Getting Started with Trend

Trend collects metrics of past experiments and summarizes them to show trends that could expose performance problems creeping up over time, which might not be obvious if one only compares consecutive versions during canary testing. This is invaluable to developers when tracking down a performance problem and pin-pointing it to a particular version that was deployed in the past.

This is an optional component to iter8 and cannot run standalone. It should be installed either as part of iter8 installation process or separately after iter8 is installed.

The following short video introduces Trend:


kubectl apply -f


Trend implements a Prometheus scrape target, so summarized metric data can be collected by Prometheus and visualized in Grafana. To enable Prometheus to scrape Trend, you need to add a new scrape target to Prometheus configuration, e.g., in Istio, you do the following:

kubectl -n istio-system edit configmap prometheus

In the list of jobs, copy and paste the following at the bottom of the job list:

    - job_name: 'iter8_trend'
      - targets: ['iter8-trend.iter8:8888']

and then restart the Prometheus pod for the change to take effect:

kubectl --namespace istio-system delete pod $(kubectl --namespace istio-system get pod --selector='app=prometheus' -o jsonpath='{.items[0]}')

Then, we use port-forward to make Grafana available on localhost:3000:

kubectl -n istio-system port-forward $(kubectl -n istio-system get pod -l app=grafana -o jsonpath='{.items[0]}') 3000:3000

Finally, we import Trend dashboard in Grafana.

curl -Ls \
| DASHBOARD_DEFN= /bin/bash -


kubectl delete -f