The best way to predict the future is to create it: Production Digital Twin
With 20M lines of code and around 600 new features released every quarter, our financial software, that has been growing for more than 35 years, is far from being considered a simple web application having several instances easily running: it requires dedicated and powerful physical machines for heavy computational work, for sometimes 1000+ users in different time zones accessing the same application instance, with an activity reaching often 1M trades per day for large banks.
It is a big challenge to predict how our product will behave when deployed in a complex client production context. However, it is crucial for us to know and adjust accordingly to maintain our clients trust in delivering the expected quality. Since we could not acquire a client environment with its users and production activity, our only alternative was to create it. Consequently, the company invested in setting up an internal production environment and I was leading this initiative.
Another key challenge that I faced, on top of the operational side, was the need to engage our entire software engineering community:
How will they access the environment?
How will they play their activity?
How will they configure their modules?
How will they raise incidents and debug?
All simultaneously, without conflicts?
I am looking forward to taking you through the creation journey: the seed environment setup, the users’ activity simulation with the client volumes and the environment enrichment process and its deployment and operating models. I will also go through the challenges of engaging the testing community in using this environment, gathering their feedback, and improving along the way. In other words, my presentation will describe our own interpretation of the Digital Twin concept, that we call Production Digital Twin.