Information Technology and the Firm 549to store information, those database management systems were, in many cases,
unwieldy and very expensive to both acquire and maintain. They were usually
hierarchical or network database systems that, alone, cost in excess of $200,000
and frequently required special database administrators just to constantly fine-
tune the system.
Today’s database technology is based on a relational model, and, on a very
simplistic basis, it resembles a spreadsheet. In a relational database, there are a
series of tables or files. Similar to a spreadsheet table, each table has columns
with attributes and rows of data. The difference is that there is only one table
in a spreadsheet, whereas there can be an almost unlimited number of tables in
a database. In addition, there is a practical limit to the size of a spreadsheet,
but databases can contain thousands of columns and millions ofrows of data.
In addition, databases also allow users to relate or connect tables that share
common columns of data.
Exhibit 16.6 is an example of a very simple portion of a payroll applica-
tion. There are two different tables. The employee table contains data about
each of the company’s employees: name, address, marital status, number of de-
pendents, and so on. The pay table contains data about every time each of the
employees is paid: their gross payroll, social security taxes, federal withholding,
state tax, and so forth.
First, notice the common column between the two tables, the employee
number. This column enables the database management system to relate the
two tables. It allows the system, for example, to print a payroll journal that has
both the weekly payroll information from the pay table and to access the em-
ployees’ names from the employee table. Why not combine all the data into one
table? Not only would the employees’ names, social security numbers, and
other information appear multiple times, requiring the unnecessary use of data
storage, but also multiple versions of the truth might occur. If one of the em-
ployees should happen to change his name or address (if address were included
in the employee table), the database would show one name for part of the year
and another for the rest of the year. Redundant data creates opportunities for
data corruption; just because data is changed in one table, that same data is not
necessarily changed in all tables. Prudent systems design eliminates data field
duplications wherever possible.
DATE WAREHOUSE
Data warehousing attempts to reconcile and live with past applications soft-
ware, while still benefiting from today’s newer technology. As mentioned
earlier, industry is rife with older legacy systems that are currently cost pro-
hibitive to replace. Most of these older systems are mission critical operational
control systems (see Exhibit 16.3) and satisfy most of the operational needs
of the company. However, they are built on technology that cannot support
the kinds of decision support tools that management requires. Many of these