Revised November 24, 1997.

This case was prepared by Chris Hance, Kevin Koeb, Lisa Peters, Brian O’Fallon, John Rokke, Carmen Scott, and Emer Dooley, from the University of Washington, School of Business Administration, as the basis for class discussion rather than to illustrate either effective or ineffective handling of an administrative situation.

The 1997 Annual Report chronicled yet another amazing year for Microsoft -Income of $3.5 billion on revenues of $12 billion, up 36% from 1996. As the world’s biggest software-only company (IBM has a bigger software business) Microsoft’s share of PC operating systems topped 90%; it seemed the war for the desktop was won. Microsoft was starting to reap the benefits of a strategy, pioneered by their CEO, Bill Gates in the late 1980s: to move PCs, running Microsoft operating systems, into c orporate America (i.e. the enterprise), to replace traditional mainframes. With sales of their "industrial strength operating system" NT booming, it seemed that strategy was beginning to pay off. Or was it?

Bill Gates nemesis, Scott McNealy, chairman of Sun Microsystems (see Exhibit #1), claimed that the Network Computer (NC) running Java was a far more cost effective solution than the Windows-Intel PC platform for big corporate customers. Quoting analysts' estimates, McNealy claimed that the Windows platform costs between $7,000 and $11,000 per year per user, while Sun's network computer, called JavaStation, will cost only $2,000 per user per year. JavaStation, is a diskless computing pla tform that downloads applications from the server on startup. McNealy reiterated his contention that PCs are too complicated to encourage the widespread use of computing technology. Despite the extensive use of microprocessors in everyday life, users are still frustrated by PCs, McNealy claimed, saying, "The only computers most of us don't know how to use are Microsoft computers."

The criticisms of Microsoft products as large, difficult and expensive to use "bloatware" were coming to a head. If Sun and McNealy could succeed in displacing the PC with a ubiquitous machine that didn’t need an operating sys tem, that was a strike at Microsoft’s core business. A computer without an operating system, was Microsoft’s worst fear realized. They had worked long and hard at establishing a toehold in the lucrative enterprise business. It became clear they had to hav e a response to the Network Computer’s promise of cheap, easy to manage computing. They had to tackle the PC "total cost of ownership (TCO)" problem quickly.

Microsoft (originally Micro-soft) was founded in 1975 after 19-year-old Bill Gates dropped out of Harvard and teamed with high school friend Paul Allen to sell a version of the programming language BASIC. Microsoft's big bre ak came in 1980, when it was chosen by IBM, over Gary Kildall's Digital Research, to write the operating system (software that controls a computer's basic functions) for IBM's new PC. Faced with a complex task and a tight deadline, Microsoft bought the ri ghts to QDOS (short for "quick and dirty operating system") for $50,000 from Seattle programmer Tim Paterson and renamed it the Microsoft Disk Operating System (MS-DOS).

The popularity of IBM's PC made MS-DOS a near-instant monopoly because other PC makers wanted to be compatible with IBM. It became the standard PC operating system in the 1980s. In the mid-1980s Microsoft introduced Windows. When the c ompany went public in 1986, Gates retained 45% of the shares, becoming, in 1987, the PC industry's first billionaire (he now owns 22% of the company).

In the 1980’s Bill Gates envisioned growth from expanding the reach of PCs into the high-end of the business i.e. large corporations and financial institutions -a market traditionally served by workstations, minicomputers and mainframes and companies like Sun, Tandem and IBM. Microsoft’s MS-DOS operating system had a reputation for being unreliable and lacked the scalability and industrial-strength networking required for large corporate users. MS-DOS was considered a "toy" op erating system by many of the Information Technology (IT) professionals and a major impediment to Microsoft’s push into the enterprise. Microsoft had to develop a robust industrial-strength operating system and scalable software solutions to capture corpo rate customers.

