Your organization's application deployed on Google Kubernetes Engine (GKE) requires high availability and uses regional persistent disks for storage. The application is stateful and cannot tolerate simultaneous downtime across multiple replicas. You need to ensure that maintenance operations do not cause downtime. Which strategy should you adopt?