Windows NT Directions

Keeping up with changes

Microsoft began working on Windows NT in 1988. The work continues. By the time you read this, NT 4.0 Beta 2 will be out, and the final release will be available in about August. During development of this new version, Microsoft and all the major computing industry players have been making significant strategic decisions about the Internet and corporate intranets that will affect the future of business and home computing. The information age has just found its latest direction.

At a recent Q&A session, Jim Allchin, senior vice president of Microsoft's Business Systems Division, said that Microsoft can "turn on a dime if it has to." Microsoft admits to some recent mistakes, but the company won't be stopped by a few miscalculations about shifting trends in this turbulent industry. Now the big M is making up for these miscalculations and has responded to the actions of other market leaders. For example, after first discounting the Internet as a key business computing component, MS has proved its ability to "turn on a dime" and shown that it considers the Internet extremely important.

Windows 95 and NT
Some have suggested that Windows 95 is all but dead and that Win95 and NT will merge by 1997. This suggestion is both untrue and true. Win95 is not dead, and these two technologies, if not their marketing strategies, will merge.

At least through 1999, Microsoft remains committed to concurrently maintaining NT Server, NT Workstation, and Win95. This way, the company can continue to address small-office and home-office environments, PC users, and corporate environments. Compatibility and reliability stand out in the minds of consumers, whereas security, performance, reliability and scaleability are necessary in the corporate world.

A chasm still lies between home and enterprise systems; the cause is not only differing levels of need and computer know-how, but also price. In both homes and businesses, most computers cannot run NT, and neither can most users.

Microsoft faced this problem with the release of Win95: To properly run it, people need a 486 computer with 8MB of RAM and 40MB of free disk space. These requirements leave many people out in the cold, but Microsoft had a commitment to push the technology ahead. Yet, to soften the blow to low-end consumers, Microsoft still supports Windows 3.11.

Now Microsoft faces whether to migrate the existing Windows and Win95 base and new customers to NT. Microsoft is not forcing the issue either way. It deals with customers one by one, helping evaluate which OS is better for their applications. So companies have to decide to support multiple platforms, stay with Windows for Workgroups, move to Win95, or go to NT. Microsoft acknowledges higher hardware requirements for NT vs. 95 and some missing NT functionality, such as plug-and-play and power management. However, late in 1995, a study by the Gartner Group showed that over five years, NT 4.0 will annually save enterprises about $146 per user over Win95. Also, NT 4.0 will have slightly lower support costs than Win95.

NT 4.0 will make the choice of one enterprise OS a little easier, and future upgrades will help even more. The user interface that 95 and NT share will make training quicker and easier. Microsoft will more seriously enforce NT compatibility for Win95-certified logo applications, so supporting cross-platform software packages (such as multimedia apps) will be easier.

In the second half of 1996, MS will present Nashville (the Internet add-on for both 95 and NT; see "The Road to Cairo Goes Through Nashville," on page 36), and mid-1997 or early 1998 will see simultaneous introductions of Cairo (NT 5.0, as Allchin let slip) and Memphis (the next revision of Win95). By the way, I overheard someone suggest the name Windows 96 for NT 4.0. For well-researched marketing reasons, Microsoft didn't change NT's name with version 4.0: Based on customer feedback about the ensuing confusion, "Windows 96" was not an option. Already, people are referring to future versions of Win95 as "Windows 9X."

The merging of technologies will happen slowly, with some core aspects of Win95 and NT coming together in NT 4.0. These aspects will include the GUI and 32-bit applications that run on both OSs. Win95 and NT 4.0 already share 32-bit APIs, such as OpenGL and DirectX (draw, play, sound, input). Win95's DOS base will disappear by the next release (Memphis), so everyone still operating in the dark ages with a command-line application interface will be forced into the light of a GUI.

A common kernel, the Windows Driver Model, and other architectural changes such as full plug-and-play (PnP) for NT are also in the works, but won't appear before Cairo. The groundwork for PnP and power management is already laid, but drivers need further modification.