With this in mind, in 1988, Bill Gates recruited Dave Cutler, formerly of Digital Equipment Corporation and arguably one of the best computer scientists in the country, to head development of what would eventually become known as Window s NT. Signaling Microsoft’s determination to enter the lucrative enterprise market, Cutler, along with a team of approximately twenty other Digital-expatriates become the core of Microsoft’s NT group.

Windows NT, like other operating systems, was designed to handle internal "housekeeping" functions of the computer – such as organizing the storage and printing of files. NT, however, would push existing technology by being t he first 32-bit Microsoft operating system (previous versions of Windows and DOS were 16-bit). NT would also allow for running multiple applications simultaneously – protecting the others if one should crash. Moreover, NT would remain backward co mpatible with all existing Windows applications, and would be designed to run on a wide array of hardware – from desktop PCs to refrigerator-sized servers.

It was hoped that NT would compete successfully with network friendly, enterprise-established rivals such as (Novell) NetWare and UNIX. The huge advantage of UNIX was that it supported a vast array of applications, and could be implemen ted on a number of different hardware configurations. Novell created server software called NetWare, that facilitated the sharing and printing of PC files over a network. By combining the functionality of NetWare and UNIX programs into a single operatin g system operable on a common PC, NT would become the enterprises’ ideal product choice.

Over the next five years, Cutler’s group grew into a 400-person development team and spent $150 million producing 6.1 million lines of code. The culmination of their efforts, Windows NT 3.1, was released in late 1993. Defying Bill Gate s’ prediction that over one million units would be sold in the first 12 months, NT 3.1 suffered poor reviews, and took three years to pass the one million units sold mark. NT 3.1 was followed however by a string of successful NT upgrades: Windows NT 3.5, Windows NT 3.51, Windows NT 4.0, and soon Windows NT 5.0. After almost 10 years and 25 million lines of code, Microsoft now had its robust enterprise-ready operating system. Ironically in those 10 years corporations began to balk at the cost of running their business on PCs and look for other alternatives.

Traditionally banks, large corporations and government agencies were "mainframe shops". The Information Technology (IT) group ran a centralized computer organization, housed in raised-floor, halon-protected, air-conditioned da ta centers. The IT budget was a known quantity allocated on an annual basis and companies understood exactly how much this was costing them.

PCs began to appear in corporate America in the 1980’s as the result of a strange grass-roots movement. Employees, driven chiefly by the use of word processor and spreadsheet applications, began requesting PCs at work. They were typical ly bought out of business-unit budgets and soon proliferated. While productivity increased, business-unit managers became increasingly frustrated with the amount of valuable employee time being spent on setting up, upgrading and "futzing" with c omputers. Once the problem reached critical mass, the IT groups were asked to help manage the problem, and now inherited a system that was out of control. IT groups in many cases now "owned" PCs that weren’t part of any IT budget. Worse still, u nlike the centralized model they were accustomed to, these PCs were "distributed" all over the company and required difficult, frequent and thus costly maintenance. The Total Cost of Ownership (TCO) was on the corporate radar screen.

Total Cost of Ownership (TCO) is the aggregate cost of all expenses associated with ownership of a personal computer over its entire life cycle. These include both direct and indirect costs, and include the following:

• Initial investment in the PC
• Maintenance of the computer system (Installation and infrastructure cost)
• User and operator training cost
• Salaries of systems personnel
• Software
• Upgrades to hardware and software
• Lost productivity and revenue due to downtime

• The "futz factor" – the cost of non-productive work hours spent getting a PC to work as intended (e.g. :if it takes a user 4 hours to figure out how to make tables in a new program they just installed).

Large corporate-user headaches with the "PC problem" started to gain voice. Exactly how and when TCO became a real issue in the information technology world is difficult to pinpoint. TCO first gained public prominence in Janu ary, 1996 when the Gartner Group of Stamford Connecticut released a study estimating the total annual cost of owning and maintaining the average PC costs as much as $12,000. (Exhibit #2 depicts the cost elements included in this estimate.) Since then est imates of annual costs have varied anywhere from $7,000 to $29,000 depending on whom you ask.

