 1980s, we encountered mainframe professionals who would insist that UNIX would never make it, wouldn't work for enterprise systems, wouldn't scale, was unproven, buggy and unsafe. So we worry when we see serious open systems professionals getting defensive about UNIX. Of course UNIX did make it despite these criticisms, but even after the rise to dominance of open systems, mainframe purchases by users are still at the same absolute dollar level as they have been for many years. Likewise, whatever else happens, UNIX products from many vendors have such strength in the market, that it isn't about to go away.

But we are seeing a greater diversity again. Some would say greater choice; others would call it greater confusion. And maybe some of the diversity of those "slim customers" - oops, "thin clients" - will affect the server environment. We can't be sure, so we have to plan for uncertainty and diversity.

The difference this time is that there is a layer of compatibility above the diverse pieces of hardware and operating systems, namely Java. That gives us hope of compatibility, but only if there is an effective standard achieved. It could come about as the result of total dominance by one software vendor, or through a vendor-neutral standard. I like the latter better because dominant vendors take us back to the really bad, really old days. In fact, though, the customers are tending to prefer dominant vendor-driven "standards".

And of course the idea of universal compatibility isn't new: it goes back all the way to good old COBOL, once touted as the language that would make incompatible mainframes compatible. Hah! This time it might just work, however. The industry and the competitive dynamic among the major software vendors is, I think, ready for a level of compatibility at the application language and network level, leaving plenty of room for them to compete at other levels. With Java-based, network applications that would be a giant step for user-kind.

What we are seeing is a new diversity that will demand that IT professionals take a broader view, not getting overly tied to one vendor's products or one technology. Our task is to ensure that applications are built to cope with, welcome, indeed exploit the changing diversity of operating environments. At least the new OS battles, if such we are about to see again, will create plenty of work for professionals. "Well, that's all a bit theoretical, isn't it? Let's get practical, what kind of thin client should we buy?" "I don't know, but with all this excessive high living going on in this household, even the thinnest client would put on weight!" "How about a nice non-fat cup of Java, then?" "OK, but what brand?"


Back | Table of Contents | Next