When preparing for a project of this size, we start with creating a high–level estimation in which all elements in play are brought to light, without focusing too much on details. Of course, there is never one way to reach your destination, so we also mark and estimate the different approaches, to have those available along the way. With all information on the table, we create a plan (or actually multiple), one being the before mentioned two-stage rocket.  

With the plan in hand, multiple teams were formed. On the side of NS, the stakeholders (like the third-parties) were handled by one team, while our team consisting of INFO and NS people started preparing the ‘internal’ migration. Although we love Scrum, sometimes it’s not the right tool for the job. So we built a backlog and started using Kanban to pick up and finish the tasks. Once the team felt comfortable enough, they started doing migration dry-runs. First on test environments, but later up to production.  

With all stakeholders on board, the third-parties on the Gateway and sufficient confidence in the team from all the dry-runs, a night in September was planned to execute the actual migration, from an NS office in Utrecht. And even though the network went down in the building just as we were finishing the migration (you can’t plan everything), the migration went smoothly and the morning (rental and parking) traffic did not even notice they had switched to the cloud overnight.