How to run 4.5 Million tests per day ... and why!
Given the size and scope of Google’s code base, and speed of development, typical off-the-shelf continuous integration are unable to meet our needs. So, we decided to create a continuous integration and testing system as a centralized service on an unprecedented scale. When fully completed and operational, it will probably be the world’s largest continuous integration and testing system, running millions of tests every single day.
In this talk, we will report on our experience running such a program in an agile manner and will also describe the basic design and features of the CI system.
The session will be a hybrid session between experience report (how to run a large agile program in a big corporation) and describing our continuous integration system itself. The session will follow the last 1.5 years from inception to development to rollout and end with the current status, findings and plans. The system itself started fairly simple and evolved over time to scale and to satisfy various requirements. By following the timeline, participants will be able to a) understand how we ran this program, b) follow the evolution of the system, and c) get an understanding of the system (how it works and how we achieved its massive scalability)
Because of its width, the talk will address a wide variety of personas and fits multiple stages. That and because participants are always very interested in Google, this talk would be a good candidate for the main stage.
- Learn about experience with initiating, planning and executing a large agile program
- How we organized distributed teams
- Learn about probably the largest continuous integration system in the world
- How to lift continuous integration to an enterprise-scale
- Research in testing on large test and code data sets
- Advantages/Possibilities of a distributed, service-based build&test system