Automated and more secure software testing with synthetic data

Adopting DevOps best practices is becoming paramount in big and small organisations to increase deployment frequency while reducing the number of issues showing up in production. Continuous testing is one of the essential elements to achieving that. Ideally, companies should test software on real-life data; however, it is difficult in many circumstances, especially when dealing with personal information data.

During this use case, a public organisation dealing with large amounts of personal data used Clearbox Enterprise Solution to build testing pipelines based on synthetic data.

The challenge

Testing software is becoming increasingly complex as the number of components and microservices used within IT products increases. We might be interested, for example, in checking that the behaviour of a product did not change after migrating to a new infrastructure or that a new user interface is properly working before it goes into deployment. Ideally, we should test each software component and the product as a whole using real-life data. Unfortunately, this is often impossible as real-life data usually contain personal information making its testing usage limited by regulations such as GDPR.

For this particular challenge, a governmental organisation had to migrate a database containing personal data to a new cloud provider. The operation presented a risk as many internal business processes were built on the database. The organisation wanted to make sure the operation would run smoothly by using test data to populate the old and the new database and to compare the behaviour of the old and the new system.

The solution

The organisation used our Enterprise Solution to ingest and clone their production database containing individual data. The cloning process generated several points representing non-existing individuals while preserving the statistical properties of the population from the original database. They finally injected the synthetic population both in the legacy and the new infrastructure databases and compared the behaviour of the two different software versions.

The result

Creating realistic data for software testing allowed the organisation to improve their Continuous Integration/Continuous Delivery processes. A virtually unlimited flow of realistic data allowed them to define a testbed for more granular tests while complying with data privacy regulations. The availability of such a data testbed corresponded to a reduction of incidents occurring with updates and new releases of their software solution.