Purpose of this section is to explicitly write down how we’re going to measure, report, and goal on our metrics so that we can have a single source of truth, and place to drive alignment. The focus here is really on the top-level project metrics, but many of the things should probably apply to more detailed metrics just so we have consistency.
Topic | How we do this | Are we aligned? |
---|---|---|
Where we measure our metrics for official dashboards | We will use prod setup, not staging environment | ✅ |
Naming of old system vs. new system | In all graphs, new system should be called “Rhea” and old system “Old” | ✅ |
Official Correctness Metric | Percentage point (pp) delta produced by subtracting success % metrics. | |
We always subtract Rhea from Old, negative delta means Rhea is more successful than Old. | ✅ | |
Official Latency metric(s) | TTFB, TTLB @ P95, P90, P50 | ✅ |
Apdex / Misery metric(s) | TBD. First candidate would be some Apdex(Satisfied/Tolerated/Frustrated) for loading root HTMLs of DNSLink websites. | ? |