The wild west of corporate computing.

Welcome to the client-server frontier, where the old rules don't apply and the new ones aren't finished yet. To find out who's on the leading edge and why, one financial executive asked his peers how their companies are pioneering the transition from mainframe to distributed processing.

Client-server computing is causing profound changes, anxieties and opportunities throughout both the information-technology and business communities. Information technology has gone through some far-reaching changes, but no prior computing movement has been as pervasive as the current migration from the mainframe architecture to the multibox, client-server approach.

In a nutshell, client-server technology allows you to distribute processing between a client or front-end device, such as a personal computer or terminal, and a server or back-end processor. The client is responsible for the user interface and data-validation functions, while the server handles the data-repository activities. A multitude of specialized servers may be dedicated to functions such as the management and storage of data bases, communication, printing and naming (for items like telephone lists and data directories).

Senior management often views the client server environment simply as personal computers linked together. In fact, client-server architecture may be a very complex structure integrating many technologies, and adopting a client-server environment entails profound cultural, behavioral and technical changes in both the user and IT areas. The migration path is uncharted and fraught with barriers and false starts. Few organizations know how to move from a large installed base of legacy systems on mainframe processors to a client-server architecture. Past strategies and techniques are unreliable at best. Call it the Wild West of corporate computing.

But although client-server technology is still in its infancy, results from information-systems organizations are starting to trickle in. As an executive interested in the vast potential of the technology, I wanted to uncover the leading-edge and best practices for managing client-server computing, so I conducted an informal benchmarking survey of more than 50 information-technology consultants and CIOs of companies such as Bristol-Myers Squibb, Nynex, Massachusetts Mutual Insurance, International Flavors and Fragrances, Carolina Power & Light, Olin, Wachovia Bank, Mead, Conrail, Unum Insurance and others. Here are some of the questions I posed to them and the answers I received.

What are the most common misconceptions about client-server technology? As early users of client-server technology, the survey participants debunked several myths, such as the notion that the client-server approach is eight to 10 times less expensive than mainframe approaches. Bob Woodrow, director of Nynex's telecommunications group, says, "If you focus purely on the cost element, you can be led astray. If companies spend roughly $3,000 per user for client-server hardware and software, they'll shell out three or four times that for intangibles such as help desks."

How do these costs compare to the costs of the mainframe approach? Bob Wagner, former CIO of Conrail, notes, "If you look at Conrail's budget for 1992 vs. 1993, the company spent more money on client-server technology than on the mainframe technology!" Many of the other survey participants agreed client-server computing can easily be more expensive. As Woodrow says, "You can make the case that client-server technology is 20 percent to 30 percent more expensive if you factor in trouble-shooting, graphical-user-interface experts and support costs."

Why are so many companies pursuing client-server computing? What's the promise of the technology? Frank Kemp, director of IT for the law firm Mudge, Rose, Guthrie, et al, believes "Information technology is pulling in the client-server direction because the legacy system is becoming integrated with the desktop environment - graphical user interfaces, spreadsheets and e-mail." Jim Weber, president of the Omicron Center for Information Technology, adds, "The user is demanding more flexibility to make changes on the fly; [client-server systems] allow the business units more freedom to do their thing." Weber also thinks "The information-systems area has a desire to push maintenance responsibility out to the user."

Client-server technology helps companies address both issues. And John Alexander, former CIO of Unum, points out, "You can extend client-server computing beyond maintenance; users want to participate in the application development processes."

Companies are also moving to the client-server environment because it enhances the user's personal productivity through user-friendly software and desktop devices, improved user interfaces and ability to use new data types, especially image. Plus, the technology is the platform of choice for new applications from system-development vendors. "People who think the mainframe is the only way to go are dinosaurs. The client-server approach is a driving force in the corporation!" Bob Weber observes. "But realize the true benefits are increased productivity and revenues, not cost savings," he warns.

What's missing from the promise of enhanced productivity through client-server technology? Most of the executives surveyed complain that for developing, deploying and managing important production applications, the client-server environment lacks the traditional control, security and expertise of mainframe approaches. Jeff Landau, CIO of Quorum, reports, "We've been implementing a number of large LAN-based, mission-critical systems, and it's like skating on thin ice all the time." To remedy that, "You have to put together a training course for local-area network. administrators, set up guidelines cribbed from the data center and have internal auditors look over users' shoulders every six months to make sure they follow procedures," Alexander suggests.

