How do I generate test data for heterogeneous environments?

Question

I am facing a new task in my job and I need to find out how to generate and administer test data. Googling led to a lot of information about specific test data generation like filling a database with random data or camouflaged production data, generating files, generating test data with multi-objective genetic algorithms to minimize test data and optimize coverage, etc.

But my task is somehow harder, because the environment is not only one database, it's a heterogeneous environment, which evolved over time, consisting of databases, files, different servers, programs, etc. Time shall also be simulated by the files aging and so on.

I am somehow lost here and need some starting points from where I can dig further into the materia.

Do you know any tools, knowledge sources, websites, books, experiential reports or something else considering the topic "Evolving testing environments"?

Answer 1

Sounds like a daunting environment; I'd suggest using a "divide and conquer" approach to identify all the test data variables. Make a list of each element of the environment needs to be varied under test, e.g.

• Database type
• File age
• File size
• Server operating system
• Programs running on the server

(I'm just guessing at the different elements here based on your question). Then, for each element, make a list of values for it, e.g.

• Database type: Oracle, MySQL, PostGreSQL
• Server operating system: Windows Server 2003, Windows Server 2008, Fedora 12 Linux

When you're done with that, figure out which values are most important to test; for example; you might want to prioritize Oracle if 80% of your customers use Oracle.

Finally, you should have a set of values for the different environment elements that you can use to create test environments by using different combinations of the element values, using the most important ones first.