J-Fall 2022: Arjen Jansen & Marije de Heus – How to migrate a mission critical railway system

This is simple, just moving trains around through the Netherlands. Making the schedule and execute this plan. Right, of course solving the schedule jigsaw is a real hard job, but when it’s done, drive around. But just being on the way, a catenary cable breaks, a tree falls down on the railway, or a broken train blocks the other train traffic. Planning messed up! How to take care that travelers are still being served. This requires action for planning adjustments in near real-time. And systems keeping track of the actual railway network state from many data sources. Use all this data to choose the best option for optimal traveler service. Additionally, zero tolerance for downtime and continuously informing many other parties about any change in train movements. To fully support this, we develop micro-services using contemporary and innovative technologies. Our services are deployed on a RedHat Openshift environment within clusters stretched over multiple data centers. Everything in a redundant setup, scalable and the strictest HA/HP requirements. Our biggest challenge is to build the new system while the trains must remain in operation. This means that the old IT system has to stay in production during the development of the new one. Due to the complexity of the domain and cooperation with many surrounding systems it is not possible to take over part of its functionality in the early stages of our development. In this talk we will elaborate how we validate the new system in production with 24/7 live user feedback. We will show our pipeline from test to production, which allows us to add new functionality and fix defects in production several times a day.