Why is the query speed of SQL statement faster on ECS than on RDS
When you run a simple SQL query directly on an ECS instance hosting a self-managed database, you bypass the network round-trip between your application and ApsaraDB RDS. That round-trip adds overhead that a local query never incurs. For a single simple query, this overhead is measurable and can make ECS appear faster.
This result is expected. It does not reflect how RDS performs under real workloads.
Why the benchmark is misleading
A single simple query is the worst case for any managed database. Nearly all of the elapsed time is network latency, not query execution.
As workload complexity grows, this reverses. Under high concurrency and sustained load, ApsaraDB RDS has clear performance advantages over a self-managed database on ECS.
What to measure instead
To get an accurate picture of RDS performance, run tests with:
Concurrency levels that match your production workload
Mixed query types (reads, writes, transactions)
Sustained load over time, not a single cold query
For detailed benchmark methodology and results, see the performance white papers for each RDS engine.