Given the complexity of costs and myriad of cost estimates, it was difficult for IT managers to know their total (true) costs. Now, facing even more pressure from corporate executives, IT Managers had to do something to reduce costs fas t. If the distributed computing model (i.e. each user in a corporation has their own PC) was going to win, the question became what was the most cost effective way to do it. Numerous competing solutions started to appear.

One of the first heavily advertised solutions to reducing TCO came from Oracle and Sun Microsystem’s Network Computer (NC). Unlike the PC, the NC is a low-cost diskless workstation, tied to a powerful server. All system maintenance, upg rades, and configuration is done on one central server and is immediately accessible to all users – cutting the cost of ownership problem at its source. Moreover, the NC is not tied to a specific CPU architecture or operating system (OS). This OS and CPU independence is driven by the requirement of a Java Virtual Machine (JVM). Designed by Sun, JVM is software that interprets applications, enabling them to work with any OS or CPU architecture.

Since the promised NC has no disk, or hard drive, everything is stored and loaded over the network. As most applications and system software reside on the server, upgrades and maintenance can be centrally located and managed. Users can be "locked out", which essentially prohibits them from "futzing" with their systems. In all, these benefits aim to reduce total life cycle costs and increase network security.

Additionally, the NC had a major non-technical selling point: it didn’t use any Microsoft products. This first rumbling of a challenger to Microsoft’s domination of the desktop PC appealed to the "Anyone But Microsoft" (ABM) group that existed within the software community. This appeal to the ABM set, and the subsequent public barbs thrown by Sun and Oracle, guaranteed that the NC received a lot of press that concentrated on the fight with Microsoft more than the substance o f TCO.

Initial analysts’ estimates claim cost savings from 25% to 79% over traditional PCs. But by its proponents’ own admission, TCO is not the major selling factor for a Sun NC. According to a Zona Research study, the major appeal of NCs in the marketplace is ease of administration 24%, versus lower total cost of only 17%. Early price estimates for the NC outlined by Larry Ellison, chairman of Oracle Corporation, came in around $500.

The combination of the Gartner study and the McNealy article drew a lot of attention at Microsoft. It became "a Bill Gates-level issue," and Microsoft launched a full-scale attack on the TCO problem, addressing software as w ell as hardware issues.

Microsoft’s hardware response was to support the Network PC (Net PC). The Net PC is a low-priced, stripped-down PC, with uniform hardware configurations in a "sealed case," so users in corporate environments can not open the box and install their own options. Moreover, the Net PC can be remotely managed and controlled, so corporate IT departments can take back the control they lost with the decentralized standard PC model. This would be a major step forward in solving the &qu ot;expert user" problem, where key employees in corporations were distracted from their jobs while "futzing" with their PCs. The Net PC would compete with directly the Network Computer (NC) standard that Sun, Netscape and Oracle were pushin g. Unlike the Network computer, the Net PC would include a hard disk to increase performance and reduce network traffic. The initial price range for a Net PC is estimated to be below $1,000.

Although the war has just begun, some early battle reports are in. Estimates of the relative market sizes and potential for success of the two opposing standards varied:

• The total market size for commercial NCs is projected to grow from approximately 1.7 million in 1997 to 6.7 million in the year 2000 (See Exhibit 3).
• In comparison, commercial PC shipments are estimated at about 100 million by the year 2000.
• Shipments of consumer NC devices are projected to expand from 2 million in 1997 to 70 million in 2000.
• The corporate market for NCs will exist primarily of replacement purchases, with NCs replacing terminals of all types.
• The TCO argument presented by the "Anyone But Microsoft" group will not be the "significant force" causing customers to switch to from PCs to NCs.

