Managing Multiple Databases Across Heterogeneous Hardware and Software Systems - High-Performance Web Databases: Design, Development, and Deployment - page 476

this problem; the problem is widespread — probably tens or hundreds of
data items. Some of those items will be used for critical business decisions.

When the information is summarized and reported to top management,
conflicts in the information will become obvious, though it will not be ob-
vious which specific data items differ, only that the information from one
department does not square with another. Confidence levels in the integri-
ty of all the databases will drop and the decision support results will be
minimized.

Although a single, central database is preferable, the reality is that mul-
tiple databases exist. They come into existence for any number of reasons:

• Independent purchases. A user department buys and uses a separate

system because it believes that is the best answer to its needs or that
IS cannot address its informational requirements within an acceptable
time frame.

• Legacy systems. The system has been in place for some time while IS at-

tended to more urgent matters. Eventually, the need for some form of
technical management of the data becomes evident.

• Acquisitions. The company has just acquired a new division that has

its own database system.

All the problems cited can be avoided if the databases, although physically
separate (and possibly residing on different hardware and software plat-
forms), are made to function as a single database. In other words, the up-
date path for one database is the update path for the others. This
minimizes all the problems except for data redundancy. Exhibit 1 illus-
trates, from a user's perspective, how multiple databases can be physically
separated yet conceptually linked together.

The remainder of this chapter addresses how to build the bridge from
one database to another so that they function as one. Although the scenar-
io described thus far considers two databases, more may be involved. The
techniques suggested in this chapter will also work with multiple separate
databases.

THE MANAGERIAL CHALLENGE

Although there are substantial technical considerations, the primary
challenge is managerial. The reasons are threefold:

• Departmental and functional areas will cross in the formation of the so-

lution.

Without senior management involvement, turf lines may be
drawn and the entire project risks becoming mired in political in-
fighting.

• The lack of detailed data definitions can cause the cost of the project to

go up and the effectiveness of the solution to go down. This activity is