I had entered a virtual lion's den. At the most recent UNIX Exposition in New York City a few months back, I was part of a panel discussing "The Current and Future Impact of Microsoft on UNIX" when a participant's UNIX-
centricity began to show. Although I have a lot of respect for UNIX, I didn't share the myopic view of many of the attendees that UNIX is the way to openness and freedom of choice. In fact, I quarreled with the use of the word "open" when used in conjunction with UNIX. At the time, I felt that most of the people in the room were in heavy denial mode and could not see the obvious: UNIX, in the face of stiff competition from Windows NT, has some tough times ahead.
I put forth the somewhat radical--at least to these people--notion that today's NT platform is in fact more "open" than UNIX has ever been. Consider the particular systems administrator who supervised 12,000 Windows-based PCs and 2500 Sun Microsystems workstations, using NT as a file server and the Solaris flavor of UNIX as the enterprise server operating system. He wasn't happy when I pointed out that while everything on his PC platform was virtually interchangeable among hardware brands, switching hardware vendors for any aspect of his UNIX environment would require major time, labor, and financial investment.
In fact, NT has become the great hope of distributed computing and is destined to take the market where no high-power PC has gone before. By combining popularity and power, NT will bring the economics of mass distribution into play in the mission-critical enterprise client/
server market where UNIX thrives today.
Open...or Not?
It's a grand irony. The so-called "open" system is actually highly restrictive. UNIX, designed and implemented by a committee in order to be fair to all participants, suffers because none of its sub-brands is compatible with the others. Perhaps this is part of the price for its being the first operating system available from multiple vendors but controlled by none: They all needed to have their own say. NT shares this multiplicity with UNIX, albeit with one significant difference: Although NT is available on diverse chip architectures from a wide variety of systems vendors, the operating system itself is manufactured at a single source and supplied with consistent user and programmer interfaces to all hardware vendors. This supplies the user with a practical "open" system.
For instance, a corporation with 1200 Compaq PCs running NT may decide to upgrade 20% of its employees. During the year since the last hardware purchase, it turns out that Digital now offers a better deal on NT hardware. This is no problem because software and users can switch to Digital hardware and still work as before.
What if this same company also wants to augment its UNIX components, manufactured by Sun, to a more cost-effective solution supplied by Hewlett-Packard? Instead of "no problem," you now have "no way," because HP-UX (HP's brand of UNIX) has programming and user interfaces different from Solaris' (Sun's brand of UNIX). This means that replacing any number of Sun platforms with HP workstations or servers requires that programs be ported, users and system administrators be retrained, and new software licenses be purchased. The cost of these changes alters the economics of the HP solution.
The objectives of "openness," as described by POSIX, are software portability across hardware platforms, interoperability among hardware platforms, and user mobility (the ability to move between applications or within the same application on different machines without retraining). It's clear that NT beats UNIX in all these areas.
This difference isn't lost on users. Everyone from the Coast Guard to Kmart--along with other former UNIX fans--is jumping on the NT bandwagon. UNIX loyalists, however, shouldn't feel left behind. Microsoft, in developing NT, borrowed heavily from the UNIX environment. The wisest course for UNIX partisans is to not lose sight of what has worked up to now. They should stay true to the software and its functions, while being open to the possibility that UNIX can take advantage of the emerging NT platform.
A Brief History of UNIX
UNIX is a powerful system that predates MS-DOS and has long satisfied more demanding, sophisticated engineers, scientists, and software developers because of its 32-bit architecture, preemptive multitasking capability, robustness, networking capability, and ability to run on powerful new hardware. In spite of UNIX's technical advantages, a mass market never emerged for UNIX software because of diversity between the applications programming interfaces (APIs) and user interfaces (UIs) among the different brands. This diversity led to higher software research and development, marketing, and distribution costs which raised the price of the software to customers and limited the size of the market, when compared to MS-DOS and, later, Windows alternatives.
NT's underlying capabilities echo those of UNIX. In fact, its architects designed it to include the major technical capabilities of UNIX. However, they omitted one significant characteristic: The diversity of APIs and UIs among UNIX vendors. As a member of the Windows family, NT has a consistent UI across all hardware vendors and applications. It also has a consistent API, known as Win32, across all NT hardware vendors which enables NT to run applications written for Windows 95.
Any system that effectively combines power and ease of use will win fans across the computing spectrum. NT can run complex mission-critical software, which required UNIX in the past, on machines that can simultaneously run the popular productivity applications available for Windows 3.1 and Windows 95. NT provides an important part of Microsoft's computing strategy and has gradually earned a solid reputation, industry accolades, and an increased user base. Also, the next interim release of NT will incorporate the new Windows 95 interface. So in one year, this already powerful and easy-to-use operating system will gain even more momentum from the popularity of Windows 95.
