Lean DevOps for automotive OEMs
Examensarbete för masterexamen
Quality and operations management (MPQOM), MSc
The accelerated software development has forced researchers and industries to reorginaze in terms of the software development process. One such recent philosophy in software development is DevOps. DevOps focuses on collaboration, continuous integration and continuous deployment amongst other things, with the intention to increase the cycle speed between feedback from end-user to an updated software. The intention with this research has been to investigate how a DevOps implementation can be achieved at an automotive OEM and what potential challenges and consequences there are of such implementation. The current software development process, from requirement to development was mapped with the help of three interviews with people that have a long background within the automotive industry. Based on the interviews, a approximated model was created to illustrate the challenges with the current centralized approach in the industry. The results show how such an approach is slow with little collaboration and the consequences are costly adjustments, slower speed to market and less customer satisfaction. With DevOps and OTA updates however, this process changes into being more responsive whilst additional functions can be added after the release of the car. Many OEMs in the automotive industry are dependent on their suppliers and to achieve a DevOps implementation, the OEMs must somehow integrate together with the suppliers in a CI/CD pipeline where the developers are located at different suppliers, and the OEM has an operations team that monitors the different car fleets and decides when a new release is going to happen. The challenges that come with an integration of DevOps is how this re-organization can happen practically since the automotive industry is highly centralized and someone with enough power must be convinced that this is the right decision. Furhermore, with this integration and closer collaboration between suppliers and OEM, a new culture will emerge. Having a cross-border culture regarding DevOps and having another internal culture within the own borders can be challenging. Finally, there must also be a way to protect the different IP that belong to the suppliers. The IP is usually different algorithms in the software that is compiled and delivered to the OEM. If this protection is not achieved in the CI/CD pipeline, it will be difficult to persuade the suppliers to agree to this kind of integration. There are certainly many challenges in terms of organization but with software being the future of automotive, there should definitely be a push for increasing the software development speed. That is however only one part of the whole infrastructure and supply chain since there are other challenges outside of this research concerning more technical detail regarding the CI/CD pipeline and what is required from the servers and cars for this to be released OTA.
DevOps , automotive , software development , continuous integration , continuous deployment