· Establish top-down support. When you're migrating large systems, organizational conflict seems inevitable. Without a clear mandate from the senior levels of your organization, IT managers will find it difficult to convince others that their cooperation is essential. A credible mandate requires that senior management be engaged in regular discussion of the issues and that there's clear linkage between IT initiatives and business goals and objectives. The resource commitments are substantial, so a statement of the strategic importance of the project, presented in a manner that everyone can understand, is important. Without such top-down support, expect chaos and endless finger-pointing.

· Co-opt the opposition. Even if the systems implementation is seen as strategic and everyone understands the need for client/server IS and business-process re-engineering, the migration will be an enormous threat to those with expertise in existing systems. Watch an experienced administrator hammer away at an obsolete transaction-processing application, magically producing needed results in spite of a cryptic interface. Learning how to use these systems didn't come easy, and many have tied their work lives to the strengths and limitations of the old system.

Somehow, you must find a way to convince these people to buy into the migration, and you need to do it early. Embracing a user-oriented approach to systems implementation is not just the fodder of IS textbooks; it's a matter of project survival. The real challenge isn't deciding whether to involve a large number of users, but rather, deciding which users must be involved and how to control them once they're in. Expect icy relations and be prepared to answer the same questions hundreds of times. Before all is said and done, they will make your life miserable, and you will reciprocate.

·Overbuild the network. Many people, including users and IT professionals, will focus on network issues as they worry about all the problems that might arise during a client/server migration. To a degree, that's understandable. Without a stable network infrastructure, you can't really expect a client/server deployment to succeed. But it's also true that you can't really enjoy success with any IT initiative without a solid network as a foundation. The fact that you're supporting client/server applications is almost irrelevant.

But trying to convince people that collisions won't melt their Ethernet network isn't a battle worth fighting. A better bet is to build a network that won't be a source of problems. Overbuild it, with more bandwidth than you could possibly use and do it with established technologies such as switched Ethernet or Fast Ethernet, reliable routers and functional firewalls. This is no time to venture onto the bleeding edge. Remember that for most managers, quality of service means providing fast and reliable network services.

· Establish a client-software distribution strategy. Among the many technical issues you will face, none will be more challenging than the decisions regarding distribution of client software, especially if you are rolling out systems that depend on fat clients. You have several alternatives. First, you can follow the path of least initial resistance and load all the applications on individual PCs. This will make you popular with so-called power users, who believe this is the way the world should operate because Microsoft told them so. You also will find this to be a long-term disaster unless you can find a way to automate updates.