Hints that the Network Computer may not quite be the simple corporate solution to the Total Cost of Ownership problem have also started to surface

• The initial cost of NCs is higher than promised. "Under $500" was the promise. "About $1000" is the reality. The first NCs cost about as much as a low-end PC.
• The maintenance cost of NCs is higher than promised. With the PC model, you spend a lot on the clients (the desktop machines). With the NC, you spend a lot on the servers and the network infrastructure. Either way, you spe nd a lot. As The Economist puts it, "..network computers may often cost a company almost as much as … PCs. Add the cost of switching, and they may cost even more…"
• The compatibility of NCs is less than promised. The first NCs lock users into proprietary systems. Sun’s JavaStation requires a Sun Netra server to boot up. (Once up, it can browse any TCP/IP-compliant server.) What happe ns to the dream of mixing and matching clients and servers from any vendor?

Even Sun’s CEO, Scott McNealy recently wondered out loud about the willingness of sophisticated corporate users to accept a Network Computer instead of their own PC: "Wouldn’t you feel insulted if you were given a $500 NC?" ; This just goes to reinforce the opinion of John Jones, a Solomon Brothers analyst who states, "The salaried workers will continue to use their PCs, while the hourly workers would go to NCs." And finally, "Forecasting that 50% of a compa ny’s users will give up their personal computers to get network computers is like forecasting that 50% of the people who drive to work will suddenly give up their cars because there is a subway line between their neighborhoods and where they work…The thin g people like about their personal computers is that they are personal, that the computers are theirs."

The need to "solve" the Total Cost of Ownership (TCO) problem became a very high-profile issue at Microsoft. The NetPC was one part of the solution, but major changes in the operating system were necessary to make that a reali ty. Action was soon under way at the top levels of the company to counter the threat from Sun, Oracle and others.

In June of 1996, Paul Maritz, Group VP of Platforms and Applications (See Exhibit #4), hired a technical assistant Kavi Singh. Singh’s sole assignment was to focus on the emerging problem known as "Total Cost of Ownership", or TCO, to investigate the issue and recommend a course of action. To Singh, TCO appeared to be the latest industry movement, the kind of issue that came along every couple of years. Before TCO it was groupware and collaboration. Before that the issues wer e multimedia and the web. Said Singh:

But this particular movement seemed to generate more media attention than usual, and some of Microsoft’s competitors were beginning to use the issue as a weapon against their rival.

Rob Short, a development manager and one of the original members of the NT team, joined the initiative later that year in October of 1996. He spent the next six months working with a team of development people, customer support personn el and VP’s to put together a plan that would address the TCO issue. Once the plan was finally ready (March 1997), teams were put together to assess various solutions. Short felt that the threat posed by Sun was the spark that really got Bill Gates movi ng on this issue:

Chris Gibbons, General Manager of Enterprise Computing and former CIO, had been dealing with TCO in his past positions and felt it was a natural issue on which to focus. Still, his perspective of the drivers behind TCO at Microsoft was somewhat different than those from Singh and Short:

Dan Plastina, a program manager at Microsoft, would agree that TCO was nothing new. He has been working on TCO and software management issues since 1991. Back then, however, while he and his team had tried to implement management feat ures (read: applications management), they often took a back seat to more visible products features:

Dan pointed to pie charts listing sources TCO factors and stated his belief that TCO for Microsoft wasn’t about transferring costs from one slice to another, but making the whole pie smaller (Exhibit 2):

Initially, Singh, Short and others spent their time meeting with people within Microsoft and visiting with customers, CIOs and MIS managers. Their approach was to have an open mind and to listen carefully for customer need without p reconceived solutions in mind. As the conversations went on, it became clear that the networked PCs solution had hit a sore nerve. There was a set of underlying problems that needed to be addressed. These problems included the difficulty of deploying new software, servicing the installed bases (both hardware and software), and controlling what people did with their computers. IT executives were further concerned about increasing retraining costs generated by upgrade rollouts.

Another issue that surfaced was the notion of soft costs. Soft costs referred to people using their PC’s for non-work purposes. This included game playing, surfing the internet, or "futzing" with the look and feel of the mac hine display. Companies wanted their knowledge workers to have full access to computer resources in order to learn new things and to become smarter. They also wanted employees with task oriented jobs to have machines with different functionality than tha t of the knowledge worker. It was clear that different levels of access and functionality were needed by different employees

Gibbons considered these costs as well as the costs of downtime, which he suggests to be the biggest cost in terms of lost revenue for any customer-oriented business. "As you get into the areas of a business where the empl oyees are closest to the customers, the employee’s costs due to downtime become much more important."

