Improve InfluxDB response time

Question

I try to improve the respond-time of Grafana and InfluxDB. Hopefully someone can give me some support.

My setup is: OS: Windows 10 Docker version 20.10.5 Docker-Container: Grafana v7.5.5 Docker-Container: InfluxDB Version 2.0.5

There are different Dashboards, with different numbers of panels. One Granfa-Dashboard for example has 24 Graph-Panels, with refesh-rate of 500ms,1s,2s which should updated the Graph-Panel fast.

I know that 500ms refesh-rate is high, but the same behavior is seen with 1s, 2s, 5s. The queries use this scheme:

    from(bucket: "bucket")
  |> range(start: v.timeRangeStart, stop:v.timeRangeStop)
  |> filter(fn: (r) =>
    r._measurement == "DEVICE1" and
    r.COMMON == "VALUE1"
    )
  |> set(key: "_field", value: "")
  |> set(key: "COMMON", value: "DEVICE 1")      
  |> aggregateWindow(every: v.windowPeriod, fn: last)

What I observe is that some graph updates very slowly.

With the Chrome-Browser I have observed the traffic.

There are some states, like "canceled" or "pending", that seems to slow down the response. A second observation is, that some queries require up to 500 ms to process.

So my main question is, how I could improve response time if a Grafana-Dashboard request a query from InfluxDB.

Update: With these lines before a query, different statistics (TotalDuration, CompileDuration, QueueDuration,...) about the execution will be enabled:

import "profiler"
option profiler.enabledProfilers = ["query", 
"operator"]

Is it possible to plot this statistics?

Thx in advance

Source: https://docs.influxdata.com/flux/v0.x/stdlib/profiler/