|
Examples of server-based computing failures are, unfortunately, not in short supply. They often occur when an organization implements a server-based computing pilot or beta with a goal toward enterprise expansion, but then forgoes the rollout. Many organizations approach a Terminal Services implementation from a PC networking perspective. Although it is sometimes possible to deploy a successful Windows 2003 Server or Novell network without extensive planning and piloting, this will rarely work in an enterprise server-based computing deployment. Think of installing an enterprise SBC environment as replacing employees' PCs with a mainframe. Both cultural and political aspects are added to the technical challenges to make unplanned deployment nearly a guarantee of failure.
IT often underestimates the impact of SBC on the various cultural and political aspects of the organization. Preparing for the organizational change as described earlier in this chapter is a key component to a successful enterprise deployment.
Many organizations skip the pilot, project definition, and infrastructure assessment steps and go straight to project planning, or even a beta. This is bound to be troublesome if not an outright failure.
The proof-of-concept pilot is essential for testing all applications under server-based computing before implementation. Proceeding immediately to a production pilot or beta can leave users frustrated with application performance or reliability or both. Even a small number of frustrated users can provide the type of negative feedback that will quell any further server-based computing expansion.
Some internal evangelists might be sold on the idea of enterprise SBC and persuade management to implement one without enough thought to the objectives, scope, roles, risks, and success criteria. Without a project definition document, the planning, project management, and implementation teams have no touchstone with which to keep the project on track.
Project design committees often like to skip the infrastructure assessment step and jump straight to planning. This tends to be the most enjoyable part of the project, when participants contribute their knowledge to build a solution. Unfortunately, it is virtually impossible to create an optimally effective plan without a detailed infrastructure assessment. Additionally, infrastructure flaws that are tolerated under distributed computing are likely to be amplified in a server-based computing environment. When users become completely dependent upon a central server farm for executing their applications, the infrastructure has to be extremely solid.
Sometimes, even large server-based computing implementations are performed without knowledge of basic tools and methodologies that can dramatically facilitate deployment. We once had lunch with the architects of a 5000-seat MetaFrame project who were complaining about bandwidth issues. It turned out that they had never even heard of the bandwidth management tools discussed in Chapter 6. Using bandwidth management from the start would have prevented their problems.
|