Because NT and 95 will share a technology core, migration from 95 to NT will be easier than now. You'll find no support for such migration, even with NT 4.0. Microsoft had too much to do in too little time, so you're stuck with no upgrade tool or strategy for this switch. It requires a complete re-install of all software on your system. Other consequences of a shared core technology will be smoother application development and simplified support for both platforms. Optimizing both the OSs and their applications will also be easier.

On the marketing side, Microsoft will now synchronize NT and 95, so major OS upgrades and release of technology updates (e.g., ActiveMovie and Direct3D) will be simultaneous. Through service packs and shell updates, Microsoft will synchronize bug fixes, too.

The Distant Future
In computer terms, the distant future is six months, and the unimaginable is one year. Microsoft has plans that extend that far, ranging from new APIs to a new 64-bit version of NT.

New versions of the DirectX APIs, besides the ones to be released in NT 4.0, will include better and faster 2D performance through DirectDraw 2.0, and improved network support under DirectPlay, an Internet gaming architecture.

Microsoft considers game development important for two reasons. First, the games industry already boasts a large bank account, and second, games and multimedia applications have always pushed the limits of OSs and hardware. Customers want realtime interaction, blazingly fast graphics with synchronized audio, and network game play. Microsoft wants to ensure that $200 game consoles don't outperform $12,000 graphics workstations. And these technologies apply equally to business applications, such as Internet-interactivity (Virtual Reality Modeling Language--VRML, Java, etc.), improved graphics for office automation, enhanced CAD and 3D animation performance, and audio/video conferencing.

Future releases of DirectX APIs will improve performance. Direct3D (due after the release of NT 4.0) will offer software emulation and acceleration to make up for deficiencies in hardware functionality. It will support full hardware acceleration for OpenGL cards and Intel's MMX multimedia architecture.

NT 4.0 will have Win95 multimedia compatibility through DirectX, and more capability is on the way with the future additions of ActiveMovie (formerly code named Quartz) for video streaming and full support of all media formats. In early 1997, the Windows Driver Model will simplify our lives: Sound cards, display drivers, and MIDI devices, for example, will all work on 95 and NT. Also easier will be providing software and hardware that operates transparently to the user. Developers will be able to write to a common API and offer the same software and hardware capabilities no matter what the implementation. ActiveVRML will improve interaction on the Internet, expanding opportunities for business and play on the Web. The new Internet Explorer 3.0 (IE3) will better integrate the desktop with the Internet (e.g., one applet will let you view material on the Web, browse your system drive, and run an office application). And Distributed Component Object Model (DCOM, formerly Network OLE) will enhance client/server operation and distributed computing.

Who's Doing What?
Apple has a version of NT that will run on PowerPC systems, but Apple is not leveraging that technology. HP ported NT to the PA-RISC architecture, but performance was not exciting compared to much cheaper Pentium-based workstations and servers. So HP is banking on the new P7 technology it is developing with Intel. Even Sun Microsystems played with porting NT to the SPARC chip, but never got past the preliminary stages.

In NT's established RISC market, NT 4.0 offers full 486-enhanced mode operation on MIPS, Alpha, and PowerPC systems, and Digital Semiconductor has released FX!32 for full 32-bit Intel compatibility on Alpha systems. This functionality will come to MIPS and PowerPC systems eventually--perhaps in Cairo, or sooner if each chip developer builds emulators, as Digital Equipment did.

Microsoft will exploit NT on RISC platforms, but the PowerPC is a problem. The 150- and 200-MHz 604 and 604e perform well, but they're beginning to lag behind Alphas and Pentium Pros. The PowerPC's price/performance point is not as good as those of the other platforms now. Unless the 620 gets to market in the near future, it may be edged out of the market. If not corrected soon, the combination of lower performance, limited Intel-compatibility, and limited native software availability (compared to these other two platforms) spells doom for this IBM and Motorola venture in the Windows marketplace.

Now What?
With all this strategic change will come improved performance, new capabilities, and new courses for personal and business computing. Microsoft has demonstrated its ability to do an about-face. Now, users have to remember that we are part of the fastest moving industry on the planet, so we have to be able to react to industry changes just as quickly.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.