For dynamic baselining, we want the following high-level behavior:

• Perf test is executed for a given workload
• If doing a performance baseline, dynamically fetch the current baseline for that workload from Elasticsearch, using some tooling (potentially custom Python script)
  • If the test passes (based on tolerancy config) then update the baseline for that workload with the new UUID, using some tooling (potentially custom Python script)
  • Otherwise, fail and do not update the baseline
• If not running a baseline, skip entirely and jut output statistics as normal