Production-to-Dev in One Button: K8s-native PII anonymization for Percona XtraDB Cluster Backups

Stop hand-curating fake staging data! This talk shows a production pipeline built on the Percona XtraDB Cluster Operator that continuously turns the latest PXC backups into sanitized, production-shaped dev/test clones - without exposing PII to development teams.

We’ll walk an end-to-end workflow using Operator resources: restore an PXC backup into an ephemeral cluster, run config-driven row-level anonymization backed by Python & Faker, apply ignore rules for known edge cases, optionally execute pre/post SQL hooks, then produce and publish a fresh anonymized backup that downstream environments will bootstrap from via a single “latest” pointer.

The focus is operational reality: leveraging PXC Operator reconciliation/status, safe cleanup, correctness, performance bottlenecks, and long-term maintainability through versioned, config-driven rules that evolve with your schema and compliance needs.

You’ll leave with a concrete pattern for “prod-to-dev” data delivery that fits GitOps/K8s workflows and scales beyond one-off scripts.