TpServer Optimization Principles

For mixed workloads involving TpServer, the SynxDB Cloud system employs the “Computation Offloading” principle for optimization:

• Read and Join Scenarios: When queries require joining data from both Warehouse and TpServer, the optimizer prioritizes moving data from TpServer to stateless Warehouse Segments for computation. This ensures that analytical workloads do not impact TpServer’s high-concurrency transactional processing capabilities by consuming CPU resources.

• Write Scenarios: When executing operations such as INSERT INTO rot_table SELECT ... FROM warehouse_table, complex computational tasks (e.g., Select/Project/Join) are completed on the Warehouse side, with only the final result data sent to TpServer for persistence, thereby minimizing TpServer’s write overhead.