If we get good tools, will control be easier? The Omicron Center's Jim Weber responds, "The security problems have to be addressed not only with tools but also with each firm's policies from the top and with user awareness. Personal computers are so vulnerable it's scary!" As Weber observes, you need to make users aware that the client-server approach is a shift in information-management culture - from data-center control to user control - and that they need to take responsibility for managing their systems. One CIO says his users think "Back-up is a reason for calling Roto Rooter," which doesn't jibe with successful client-server implementation.

What can you do to ensure success with client-server technology? First, think in terms of your entire organization when implementing client-server systems in your company. Addressing one island, fiefdom or LAN at a time doesn't work, because then you're simply plugging in one-time solutions.

Plus, all the executives surveyed say you should have some type of plan that covers not only the traditional components of information-technology plans, such as platforms, operating systems, software and packages, but also tools, controls, security, back-up, data integrity, training for both users and IT staffs and infrastructure support. Tom Dwyer, vice president of Carolina Power & Light, sees the shift to a client-server environment as a "major change that needs to be planned and implemented well over three to five years."

Be aware that you'll need to balance many aspects of the client-server environment, especially the configuration of your LAN's software releases. "You must control your release levels for your systems to be in synch. Otherwise, your client-server approach will be alphabet soup and chaos," warns Peter Daboul, senior vice president of information systems for Massachusetts Mutual Insurance. In other words, as the distributed environment grows, you need to ensure first that all your LANs use the same software for their operating systems and second, that they all use the same version of that software. The greater number of desktops and servers increases the complexity of your computing environment and thus the potential for incompatibility.

DON'T ROMANCE YOUR USERS

Senior management and users need to understand and buy in to the company's client-server plan. The best way to get this buy-in is to involve your senior management and users in developing the plan so they become knowledgeable about it. Some firms, such as Carolina Power & Light, have employed company-wide councils or task forces of users, management and IT staff. If you don't take the initiative, you'll be forced to do what Ross Anholz, CIO of Olin Corp., calls "romancing each user," or you'll be left with senior-management edicts.

While you're at it, you also should develop a strategy for dealing with your legacy systems. Most firms select.an individual approach for each application in their portfolio rather than one approach for all systems. Two of the most common approaches are re-engineering the business process first and then redesigning the application, and replacing the legacy application with packaged or externally developed software.

Your client-server approach will require new software for application and development tools, performance and network management, and security and control. Individual "point products" are available, but fully integrated product suites are two to three years away from full roll-out. "Client-server technology available today is not industrial strength," Quorum's Landau says.

What are companies doing now to address the need for new software to support client-server systems? Three strategies are emerging: using a software vendor that's pursuing an integrated product suite; developing custom interfaces between "best of breed" point products; and living with the current products now and planning to change in three years when full product suites become available. Most firms are going with the first or third alternatives, although a few cite significant gains from the second.

Again, client-server technology isn't a cost-reduction strategy, and you shouldn't promote it as one in your company. The advantages of lower hardware costs are more than offset by higher support, training and control costs. Support is very labor-intensive because of the lack of management tools and the immaturity of the client-server environment. However, some firms report large reductions in application-development costs and time. For example, John Alexander cites these statistics from his company: "The function points, which are a measure of application-development productivity, run in the 50 to 80 range versus 15 to 30 for mainframe applications - a three- to five-fold improvement in application-development productivity."

Client-server training is extensive and could easily be classified as re-education. Users need training in productivity and application-development packages and tools, and since they may be accustomed to only working with the mainframe, they also need guidance in handling operations and controls. Application developers need training in tools, methodologies, processes and languages. LAN administrators need similar training, as do data-center and network personnel. Kalish Khanna, head of information technology for the Society for Worldwide Interbank, puts it in perspective: "The costs of hardware and software for client-server computing are trivial compared to the training and control costs. You should plan on a major investment in training if you want your program to be successful."

