A database team needs to deploy a clustered database where each replica requires a stable, predictable network identity (e.g., db-0, db-1) and its own persistent storage that survives Pod rescheduling. Pods must also be created and terminated in a defined order. Which Kubernetes workload resource should they use?