TiDB Cluster Deployment: Machine Deployment Requirements

This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: tidb集群部署,机器部署要求

| username: HACK

When deploying a TiDB cluster, to minimize the impact in case of server node failure, what are the physical machine deployment requirements? For example, the deployment location of servers in the cabinet, rack, and similar requirements.

| username: TiDBer_jYQINSnf | Original post link

This depends on your hardware environment.
Divide the fault domains into three. Place one replica in each fault domain.
If you have three data centers, then the fault domains are at the data center level, ensuring that one data center failure does not affect the system.
If you have only one data center but at least three racks, then set the fault domains at the rack level, ensuring that one rack failure does not affect the system.
If you have only one rack with three hosts, then you can only achieve host-level fault tolerance.
If you have only one host, then if it fails, everything fails. There’s nothing you can do about it.

| username: tidb菜鸟一只 | Original post link

The best option is definitely a two-site, three-center setup.