- failover
- query routing (e.g., for sharded deployments)
- caching
- workload stats/metrics
- query rewriting
etc.