Scenarios for using SelExL

SelExL can be used in many situations - here are a few examples:

Bug Fixing

• A problem is identified in the production system e.g. a number of accounts fail to bill in a telephone billing system during an overnight batch run.
• The next day SelExL is used to extract these "problem accounts" from production and load them into a small test database.
• A member of application support now analyses the problem using the "real data" in the test database and develops a fix (code and / or data fix).
• The fix is tested against the "real data" in the test database and regression tested if required.
• The fix is safely implemented into production.

This approach is much more efficient and reliable than a programmer using a test system that does not contain the specific data causing the problem.

User Training

Create appropriately sized training databases with data targeted at specific training scenarios. Trainers can decide exactly what data they require in the training database. For example in the case of a banking mortgage system the trainers can ask IT to provide a training database with a spread of mortgage accounts across all types of mortgage and account state.

Data Warehouse Development

Use SelExL to populate test databases for each of the source systems for the data warehouse. This allows the development team to test the extract and transform code against real production data for each source system. The data chosen to populate the test databases can be "random samples", say 5% of production, a specific set of data based on a provided list of account numbers, customer ids etc. or even a combination of both.

System Test

Generate a "random" sample of representative production data for system test. Use SelExL to extract a small (say 1%) amount of data from the production database e.g. all account numbers ending in 01.

End User Investigations and Analysis

Occasionally end users need to conduct some initial investigation work on the system. For example to look at numbers of customers with various attributes such as sales profile, location and alike. In this sort of case it maybe inappropriate to carry this work out against the production database due to potential performance issues. Use SelExL to extract and load a sizable sample (say 10%) from the production database into a test database. It is then a simple case of performing the investigation against the sample database and scaling up the results.