Bargains and bombshells: calculating the costs of client server.

Does client/server save money? While many benefits accrue to companies deploying the technology wisely, upfront savings is almost never one of them.

Fortunately, many CEOs have abandoned the pretense that client/server computing is cheaper. They have benefited from the experience of the client/server pioneers who believed the promises, made the investment, and measured the results. Many CEOs simply won't accept the same tired promises ever again.

Nevertheless, revolutionaries who advocate change for change's sake continue to beat the drums for cost savings. We must not underestimate the power of their drumming. They have one powerful and compelling argument they have amplified to support their position.

The argument is thus: Client/server computing is less expensive than centralized computing because of the emergence of cheap yet powerful microprocessor-based computers. Yes, if you measure the costs of processing data - a common unit of measure is MIPS (millions of instructions per second) - microprocessors represent a less-costly platform than mainframes. When comparing the raw costs of doing the same work on mainframes versus a network of microprocessors, client/server wins every time.

Unfortunately for many organizations that blindly started client/server projects in anticipation of lower costs, the argument fails to reflect reality. A comparison of costs for client/server and mainframe environments compiled by three independent research organizations (see chart) shows that client/server costs per user are either significantly higher or on a par with mainframe costs. In both environments, hardware costs represent a small fraction of information technology expenditures. People and systems management costs consume the largest shares of any budgeted activity. It is no different in client/server technology. When broken down by function, the infrastructure costs of client/server pale in comparison with costs for training and systems support and maintenance (see chart, next page). Initially at least, client/server computing adds expense to every element of processing and delivering corporate information. In the context of real people doing real work, client/server imposes costs on an organization at least equal to and, in many cases, more than current expenditures.

Nor should it be assumed there is any inherent incompatibility between client/server and mainframe systems. In most cases, new PC-based applications will be integrated with existing mainframe-based systems.

In this common scenario, costsare a function of the resulting complexity, in which desktop clients, Local Area Networks, and midrange and mainframe servers cooperate to give users seamless access to information across the enterprise.

Client/server computing merits attention not because of what it costs, but because of the value it adds. It is for this reason that even if a client/server implementation does cost more than expected, the effort is nevertheless frequently welcome. That's because the benefits of client/server applications are so compelling that organizations willingly make the investment.

WORST-CASE SCENARIOS 

CEOs prefer to know the worst-case scenario and have the most conservative cost figures when evaluating a project. When executives are prepared in such a way, they can be quite tolerant, even when actual results fail to meet projections. This surprising result is reflected in a 1994 IBM Consulting Group study of 24 companies that completed at least one client/server project. The participating companies were drawn from a cross-section of industries, including manufacturing, transportation, insurance, and finance.

More than 84 percent of the executives surveyed expressed satisfaction with the results of their client/server projects. This is astonishing in light of the following reports from the same survey:

* 68 percent said their client/server projects took longer than anticipated.

* 88 percent said the projects required redesign.

* 55 percent said the projects cost more than expected

Why did many CEOs express satisfaction with client/server even when projects took longer than expected, ran over budget, and did not immediately reduce information technology expenditures? They recognized that business units derived real value from their client/server initiatives. The study found that companies realize the most significant client/server outcomes when users take advantage of their access to enterprise data to create new information. This can result in new ways of enhancing the management of existing businesses. It also can lead to the identification and creation of new revenue opportunities, broadening customer relationships or promoting new products or services to existing customers. The study clearly shows these benefits - not projected cost savings - are driving the move to client/server.

The implication is that if an organization is asked to justify a client/server investment on the basis of saving money relative to current information technology expenditures, it almost certainly will be disappointed. However, a key step in the survival process for implementing client/server computing is understanding how much it will cost. This is not a trivial task, as the elements needed to develop a solid client/server costing model are not well-understood.