Gibbons was also quick to look beyond TCO. He felt that TCO, although relevant, was only one part of the equation for Information Technology (IT) decisions. Corporations were still looking at IT as a cost to be contained, rather than as an investment whose costs should be evaluated in relation to benefits.

Once Singh had a clearer picture of the TCO issue, he set about addressing the solution within Microsoft. This typical Microsoft approach to a major initiative, involves assigning someone smart to the issue who collects information and analyzes the problem. They then evangelize a solution within the organization. It is an engagement process which sometimes gains support and momentum, and sometimes not. Said Singh, "In this case…the solution was resonant enough with people…that we were able to go a long way."

Building that support was helped by a number of key people. Bill Gates asked his research division to weigh in on the issue and a number of "concept" papers were written by some of the top research scientists in the organizati on. As Short describes, vision emerged of a PC that was capable of doing two very different and distinct things:

The responsibility ultimately lay in Paul Maritz’s (Group VP, Platforms and Application) organization, the group that owned Windows95, NT and Office. A group of people there had pushed the idea, so a team was formed around Moshe Dunie, VP Windows, Rob Short, Dan Plastina, and Frank Artale, director of program management.

The system management people, such as Dan Plastina, and the customer support group under Chris Gibbons had been compiling data for years. "We knew straight away there was low hanging fruit", Rob Short said. "If we could solve the application installation problem, so customers can just put new software on their machines and know it won’t break anything else; If we could fix the problems with networks, and make application management simple (so corporate customers can just upgrade 20,000 PCs to a new version of an application without worrying about "breaking" something else) –that covers about 40% of the customer support calls. Right there".

With Bill Gates pushing the issue, the pressure was on. The team had meetings with Gates every three to four weeks (many of them lasting over three hours) and received top-level aid to clear roadblocks – there would to be no excuses for not pulling this off. Rob Short said that made an amazing difference:

Managing the team itself required a major effort in coordination across multiple product groups. Both NT and Windows95 (Microsoft’s two major operating system products) had to be involved as fundamental operating systems issues had to b e resolved. The Office group (home of Microsoft’s applications products such as Word, Excel & PowerPoint) was a key part of solving the application management problem. The Application Tools group was required to make sure that there would be a sing le standard for application development going forward. The customer support group was a key source of information on what had customers frustrated enough to phone Microsoft. Rob Short took the lead on development and established weekly meetings for the te am of people across all the groups. The number of people involved in TCO at any stage was hard to gauge. Rob Short described the situation:

The development process was made more effective by bringing the technical side closer to the customer. Microsoft had been sponsoring global and regional events to review product ideas. These events, normally organized by the Developer Relations Group (DRG) bring sales, product, and technical people together with customers, in order to formalize the customer feedback structure. In May 1997, Microsoft organized a huge two full-day events and brought in a few hundred developers and corpor ate IT managers to get feedback on the approach to solving the Total Cost of Ownership problem.

In addition to these events, Microsoft has adopted a number of different measures to decrease the cost of PC ownership – not only in the Windows NT operating system, but also in a number of desktop applications. Three of these are Zero Administration Windows (ZAW), Windows NT 5.0, and the development of applications standards for Independent Software Developers (See Appendix 2 for details).

The deadline to ship product was cast in stone. Microsoft scheduled a Professional Developer’s Conference (PDC) in San Diego in September 1997, and promised all 6,200 attendees a beta version of NT 5.0 (which incorporated the TCO featur e set) to take home. With a firm date set the pressure to get the features in to NT 5.0 became intense, and the group realized that they might not get all the features they wanted in to NT 5.0 beta.

