Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: tidb Coprocesso读取限制
[Overview]: Long network latency for Coprocessor reads
[Background]: A cluster composed of cloud provider in Beijing, on-premises data center, and overseas nodes. Each location has its own service connecting to the corresponding TiDB.
[Problem]: For example, when making a select request to the on-premises data center, viewing through TiDB dashboard shows that Coprocessor reads are connected to the overseas nodes, causing slow queries. Each table has a small amount of data, and the current bottleneck seems to be the connection to the overseas nodes.
[Version]: v6.1.0
Is there a way to ensure that when the client connects to the on-premises data center, Coprocessor reads are also connected to the on-premises data center, and when the client connects to the overseas nodes, Coprocessor reads are connected to the corresponding overseas nodes?
Tested Placement Rules and as of timestamp in select, but neither solved the problem.