It's not an ordinary release; it's a BIG BANG!
The big day has come at last!
It's finally the release day of the Baumappe project we developed for our long-term partner Heinrich Schmid. But it's not an ordinary release; it’s a BIG BANG!
This weekend, we implanted four new major modules into the existing system. And this was the result of a year-long hard work of two different teams and 14 people in total. Here at Atolye15, most of us work remotely. As a matter of fact, our teammates involved in this project live in five different cities. But we wanted to gather everyone at our headquarters and turn the BIG BANG into a festival. Because we knew it would be quite a stressful release that would affect thousands of employees and partners of a giant company in Germany. It had to be a release with zero room for error.
Our preparations started weeks in advance, and we had full confidence in ourselves. In a period of almost one month, we successfully eliminated all of our deficiencies, made effective plans, and created simulations to make the transition in the safest way and in the shortest amount of time possible. We carefully dissected and closely examined everything to extract the valuable data within the system. With the simulations we ran, we had the chance to foresee and get rid of all the errors we might have experienced during the transition. Overall, the source of our confidence was these tests and simulations.
In the early hours of the morning, the whole team got together in the office. Following the breakfast prepared by our dear chef, Seço, we had a short meeting and got to work right away. After days of meticulous planning, we started the process by turning on maintenance mode with LaunchDarkly. Upon completing every single of our backups first, we updated all of our environments using Vault. Then we proceeded to gradually deploy our microservices and micro-frontends using ArgoCD.
In addition to the release of these modules, the 3rd party service, which is our main data source, was to be replaced with another 3rd party service. Therefore, the ID references in more than 1 million records in every database of all our microservices had to be substituted with the IDs of the new service. In order to achieve this, we had to transfer IDs from one to the other thanks to the records common to both data sources. And to make that possible, we wrote various scripts and repeated their simulations time after time.
With the help of our high-speed CI/CD process and our dynamic team, solutions to the issues we encountered quickly reached the production environment just minutes after going through our QA process in our staging environment. Meanwhile, our DevOps team contributed by monitoring all the tracings and logs on each of our systems utilizing New Relic.
Besides the manual tests, our QA Team also ran several stress and automation tests on the product. As the final step, our DevOps team updated the resource limits of all our nodes on Microsoft Azure to improve performance.
In the end, everything went off without a hitch. And this awesome team achieved yet another victory. Once and for all, the winners were all of us: our team, our client, and the users.
In a couple of months, another team of ours working on Bookyourdata.com is also going to have a big launch. And in that project too, we’ll have many different concerns and problems to be solved. And we’ll make sure to share the process with you in the same way. Stay tuned!
P.S. A huge thanks to our Marketing team and Chef Seço for turning this stressful time into an unforgettable festival for all of us.