PayPal moved from a single large Jenkins instance to using containers to improve scalability and efficiency of their continuous integration and delivery (CI/CD) pipelines. They deployed Jenkins masters and slaves as Docker containers on a Mesos cluster. This reduced resource usage by 10x, saving over $2.7 million annually. Issues with inconsistent environments and slow tooling updates were also addressed by using containers to standardize environments.
Varied workloads can co-exist without impacting one another
We can ensure their’s isolation and the tasks are not tinkered
Before containerization, it’s a painfully slow process for someone to have a new tool install for their builds – they have raise requests, we would go through testing, ensure it doesn’t result in any conflicts and push it to all the slave VMs. With containerization, the users can create their own jenkins slave docker images and use them after going through automated approval process. The containerization/individual jenkins slave docker image ensures no one else is adversely impacted.