Three Signs of Trouble in Your Database

A relational database is different from flat files that you might create with a spreadsheet or a word processor. Flat files are easy to set up, but difficult to maintain. They frequently cause problems such as these:

1. You store some pieces of information in several places
2. You find "impossible" entries in your files
3. You spend hours comparing printouts when you use your data

If these look familiar, you should get a full diagnosis of your database system.

1. You store some pieces of information in several places

• customer addresses
• employee names
• prices of your products

When you update these items, it is tedious to find all of the places they occur and type in the same information over and over. Afterward, you are never completely sure that all copies are the same. There is also a risk of deleting important data, then finding that there were no copies elsewhere.

A properly designed relational database stores each item in only one table, pointing to that table from other tables of related information. Data can be pulled from sets of related tables as though they were all stored together.

Generally, you can delete a pointer without disturbing the data it represents. On the other hand, a carefully planned cascade delete can keep the database from accumulating mountains of obsolete and misleading data.

Top of page

2. You find "impossible" entries in your files

• orders shipped before they were placed
• employees born in 1999
• orders for nonexistent products

A carefully constructed relational database includes rules for verifying data, so that mistyped or inconsistent entries are caught before they are stored. The same system can also flag entries for special attention; for example, a customer who orders from a new product category or one who waits longer than usual to re-order.

Top of page

3. You spend hours comparing printouts when you use your data

Suppose you want to know whether customers in category X are buying more of product Y than they did last year. You know what you sold and who your customers are, but

• Your customer descriptions are in a word processor file arranged by company name.
• Your completed orders are in spreadsheets, with a separate file for each sales person and calendar quarter.
• These spreadsheets identify customers by codes from another spreadsheet.

Relational databases do queries like this very easily. Query results are not stored, but reassembled each time you call for them. That way, you always work with up-to-date information.

Top of page

Copyright 1996, 2000 by Thomas David Nichols. All rights reserved.

Return to HQS home

Heliotrope Quality Systems

www.hqswebsites.com