## Conclusion

In this lab you used **TiDB Self-Managed** to see the architectural properties that let **TiDB Cloud** survive data center or availability zone failures:

- **Multi-AZ placement**: PD, TiKV, and TiDB run across three zones, so losing one zone does not take down the entire cluster.
- **Raft replication (3 replicas)**: TiKV keeps data available when one copy of a Region is unreachable.
- **Stateless TiDB SQL layer**: Clients connect to any healthy TiDB node; they do not wait for a single-node failover to complete before resuming work.

**TiDB Cloud** applies these same principles as a fully managed service. TiDB Cloud operates placement, replication, and recovery for you. When a zone or data center fails, your applications can keep using TiDB Cloud endpoints that route to healthy capacity—without you logging into PD or TiKV nodes. The hands-on steps in this lab make that guarantee concrete.

For more on TiDB Cloud, visit [https://tidbcloud.com](https://tidbcloud.com/).