A list of priorities became obvious. Without the application installation improvements there was no product. This particular problem was known to customers and developers alike as "DLL hell" and was one of the group’s top prio rities. All Windows applications use software components called Dynamically Linked Libraries (DLLs). Unfortunately a customer could install a new application that wiped out the DLLs used by another program, so the old one stopped working. The developer t ools that the Independent Software Vendors (ISVs) used to develop applications that run on Windows were also considered crucial. Microsoft had to ensure that ISVs could not develop any more applications that did not conform to the new TCO standards. Thes e prioirities were all included in the first beta, and other features were traded off to ensure that.

In September 1997 all 6200 PDC attendees did get the beta release, and by November 1997 100,00 copies of NT 5.0 beta were released to the Microsoft Developers Network (MSDN), a group of software developers, big customers and other Micro soft partners.

After 18 months working for Paul Maritz on the TCO issue, Kavi Singh felt that TCO had been embraced by Microsoft, and that a threat had been turned into an opportunity. As Singh prepared to move on to his next assignment, he considere d the current challenge facing Microsoft to be "application unification." This was a large and difficult undertaking involving the integration of every file, in every application and operating system. This issue, when properly addressed, would further reduce the total cost of computer system ownership.

Short looked at the technical challenges still facing Microsoft:

Plastina talked about the continuing market dialog between Microsoft, Customers (end users), and ISVs:

Gibbons chose to look at the challenge for Microsoft from the "big picture" of the customers:

The tremendous amount of attention that TCO garnered with McNealy’s first pronouncements on the subject only a year and a half ago, and the fact that it has stayed a hot issue since then, are indications that computer users feel TCO is a big issue. Many customers have had a gnawing feeling that IT costs are spiraling out of control for some time, but they have had no methodology for measuring it. The Gartner model, which has become the de facto standard for measuring TCO, quantified t he soft costs that frighten CFOs because they have grown unchecked and unmeasured. As grateful as they are that the issue is finally on the table and being addressed, some customers remain skeptical that Microsoft’s TCO approach will really solve their c ost problems.

An IT manager at a major Microsoft enterprise client believes that while well intentioned, Microsoft’s initiative is years away from really simplifying his biggest PC cost, configuration management. The open architecture and the lack o f good design discipline by application vendors have made the Windows NT environment a conflicted mess. Basic steps that he would like to be able to take like locking the NT directory so users can’t modify it (necessitating later IT fixes), can’t be done . Some programs crash without the ability to modify the directory. While developers are being encouraged by Microsoft to follow stricter TCO compliant program design, it will be at least two years before these cleaned up applications are in use. When this is looked at in conjunction with the large base of legacy Windows 95 applications, and non-compliant Memphis applications to come, the situation is, "so bad that its not really fixable" with current PC architecture.

On the other hand, there seems to be a general customer feeling that while NT 5.0 will not solve the whole TCO problem, it will help with the administrative costs of updating software and operating systems.

The Sun/Oracle response to Microsoft’s TCO proposal is as scornful as the original criticism that brought the issue to the public’s attention. Sun’s CEO McNealy continues to refer to Microsoft’s lineup of applications and operating sys tems as a "hairball." Sun is advocating the JavaStation, its stripped down NC that will run Java programs on Sun computers linked to Sun servers. The end users will not be able to modify the system files or applications, as those will reside o n the server. This configuration would greatly reduce both the initial setup of new applications/machines and their long run maintenance. Currently there is a dearth of applications written in Java, but Sun is confident that this will change. The event ual JavaStation is predicted to cut TCO by 90%. Note that when Sun states that, "The Network is the Computer," they are referring to Sun networks and computers. As one industry analyst noted, "everything [Sun] does is to sell hardware.&qu ot;

Oracle, while nominally aligned with Sun against Microsoft, has its own agenda. Oracle approaches the TCO problem from their position as the leading supplier of relational databases to PCs. They are not so much pushing Java as they ar e pushing their database and server technology. Oracle recently announced that they have developed technology that will allow NCs to use a myriad of applications that are currently compatible with their database. This could kick start the whole platform and make it a viable alternative.

