One only moves a data platform once
More than three years ago, we chose to move our platform to a new datacenter. Our demands and those of our customers had become higher than our old platform could handle. Something had to change structurally to be able to meet the highest current requirements.
We realize that such a data platform move is a time-consuming, complicated, and risky, presenting a high impact on our core business and that of our customers. Our platform consists of twenty years of product and order data for all our customer and links with all the data processes we have built over the years. Such a move is exciting for a company like Onetrail, where everything is about data. So a relaxed relocation was not possible, but we started on time with a dedicated team of employees. The atmosphere in the operations and development teams were both thoughtful and focused, and big vacations were postponed.
Preparation and testing
During the preparation, we looked critically at the efficiency of the platform, determined what could be improved, and examined our security. We established a new server structure, extensively tested it, and guided our customers in adjusting their systems.
The weekend of the move
At the end of May, we completed the final preparatory work and had a go for the actual move. Of course, we had planned this at a quiet moment, so the business suffered as little as possible.
On Saturday of June 29th at 8:00 am, we shut down all our systems to carry out the actual move. Every team member knew his job. At the end of the weekend, somewhat delayed but well in time for the Monday morning deadline at 6:00 AM, everything seemed to have moved well. On Sunday morning, all data was on the new platform, and all messages could be processed. We started the processing in our new data center and monitored the progress accuracy and remained alert because such a first success did not mean we were there yet.
Moreover, that became clear when the message server crashed on Monday afternoon. Our team started immediately with a setback. Later that evening, everything worked again. That week the operations team continued to monitor the systems with extra attention, and fortunately, no unexpected circumstances occurred. We succeeded, our platform had moved!