25 Mar 1996
When Microsoft was writing the first version of DOS, it was a company with 100 employees. The first PC had no hard drive and no network. Microsoft succeeded by understanding the requirements of the inexperienced end user running a standalone machine. It became the primary software vendor to small companies that might be able to afford only one machine.
IBM's best customers are the Fortune 500 companies. After all, IBM is itself one of the world's largest corporations. Large information systems are installed and managed by professionals. Central control and uniform standards are essential to avoid chaos. IBM, like its customers, confronts a mixture of heterogeneous computer systems including mainframes, minicomputers, RISC systems, and PC's.
As Microsoft did the analysis several years ago, the market for PC software was $10 billion. Microsoft controlled 40% of that total. Even without a Federal antitrust investigation, this did not leave much room for growth. However, the larger software market was $60 billion. The additional $50 billion was being spent on programs for mainframes, minicomputers, and Unix, mostly for database and data communications. Microsoft clearly has better prospects if it can carve out a share of a new $50 billion marketplace rather than growing its share of a $10 billion market it already dominates.
Microsoft needed a system that could compete with minicomputers and low end mainframes, at least as a database server. This requires lots of CPU, and therefore support of RISC processors, symmetric multiprocessing, and the ability of the DBMS to handle more than one transaction at a time through multithreading. The system also needed to support very large datasets, files larger than one disk, and files duplicated redundantly across disks (to keep the system running in case of hardware failure). The data had to be secured with at least a US Government "C2" security rating (comparable to most mainframe systems). Since Microsoft systems would still run on smaller, less expensive machines they needed the network to be integrated directly into all administrative functions so that a collection of servers could be managed as a single system. Given a Microsoft bias to small organizations that lack professional information processing staff, the system had to be easy to install, configure, and operate. Ideally it would look much like ordinary Windows. These requirements define the basic features and objectives of Windows NT.
Microsoft can only grow by undercutting at least IBM's AS/400 and RS/6000 strategy. But while IBM may feel a pinch, the most direct targets are the Unix platforms that currently run Oracle, Sybase, Informix, and Ingres. Although there is a clear business opportunity here, it is also true that Microsoft cannot afford to be a "nice guy". There are some substantial issues underlying the catch-phrases of "Microsoft at Work," "Information at your Fingertips," and "Object Linking and Embedding." Microsoft aims to deliver a simple, integrated, intuitive networked information system. However, once an organization becomes committed to a database system, the database vendor sets the rules. If the database tools don't use OLE 2, then Microsoft's strategy is ineffective.
In 1989, customers and vendors were uncertain about the OS/2, Windows, and other user interface shells. OS/2 had not been accepted. Two releases of Windows had been largely ignored. The most popular application programs, the Lotus 1-2-3 spreadsheet and WordPerfect word processor, ran under plain DOS. When Windows 3.0 was release in mid-1990, nobody expected that it would be such a monster success. However, Word for Windows, Excel, and PowerPoint showed users the advantages that a GUI could provide. The success of Windows, in turn, allowed Microsoft to capture a dominant market share of PC applications.
Today, customers and vendors are similarly confused about "Client/Server" and "Object Oriented." What is the proper way to grow small companies with PC workgroups into more sophisticated information systems? What is the best way to "downsize" large corporate MIS applications to use the power of distributed system? It is reasonable to suggest that a vendor that proposes the right system for application development, database, and distributed information management will define the next cycle of computing.
Microsoft offers a fairly clear set of products. The server is Windows NT on anything from a 486 up to a multiprocessor RISC or Pentium Pro system. The desktop client is Windows 95. The database is SQL Server accessed through ODBC. Applications can be developed in C++ (hard), Visual Basic (easier), or by using end-user tools connected by OLE 2. Software management is by SMS. Wide area communication uses the TCP/IP protocol used on the Internet or Intranet.
As computer prices drop and power grows, sophisticated processing comes within the grasp of every company. Greater sophistication, however, threatens to overwhelm the user community. Microsoft confronts complexity with a family of internally diverse but externally homogeneous systems. Users can choose simple desktop systems for office workers, or complex server platforms shared by an entire department. Workstations can be isolated, locally interconnected, or attached through the Internet to the entire world. Across this entire family of products, there is a common user interface, a common set of applications, interoperable file systems and networking, and an evolving common programming interface and set of development tools.
IBM is a participant in several less clearly focused industry counterproposals. It has joined with Apple, Novell, Lotus, and several Unix vendors in several industry-wide initiatives:
IBM proposes a strategy that allows customers to tie together existing IS resources on legacy systems. Microsoft proposes a new set of platforms that are powerful enough to replace the legacy systems at a fraction of the cost and are easy enough to use that the conversion is feasible.
The IBM policy of working with industry-wide initiatives makes it the captive of committees, joint-ventures, and the standards making bureaucracy. The recent success of Netscape demonstrates the importance of running ahead of standards when the industry itself is moving too quickly. Microsoft makes the occasional mistake (as in the case of MSN) but it is equally quick to abandon its failures (Blackbird), turn 180 degrees, and still arrive with a full set of products before anyone else gets out of Beta.
IBM is rapidly running out of any remaining credibility that it has with the customer base. Although it has long term strategic elements, they do not form a coherent strategy. While the industry is changing almost daily, IBM has announced no major new products in over a year. OS/2 is a good example. Although there are claims of a future release, IBM is not funding meaningful support for the existing system. It cut off its relationship with its customers (TeamOS2) and apparently with the vendors of OS/2 native products. Yet it will not admit what is happening or propose a new policy.
Copyright 1995 PCLT -- Surviving the Next Operating System -- H. Gilbert
This document generated by SpHyDir, another fine product of PC Lube and Tune.
Continue
Back
PCLT