The original application had been built by IBM but then moved to another support provider. As a result, when we began working on the application, it was clear that no-one understood how the code worked. Therefore our initial approach focused on understanding the codebase and documenting how and why the application worked as it did.
In order to get the application to a maintainable baseline, we started by replatforming the existing IBM stack on AWS, using test harnesses to verify that the behaviour had not regressed. This involved using Terraform to implement an Infrastructure as Code pipeline, which meant we could boot up the Bichard 7 application with a single command in Docker.
This allowed us to test the old artifact in production, while ensuring that all we had done was documented in the code.