Session abstract

The Operator SDK supports 3 ways of implementing your app's operational knowledge: using Go, Helm, or Ansible. A compiled language like Go is usually faster than an interpreted language like Ansible and Python. But how much does it matter when implementing an Operator? How much speed is sacrificed for faster iterations and easier maintenance? We tried to answer these questions by benchmarking Operators written with the three Operator SDK types to manage the life cycle of a large number of application custom resources on a cluster. We then looked at ways to optimize execution speed of Ansible-enabled Operators to close the speed gaps with the other types of Operators. In this session, we’ll present our findings, lessons learned, and ways to optimize execution speed with development efficiency. You’ll see how to evaluate these tradeoffs for their own clusters.