Uploaded image for project: 'Network Observability'
  1. Network Observability
  2. NETOBSERV-1225

Enablement of perflow RTT latency analysis

    • Icon: Epic Epic
    • Resolution: Done-Errata
    • Icon: Undefined Undefined
    • netobserv-1.5
    • None
    • None
    • rtt-enablement
    • False
    • None
    • False
    • Not Selected
    • To Do
    • OCPSTRAT-861 - Passive Per-Flow Latency (RTT) Observability
    • OCPSTRAT-861Passive Per-Flow Latency (RTT) Observability
    • 100
    • 100% 100%
    • S

      Followup on https://issues.redhat.com/browse/NETOBSERV-954

      This epic will allow RTT enablement for Docs and QE

      Feature gate will be removed (see if needed according to implementation)


      Feedback:

      • In Overview, Donut: it shows "Ms", which would stand for "Megasecond" ... should be a low-case "ms"
      • Overview, Timeseries chart: Y axis should show the unit (ms)
      • Overview, Timeseries chart: it seems there's no tooltip?
      • Topology, Edge labels: it's also showing Ms instead of ms

      On a more general note, I started testing with the default sampling, and this is really impractical as very few RTTs come up. After switching to sampling 1, it started to look good. I guess that's going to be a known issue, at least for some time, but we should probably document that (not only in the downstream doc, but also in the CRD doc). We can create a new task for that in the 1.5 epic.
      Also, maybe also in the 1.5 epic, we could add a task to show things like p99 latencies rather than averages, which I think is more relevant to look at for SRE

            jpinsonn@redhat.com Julien Pinsonneau
            jpinsonn@redhat.com Julien Pinsonneau
            Amogh Rameshappa Devapura Amogh Rameshappa Devapura
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated:
              Resolved: