We need to store benchmarks somewhere and be able to analyze the results even if time has passed.
I suggest to create a repo called eve-performance and store results there.
Proposing the following structure:
Performance benchmark structure
Level 1: Name of the configuration (e.g. Raspberry Pi 4)
- README.cfg - full spec of the HW
- SUMMARY.csv - All results in the table
Type | Config name | Config descr | Io_test | Result |
- HOWTO.me - How to run
- Level 2: CONFIGS
CONFIG_NAME : where we run (which env : Container/ eve/ ubuntu)
- README.cfg - some description
- default.yml - eden config (from ~/.eden)
- Qemu.conf - qemu config (from ~/.eden)
- Dev….json -after test is finished (from ~/.eden)
- Level 3: TESTS
IO_TEST: name of the fio test
- test.cfg - full test configuration
-README.cfg - test summary
-fio (or some test) - RAW results
-iostat
-perf
-sar
All results should have time and raw data.