InformationWeek

Microsoft's Operating System Strategy:
Windows NT May Be The Future, But Windows 95 Dominates Now

By Stuart J. Johnston
January 13, 1997

In the past year or so, there have been a lot of breathless articles about the rivalry between Windows 95 and Windows NT Workstation in the battle for the business desktop. Much of this has been based on a fairly fallacious argument that Windows 95 is "merely a transitional operating system" and that NT is the real McCoy.

Why is that a red herring? Because in a very real sense, all desktop operating systems are transitional, and the real question for corporate IT managers is whether an operating system provides users with what they need when they need it at the right price.

But to start with, let's get completely away from computers for a minute and talk about a different but related technology: recorded music. The first commercial format for delivering recorded music, from its invention in 1877, was wax-coated cylinders. Later, the drive mechanism was reoriented and recordings became available on acetate disks. These platters played at 78 revolutions per minute (RPM) and could only hold a few minutes of music per side, but the reproduction quality was much improved.

With improved plastics came vinyl records, which were much less brittle and more durable than acetate. Vinyl disks, which spun at 33 RPM, could hold more minutes of audio and had better sound quality than 78s. In the mid-1960s, eight-track tape cartridges were introduced and became extremely popular for use in automobiles as an alternative to AM radio. But eight-track audio quality was not as high as vinyl, so they did not displace LPs. However, another playback technology did -- the cassette tape. While the audio quality was lower than vinyl, cassettes were smaller and easier to use. Within a decade, cassettes eclipsed LPs in sales.

In the early 1980s, compact discs arrived. CDs had digital quality that for most listeners far surpassed even vinyl and had other advantages, such as random access to songs anywhere on the disk. They also had many of the benefits of cassettes, such as a smaller format and resistance to damage.

Which of these media were in u se the longest? The answer, by several decades, is the wax cylinder. In fact, the life cycle got progressively shorter for each successive generation. As a technology matures, the interval between major changes gets shorter.

The same can be said of operating systems for desktop computers. DOS was the dominant operating system on PCs from 1981 until sometime in the early 1990s. An argument can even be made that, since DOS was required to run Windows, DOS's dominance continues.

Along the way, we got used to Microsoft's shipping new versions of DOS and Windows about every 18 months. But the systems remained backward compatible. Windows 3.0 applications run under Windows 3.1, Windows for Workgroups, and even Win95. In fact, providing backward compatibility in Win95 was one of the biggest reasons its release was delayed nearly nine months. Windows NT had less of a charter to provide backward compatibility than Win95, and it accomplished even less than promised.

Now, there have been lots of articles in the press, both trade and mainstream, saying that Win95 was bound to be and has been a big flop with corporate America. That, however, is not a correct assessment of what is happening. In fact, many members of the press are either too young to remember what made Win3.x popular in the first place or they have selectively forgotten. As sales of PCs grew each year, Microsoft negotiated deals to bundle Windows on the vast majority of new machines. And as corporations phased out older, DOS-based PCs, the new ones came with Win3.x preinstalled.

Initially, a lot of IT managers simply took Windows off new PCs as they came in. But momentum gradually turned in favor of Windows, partly because of pressure from early adopters and partly because pilot projects successfully reached completion.

At that point, corporate IT resistance to Win3.x slackened and more companies started leaving it on new PCs than taking it off. It took six months to two years for resistance to fall to the point that Windows saturation of th e corporate desktop became an inevitability. The arrival of applications that exploited Windows' features also helped, and those products took between six months and two years to ship.

So where are we a year and five months after Win95 shipped? By now, most corporations have made a decision to go to Win95 or skip it and go to NT, and the largest percentage have chosen to go with Win95. One reason is that the Win95 user interface is easier to use than Win3.x's (There are advantages to incumbency, so switching to a non-Windows is out of the question in most cases). Another reason is that Win95's hardware requirements more closely match the capabilities of the new PCs that are shipping. Still another reason is that support costs for Win95 are generally lower than for NT.

There are those who insist that Win95's popularity is almost entirely based on all the hype surrounding it. But while some IT managers might be susceptible to hype, the vast majority of corporate IT managers are not lemmings. They make d ecisions for sound business reasons, not whimsy. Most corporate customers make tough evaluations before choosing technologies, and if they get duped once, they don't tend to buy from that vendor again.

By the end of 1996, those few market researchers whose predictions I trust estimated Microsoft had sold around 60 million units of Win95. Since there are only about half that number of PCs in homes and many of those are not running Win95, a lot of those Win95 units pretty obviously went to corporate customers. Additionally, a lot of large companies have recently committed to large-scale Win95 deployments. That indicates to me that we have reached the point where more companies are leaving Win95 on new PCs than not, and that Win95 is an inevitability on most corporate desktops over the next couple of years. The arrival of Office 97 -- the first major update to Microsoft's dominant desktop productivity suite since Win95 shipped -- will also hasten corporate acceptance.

Now that NT 4.0 is available, a lot of companies have also begun evaluations and pilot projects to see if they want to use it on some or all of their desktops. Some have already chosen to go that route. In fact, sales of NT are ramping up, though not at Win95's explosive pace. And you can bet that if Microsoft was selling millions and millions of copies of NT Workstation that they'd be trumpeting it at every opportunity. That, however, hasn't happened. In fact, several companies that I've talked to wanted to go to NT on the desktop but found the costs prohibitive.

Eventually, Windows NT will probably overtake Win95. Microsoft may even be tempted to discontinue Win9x in order to move NT onto desktops sooner, but I personally think that would be stupid. Of course, at a point near the end of the century, Microsoft's aim is to subsume Win9x into NT Workstation so that all desktop PCs will be running on the NT kernel. But a major switch to NT on the desktop would require that NT become more manageable and supportable as well as that there's a la rge proliferation of PCs with the hardware capabilities to comfortably run NT at today's average price point for a desktop PC.

That will most likely be in 1999. Microsoft has at least one more major release of Win9x -- the "Memphis" release coming in the second half of this year -- before Win9x's kernel is swapped out for NT's. Plans change, but to cut off Win9x's life cycle before users are ready in order to consolidate everything under NT would be short-sighted.

This can be illustrated in the earlier analogy to music playback formats. After the huge success of CDs, the hardware vendors thought they could tempt consumers to transition to a new format. CD player sales had slackened and become a replacement market, and the vendors wanted to jump-start new hardware sales. To date, digital mini-discs have been introduced to the market twice, the format has garnered little customer acceptance. The vendors blame it on the shortage of titles on mini-disc, but I believe there is a more pragmatic reason.

Consumers have a large installed base of CDs, and switching formats means tossing out that investment. After all, CDs do not deteriorate every time you play them. Mini-discs offer a smaller form factor but lower audio quality, and they won't play CDs. There simply are not enough compelling reasons to switch.

So, at some point, I think, product life cycles reach a point beyond which they cannot become shorter due to user or customer resistance. Still, the original statement that all desktop operating systems are transitional mostly holds true. After all, about the time that Win9x converges with NT Workstation, Microsoft is due to come out with a 64-bit version of NT to compete with the high-end 64-bit versions of Unix that are coming out now.

Of course, there are those that believe that network computers will make this whole discussion moot. I have to disagree based on my own observations, but that will have to be the subject of another column.

Return to Aut horITies or view AuthorITies archives

Comments?

http://www.informationweek.com