Wyse Technology, the maker of X-Terminals, claim to be the only company now selling NCs. Wyse claims that TCO can be reduced by 50% by using X-Terms instead of PCs. Through the fortuitous arrival of TCO and NCs, this maker of older du mb terminals has suddenly found itself in fashion again. While this does give Wyse some increased visibility, they are not bringing any new technology or proposal to the market.

The popular press’ coverage of the TCO battle is fueled more by McNealy’s quotes than by the technical merits of the proposals. Americans love an underdog, and McNealy is playing the part for all he is worth. Given the fact that neith er the NCs nor their Java applications exist, the battle of the quotes is mostly being waged to win mind share for Java in a Microsoft world. Nonetheless, the press continues to be fascinated by the idea of a credible challenger to Microsoft. It is wort h noting that the TCO underpinnings of the debate and the specifics of NT 5.0 are being more or less ignored in the popular press. In computer related publications, there appears to be appreciation of the new features of NT 5.0 as well as more focused an alysis of the TCO problem.

The consultants had almost as much to do with the rise of TCO as did McNealy and Ellison. Back in 1987, Gartner first sounded the TCO alarm, reporting that the annual TCO of a PC was almost $4,000. This cost increased to almost $9,000 by 1996. When the PC is connected to a LAN, the TCO estimate rises to almost $12,000 annually. Many other consultant have come out with TCO estimates, but Gartner’s estimate remains the standard and amongst the highest. Curiously, Gartner and the other consultants have remained ambivalent about the industry they gave footing to. They have advised their clients that NCs are too risky, and that companies should wait at least a year for some standards to emerge before casting their lot in with either the Sun/Oracle NC or the emerging Microsoft/Intel NetPC. Additionally, Gartner claims that NCs are not really needed to reduce TCO. Simply by using current administrative technology and present IT best practices to more tightly control their current networ ked computers, savings of 25% to 30% can be realized.

Oracle is the leading developer of database management systems software . Oracle8, a faster, more flexible version of the firm's database software that was released in 1997, also provides support fo r a system of network computers -- stripped-down, low-cost computer terminals operated through a control server that company CEO Lawrence Ellison has touted. Oracle also makes application development productivity tools, computer-automated software enginee ring products, and document automation products.

Sun Microsystems is a leading manufacturer of UNIX-based workstations, and a leader in network computing solutions. CEO Scott McNealy was an early advocatoe of "The network IS the computer". Sun is both a hardware and software company selling microprocessors (UltraSPARC), software (Solaris UNIX), Internet/intranet servers (Netra), and workstations (SPARCstation).Sun is the originator of Java, a language for creating applications on the Internet which has become the unofficial industry standard.

Netscape was built around the amazing accessibility it offered users to the web through its browser. Netscape Navigator, the most popular tool for accessing the World Wide Web, has a 50% to 80% market share. Netscape also produces Netsite Comme rce Server, a secure method of paying for Internet purchases. Netscape is battling omnipresent software maker Microsoft to hold its lead in that market. CEO, Jim Barksdale.

Computer Associates , the company that grows by acquisition. It bought 60 companies during the 1980s and 1990s and is now the number 3 independent software company after Microsoft and Oracle. Computer Associates has a huge product range. It’s ma in product, Unicenter, gives corporate organizations centralized control over their software, hardware, and networks. CEO Charles Wang, a competitive billionaire who provides his employees with day care facilities and free breakfasts, owns about 5% of CA. Swiss billionaire Walter Haefner owns over 23%.

10/96-3/97 Formal TCO planning with development, customer support, and

3/97 Development team readies first release build of NT with ZAW

Microsoft’s solution consisted Zero Administration Windows as part of the next release of NT (NT 5.0), and Software developer guidelines and tools to make sure applications developed for Microsoft software "behave" i n accordance to the new guidelines set out by Microsoft.

