PolarDB:What is the query speed difference between the cross-database query in the same RDS instance and the cross-database query among different RDS instances?

If the volume of queried data is the same in both query scenarios and the hardware resource limits of the RDS machine are not taken into consideration, the latter query scenario is theoretically faster than the former query scenario (this is more typical in aggregate query scenarios). The CPU and I/o resources of apsaradb for RDS can be better utilized because the RDS instance has a higher parallelism.

Ideally, the RT of the latter is one in N of the former (N is the number of RDS instances). If the RDS instances have the same specifications, they are more likely to reach the RDS system resource bottleneck, because data is computed and transmitted over the shards in a single RDS instance.