At Salesforce, our team does Kubernetes deployments at the petabyte scale, with 100+ clusters, 10000+ pods, and 25+ PiB of provisioned storage and innumerable Kubernetes jobs. Operating at this massive scale necessitates an approach that reduces engineering toil. Maintaining such mammoth clusters while keeping security in mind is a whole different challenge.

We specifically talk about how and why we picked immutability and Infra-as-code as the cornerstone of our design and the challenges around it. Immutability provides security and scalability and eliminates config drift. Infra-as-code guarantees that changes to our infrastructure follow the same software lifecycle. but the trade-off is a lack of flexibility to adapt to changing requirements.

In this talk, we go over the unique design choices that help us manage, grow, and configure petabyte-scale deployments at Salesforce.