3 ms·
Signals of API Health and Performance in Cloud-Native Applications
- emfree 10y agoLooks cool. Instrumenting at the network layer is certainly a promising approach. Are you recording latency distributions, and not just averages? The screenshots only show mean and median latency, which isn't enough to spot many anomalies.
- ackerman80 10y agoRight now we maintain few select percentiles from the latency distribution over 1 min time-period. We plan to maintain latency histograms which will allow you to look at latency distribution on arbitrary time intervals.
- conikeec 10y agoBig fan of the visual devOps paradigm. Also installation is painless and non-invasive