Managing these systems is perhaps the most formidable challenge of client/server and results in the greatest unexpected costs. Much of the cost of any new client/server implementation derives from management issues such as data security, disaster recovery, storage management, software distribution, and the related labor costs involved in these indispensable disciplines. These costs are well-understood for mainframe applications, but they often are overlooked and nearly always underestimated when planning for client/server computing. As enterprises develop client/server applications, they reap a number of economic and performance benefits, but they also must struggle with an equal number of thorny systems management issues.

The cultural issues of client/server computing, for example, are just as complex as the technical issues. As organizations evolve away from a centralized computing environment, business users from CEOs to telephone operators need to formulate a new way of thinking about using information, computing blurs the distinctions among employees, customers, and suppliers. The concept of computer professionals begins to lose focus. In a client/server world, every member of the organization takes personal responsibility for a portion of a system that previously was administered by computer professionals and off-limits to everyone else. The role for CEOs is daunting: They must prepare their organizations to navigate these changes if they are to remain at the forefront of adding value through information.

WESTPAC'S HARD LESSON

When new technology is not well-grounded or partnered with an appropriate business application, failure that impacts every corner of the organization is assured. That's what happened at WestPac Banking Corp., a $4 billion bank holding company in Sydney, Australia. Following setbacks including a painfully expensive runaway software project, the chairman and four members of the board of directors of Australia's largest bank were forced to resign in 1993.

The old management team's five-year plan, according to a four-color brochure promoting the project, was to "create a showcase of decentralized data systems, allowing branch managers to spin out new financial products rapidly using client/server, CASE (Computer Assisted Software Engineering), and expert systems technologies." Before we go on, check your buzzword counter. Anytime you encounter three buzzwords in a row, you know it's time to grab your wallet.

Before it was over, the $85 million project had turned into a $150 million debacle that produced nothing except massive layoffs. In 1993, WestPac reported its first loss in 175 years of operation. Analysts blamed much of the bank's poor position on ill-advised property loans, but the fact that the bank trimmed 500 jobs from its computer department is evidence that the project was so complex it was probably impossible to manage. Industry observers speculated about the failure of the WestPac project. Some blamed the project management, and others blamed the tools. But ultimately, as it must be, the chairman and the board accepted the responsibility. Now with a new CEO and CIO working together, with a vastly trimmed IT staff, WestPac is back on track and reporting good results.

RELATED ARTICLE: FOUR WAYS TO SAVE

Client/server may not cost less, but that doesn't mean it has to cost more. Following are four approaches that can minimize some of the costs associated with this environment.

1. Minimize the number of vendors. When things go wrong, fingerpointing is expensive. Much of the extra cost of client/server management comes from incompatibility among multiple vendors' products. Minimize the number of fingers and spare yourself frustration. You'll also save by having to hire fewer specialists.

2. Centralize systems management. Every site that has a computer requires a manager. The easiest way to minimize the number of managers is to minimize the number of management locations. If you can get a single computer to intelligently monitor a whole network, you're really saving money.

3. Use integrated management tools. You can avoid the need for some high-cost people by using easily customized off-the-shelf tools that monitor all your software platforms. Using such tools is less expensive than developing your own.

4. Train existing staff. It's tempting to avoid steep retraining costs by hiring consultants to run the new systems. But you'll save in the long run by enhancing the value and loyalty of existing employees who, after all, then will be able to manage all your enterprise systems, the old as well as the new.

Even small efforts in these areas can have a significant long-term impact on support costs. And support tops the list of unexpected costs.

RELATED ARTICLE: THE CALCULUS OF CLIENT/SERVER

The concealed, ongoing costs of client/server management must be considered in any economic analysis of a software project. When a system's management and administrative casts are factored into the equation, client/server systems often ore seen fo be as costly as traditional systems. In other cases, client/server systems actually represent a bigger hit on the bottom line. To get a ballpark estimate of the costs of deploying client/server, here is a simple worksheet. All you have to know is the number of end users to be supported by the system.

1. Estimate the number of end users: _____

2. Determine the cost factor based on the level of sophistication you want.