The NT vs. UNIX battle is a washout when it comes to technology. Compare the two, and you'll find more operational similarities than differences. Both are robust 32-bit architectures with preemptive multitasking and built-in networking. Each runs on multiple-vendor platforms and chip architectures and can accommodate mission-critical applications.
So the battle will occur on economic grounds. When technology is equal, users turn to more practical factors to help make their decisions. Which system has the lowest overall cost? Which environment supports the most applications? Which system is easier and less expensive during the upgrade process? Which system allows mission-critical applications to run alongside productivity applications on one machine? When you look at it this way--outside the fog of UNIX bias--NT wins in every column.
Both Oars in the Water
UNIX dropped the ball. Imagine a sunny day on the Charles River. A UNIX crew is in one boat, and a Windows crew is in another. The Windows boat is coxswained by none other than Bill Gates, whom the crew follows with diligence and loyalty as it rows in one direction. No one is quite certain who is coxswain of the UNIX boat or whose directions the crew is following--especially as the crew members appear to row in different directions.
Winning large markets requires a focused commitment. The standard architecture for centralized enterprise computing, MVS, was set by IBM with that same singularity of purpose that Microsoft has today. Now, Microsoft effectively controls a virtual corporation of ISVs, Solution Providers, and Systems Integrators who are all rowing in the same direction. When they compete, everybody wins: Customers and vendors all benefit from the consistency of products and the economics of mass distribution. NT gives Microsoft the ability to bring this business model to the enterprise. Although many people think that UNIX is still the heir apparent to this market, it lacks the singularity of purpose to compete with Microsoft and NT.
Windows NT offers freedom of choice, which is of significant economic value to users. It is the result of an absolutely consistent operating system in terms of APIs and UIs. Ironically, this consistency is born out of one vendor's control over these critical interfaces. Although UNIX enthusiasts disdain this control, the resulting consistency creates a high-volume market and volume brings value to customers. Hundreds of vendors sell products based on the consistent NT interfaces, so NT users have more choices to switch brand loyalties and retain their hardware, software, and training investments, especially as they make subsequent purchases.
A New Mass Market
We need to look outside the computer industry for adequate comparisons. Few people admitted shopping at Kmart 10 years ago. Now price makes the difference in most buying decisions, and mass distribution of brand-name products has been popularized by stores such as Price Club and Home Depot. It's even fashionable to shop there. By selling in volume, these vendors can cut prices and beat the competition. The Japanese did much the same thing with cars, offering simple value and price for years. In 1990, with the introduction of the Infiniti and Lexus lines, they dramatically cut the price for luxury cars and showed customers that they could buy high-end quality for significantly lower prices.
Enterprise computing will be the next beneficiary of this business model. Microsoft has created a high-volume market where competition brings quality and value to customers. Mass distribution of NT products will provide distributed enterprise solutions that are built from commodity components and will essentially change the face of the industry.
In addition, the plethora of Windows applications now available, along with those that will arrive with Windows 95, make NT an even more attractive investment. Businesses choose their operating systems on the basis of available applications, and NT can support mission-critical applications on the same machines that run popular Windows programs. Furthermore, although both UNIX and NT Server are powerful application servers, NT Server excels in file and print services--areas where UNIX is traditionally weak. NT's ability to support this range of services on a common server architecture creates additional cost-saving opportunities for organizations. Companies can consolidate their UNIX and Novell NetWare server architectures, removing unnecessary complexity and saving on system administration, software licensing, and training.
NT is destined to become the standard architecture for distributed enterprise computing, creating a dramatically larger market for workstations and small to midsize servers. Any UNIX vendor who wants a piece of the action needs to be aware of market realities.
UNIX has a valuable base of enterprise client/server applications and developers that NT needs in order to win the enterprise. The UNIX community--software vendors, hardware vendors, and corporate users--needs to leverage these valuable assets as the NT market grows. How individual members of this community do this will determine their future in enterprise computing. This leverage will need to take into account market timing, transition costs, the ability to fully exploit the NT architecture, and the preservation and acquisition of customers. UNIX vendors can thrive in the NT era, but they must recognize the opportunity now.
Such an evolution is not unprecedented. Ken Olsen, Digital's founder and former president, once described UNIX as "snake oil" when Digital's VMS was on top of the world. NT now brings the same advantages over UNIX that UNIX once had over VMS. Consequently, each member of the UNIX community needs a strategy for moving to Windows NT that will leverage its current investments and assets. The time to act is now.
