An integrated applications and data environment across the enterprise is the vision of many a CIO (or CEO). How wonderful it is to enter data once and have it immediately available everywhere. Consistent rules, appearance, documentation, and training are excellent goals, even more so if the organization simultaneously adopts best business practices. If this only requires purchasing and implementing a family of applications (such as Baan, Oracle, Peoplesoft, or SAP) rather than building it--wow! If this is your dream, or your management’s, or if you work for an integrator--especially if you think getting there is easy--you need to read this book almost as much as every project manager needs to read The mythical man-month [1].
Enterprise-wide integration is made difficult by the same things that always make software implementation difficult. They have little to do with the software itself. Rather, as the author discusses, understanding the problem, defining the solution, developing new procedures, converting, user support, and staffing are the hard part. As the author states again and again, proper project staffing is critical.
Two areas that lead to the failure of many large projects could have received more attention: deciding how much of the enterprise to include in the project, and coping with resistance to change. Making the scope too large risks having requirements change before the new system is implemented and increases the number of interfaces. The pervasiveness and strength of resistance to change are usually seriously underestimated by logical and rational systems personnel. There are no easy answers to either, but failure to continually address them is dangerous.