The companies I surveyed use several strategies for handling their staff's client-server development. Some firms develop their own internal expertise via training and project experiences. Other firms supplement their internal staff with experienced hires from outside the company or use external contractors for major new application developments. Cecil Smith, senior vice president of Wachovia Bank, reports his company is trying the latter tactic. A few firms are taking the "Vince Lombardi" approach of having two teams - one planning for today (legacy systems) and a separate team for tomorrow (client-server systems).

A PINCH OF THIS, A DROP OF THAT

However, most firms are using a "blending and sorting out" philosophy, in which they limit training to selected staff, usually those with high potential; use contractors for major projects; hire experienced external personnel; and focus some of their existing staff on legacy systems. Smith has a small core group of highly technical client-server staffers who also have experience in mainframe technology to help his organization bridge the two worlds.

Some early client-server users have experienced a high failure rate (40 percent to 70 percent) in converting their traditional Cobol programmers to object-oriented methodologies and client-server programming, particularly older programmers. One firm with a longstanding Cobol staff even offered a bonus to its development personnel if they completed a client-server training program!

This situation really isn't all that surprising, because the re-education and behavioral issues associated with the transition to a client-server system are profound. For example, employees must go from being individual performers to team players, from having specialized skills to having a broad array of skills, from focusing only on technology to focusing on both technology and people, and from simply upgrading their existing skills to extensive retraining in many different areas. Most executives report that getting the whole organization on board works better than wooing users one-on-one, because you have a better chance of implementing a consistent strategy enterprise-wide when you appeal to users' sense of "team spirit." When you try to win over individuals, they're more apt to press for concessions to their own needs and goals.

Despite some of the difficulties associated with integrating client-server systems into the corporate environment, the technology is, alive in many companies and on users' desktops. It will continue to grow exponentially for the foreseeable future, because the potential to improve user productivity is great. But remember you're heading into uncharted territory. You and your information-systems managers will need to consider computing platforms, application development, programming languages, skills, security, management and a host of other issues before you can successfully settle client-server technology into your information-management systems.

RELATED ARTICLE: THE TOP 10 CLIENT-SERVER MYTHS

Senior management isn't always attuned to the realities of integrating client-server computing into the corporate environment, according to an informal survey of 50 CIOs and information-technology consultants conducted by consultant Sidney Diamond. Here's their list of management's most common misconceptions.

1. Senior management understands the CIO's concerns about client-server implementation, user control of processing, and training and security issues.

2. Adequate client-server staff exists in the marketplace.

3. Mainframe-application developers can readily adapt to the client-server environment.

4. The client-server environment is easy to implement.

5. End users can run critical production systems.

6. End users can secure their own environments.

7. Client-server technology is the same as a LAN operating system.

8. Mainframe approaches will work.

9. Mainframes don't matter.

10. Client-server technology saves the information-technology group money.

RELATED ARTICLE: SERVING UP THE LEFTOVERS

Like many executives whose companies are converting to client-server systems, you're probably wondering what to do with your legacy system. You have plenty of options, ranging from a face lift to a complete overhaul of the legacy system:

* Retain. This means maintaining your legacy system as is. Some companies aren't willing to spend the money to convert their legacies to the client-server environment, particularly if they implemented the system only a few years earlier. They continue to use the system for its intended purpose, but they don't move it onto the client-server platform.

* Renovate. You can marry your legacy system to the client-server platform by incorporating a graphical user interface. In other words, you take the front end of the system - the part that interfaces with the user - and put it on the client-server system, without disturbing the back end.

* Relocate. Here you're transporting your legacy systems from the mainframe platform to the client-server platform, without making any structural changes to your software.

* Surround. Creating envelope or "middleware" for your legacy system is more complicated than relegating. Basically, you take the old application and surround it with the new client-server technology, leaving the core portion intact.

* Re-engineer your code. This is the next step up in complexity. As you shift your legacy system from the mainframe to the client-server platform, you also redesign the software to eliminate any inefficiencies or problems.

* Replace. Replacement usually means completely scrapping your legacy system and putting in a preprogrammed software package.

* Re-engineer your business. Once you do that, you can then redesign the application to suit your current needs.

Mr. Diamond is an information-technology consultant in Stevenson, Md. He was formerly vice president of information services for Black & Decker Co.