In the health sector we have lots of life and mission critical systems. Changes must happen, but they cannot happen in a way that puts people's health in jeopardy. We also need to deal with loads of unstructured data which needs to move quickly in various unpredictable patterns. Through trial and error we have learned a lot, and we hope that our experience can be of use to others. Therefore, we want to show how our deployment process ensures that critical systems don't need intensive care immediately upon hitting production.

Using Kubernetes to enable our new untested changes to live side by side with the current. Processing the same data and leveraging Kafka to assess the differences in real time. We have built a system that lets us catch the subtle deviations arising from our organic data and usage without need for rollbacks.