PolarDB:Performance comparison: PolarDB for MySQL vs. ApsaraDB RDS for MySQL

To ensure an accurate comparison:

• Use identical instance specifications for both products.

• Use the same database versions.

• Use a simulated production workload or Sysbench to reflect real-world conditions.

• When comparing read performance, do not use single SQL queries.

  PolarDB for MySQL uses a compute-storage separation architecture, so a single query may show higher network latency than ApsaraDB RDS for MySQL. In production, cache hit rates typically exceed 99%—only the first read triggers disk I/O. Subsequent reads come from the buffer pool, delivering equivalent performance.

• When comparing write performance, use a stress test rather than single SQL queries.

  For a fair comparison, test a PolarDB for MySQL cluster (primary node + read-only node) against an ApsaraDB RDS for MySQL instance (primary + read-only with semi-synchronous replication). PolarDB for MySQL uses Quorum-based writes—a write succeeds only after data reaches a majority of replicas (at least two of three), ensuring data redundancy and strong consistency at the storage layer. Comparing against semi-synchronous replication rather than asynchronous provides a more meaningful benchmark.