Section I - Overview

The availability of Test Data remains to be a difficult challenge for any testing framework - manual or automated. In order to be effective test data, it must be:

• Realistic

• Timely

• Subsets

• Repeatable

Realistic

In order for the Test Data to be realistic, it must effectively represent the variety, size and complexity of the original data. It must be a accurate profile of the data you wish to mimic.

Timely

Untimely Test Data can directly impact our ability to test our solutions and hinder our deliver time. Waiting for Test Data to be reset after running tests, (e.g.: scheduled batch jobs) is a common point of frustration.

Subsets

Too often, refreshing the Test Data is an cumbersome process. Because of data modeling constraints, (e.g.: relational constraints) and poorly automated designs, (e.g.: database schema or table restores) development teams find themselves parasitically coupled by the data sources they share - even if they don't use the same data records.

Repeatable

We all know that tests fail - otherwise there wouldn't be a need for testing. Resetting Test Data back to its desired state is a basic property of any testing platform. One sniff that can be used to identify this inability, is the continuous growth of data in the testing data source, (e.g.: database size grows at a 15% rate per day). This is because tester create new test data instead of reusing the data.