Benchmark archive
Interpreting benchmark results
Apply the correct caveats to historical AponiaJS performance measurements.
Historical synthetic results can compare two implementations under one controlled CI protocol. They cannot establish:
- production latency under real application work;
- tail latency or memory behavior;
- database, network, or external service cost;
- capacity across different machines;
- a stable performance ratio across framework versions.
When citing a result, include:
- the AponiaJS version and commit;
- the benchmark protocol and workloads;
- runtime and dependency versions;
- connection count and duration;
- raw treatment values;
- the number of observations;
- the limitations above.
Prefer application-specific load testing over archived framework microbenchmarks for deployment decisions.