Identifying the most severe trouble areas for IT managers, Dan already had written a document describing what the NT developers should be working on. Dan’s original document became the basis for the first Zero Administration Windows (ZAW) Scenarios pap er. Rob Short contributed to the ZAW paper and this in turn gave it high visibility and helped propel the initiative. According to Rob:

There are essentially three prongs to ZAW: the Zero Administration Kit, the System Management Server, and Microsoft Office.

The ZAW initiative sought to address TCO issues on existing platforms by adding augmented functionality to an established install base. The next phase of Microsoft’s TCO response is the development of Windows NT 5.0, which is expected to reduce desktop TCO by as much as 50% over any other Windows-based platforms. A number of Windows NT 5.0 elements that will be leveraged to obtain these cost reductions are outlined below.

• Automatic System Upgrade: The client operating system automatically updates on boot, essentially downloading the most up-to-date version of the software. It is also possible to remote boot a client from the network server, m aking network maintenance easier for the system administrator.

• Automatic Application Installation: All software can be distributed and installed remotely from the network server, transparent to the client user. Microsoft has essentially integrated into the operating system a more powerf ul version of the installer application developed for the Office Suite. The operating system provides improved management of application files, version, associations, and references for consistent installation and removal.

• Microsoft Management Console (MMC): A single GUI interface is provided for systems and network managers. System management tools can be developed as applets that "snap" into the MMC interface, providing a central m anagement tool for network administrators. A number of other system management tools, such as Windows Management Instrumentation (WMI) and Web-Based Enterprise Management (WBEM) also provide more central power and reporting ability to information systems administrators.

• IntelliMirror: As stated in Microsoft’s press release introducing the technology, "IntelliMirror PC management technology will give customers of 32-bit Windows the advantages of centralized computing with the performance and flexibility of distributed computing. It mirrors users’ data, applications and customized operating system settings to a Windows NT 5.0-based sever using intelligent caching and centralized synchronization. This means that users have access to all of their information and applications, whether or not they are connected to the network, with the assurance that their data are safely maintained on the server." This technology, in essence, saves all user settings, data, and applications on a server, leaving only cached versions of this data on the local client. Should hardware failure occur, or the user roam to a different site on the network, all information is cached to the new client and replicates the user’s home environment.

According to Dan, what some at Microsoft wanted to do was to hype up ZAW and promises of NT5. Instead, Dan argued for a quite approach so that: 1) they were able to listen to customers and deliver what is was that customers wanted, and 2) that they could make sure Microsoft could deliver what customers wanted before anything was promised. Applications developers too didn’t want Microsoft putting empty performance promises out in the market.

Microsoft has tried to be as transparent as possible when replying to customers wants. "Look here’s what we can and can’t do with NT." But Microsoft also acknowledged that they couldn’t satisfy the customers without industry compliance and support.

Having put considerable efforts into reducing TCO through a well-planned operating system, the third portion of the TCO initiative was directed at independent software providers with the development of "clean" applications. T hese "clean" applications leverage the TCO characteristics inherent in the Windows NT 5.0 operating system to offer "optimal end-user experience and functionality". Microsoft holds developers’ conferences each year, during which it pr ovides recommendations for the development of "clean" applications – such recommendations include utilizing the Microsoft installer for application deployment, building MMC snap-ins (or augmenting the functionality of existing snap-ins), and uti lizing management data generated through WBEM. Microsoft challenges software developers to identify the Top-10 Help Desk calls for their products, and to address these problems in subsequent product releases. Microsoft has implemented a branding program in which applications that meet minimum requirements for addressing TCO concerns can be marketed with a Microsoft Windows logo. This logo, it is hoped, will be used by IT managers to identify "clean" applications for enterprise network environ ments.

While logo’s have historically been used mostly for marketing purposes, Dan and others at Microsoft see their new Logo initiative as a way to encourage working guidelines in the industry. In discussing ISV incentives to comply with the new guidelines Microsoft personnel cite: 1) customers demand reliability, 2) cost to serve is too high for Microsoft and ISVs (tech support, etc.), and 3) ISVs can sell upgrades (new feature now being replaced by new reliability). As Dan puts it: