Not long ago, Kubernetes was only for stateless apps. Now it's common to run stateful workloads and databases on Kubernetes. But with Custom Resources, we can store just about anything in Kubernetes itself. Maybe we don't need those databases after all? With a bit of work, we can turn the Kubernetes API into a database and metastore, without any additional infra. All you need is a Kubernetes cluster to get started, and you can install "real" databases over time. For some applications, you may find you never need a real database. What would a SQL interface to Kubernetes look like? In this talk, we explore what happens when you store data and metadata directly in the Kubernetes API, and why you might want to do that!
Ryanne Dolan is a senior staff software engineer at LinkedIn, where he works on data infra at massive scale. He is singularly focused on CDC, ETL, rETL, replication, ingestion, backfill, etc. He has a rare condition that makes everything look like a database. He lives in San Antonio with his six pinscher pups.