Wiley - Data Modeler’s Workbench
更新日期: 2005年12月04日
文件大小: 2.669MB
图书格式: Adobe/PDF
图书语言: 英语
内容简介
It would be interesting to draw a map of all the challenges faced by professional modelers in their day-to-day work and then to assess how well each one is addressed by the literature. It would be a very uneven picture.
Normalization would probably win the prize for the most-addressed topic. Indeed, an outsider could easily conclude that data modeling was primarily about the resolution of complex normalization problems and that the prime tools of the data modeler were relational algebra and calculus.
Languages and diagramming conventions would be more than adequately covered. There has been a steady stream of alternatives proposed over the years, and a great deal of debate—sometimes passionate debate—as to their relative suitability. Thanks to more recent publications, largely by practitioners, we would see some useful contributions in the previously barren areas of data model patterns and business rules.
These are important topics, but they address only a fraction of what data modelers actually do. The choice of conventions is usually a one-time decision, often made by someone else. If normalization is done at all, it is usually as a final check on what has been done intuitively. Thirty years of research since the original articulation of the first three normal forms has made little difference to the models that are produced in practice.
What is missing in this lopsided map is material on the process of data modeling: the practicalities of project planning, communicating with end users, verifying rules and assumptions, presenting models to diverse audiences, even convincing stakeholders that data modeling needs to be done at all. This is the messy, inexact stuff that consumes the time of real data modelers—and very often determines whether a project will succeed or fail.
The glib response is that we learn these things from experience, rather than from books. This should not be an acceptable answer. True, many of these topics may be unattractive to researchers: they are not easily compartmentalized, they require a detailed knowledge of practice and organizational settings, and there is seldom a single solution. But there is a real need to codify experience if we are to avoid reinventing the wheel or paying for data modelers’ education with failed projects.
Steve Hoberman has tackled a large uncharted area with this book. He identifies a range of challenges facing data modelers, and this is an achievement in its own right. Many of them are common problems that we all face, but have not always stopped to think about or discuss.
