iter8

Iter8

Metrics Driven Experiments

Open-source platform for metrics-driven experiments and rollouts.

Built for DevOps/SRE/MLOps/data science teams.

Use Cases

1. Load testing with SLOs
2. A/B(/n) testing with business reward metrics
3. SLOs with metrics from any backend
4. Traffic mirroring
5. User segmentation
6. Session affinity
7. Gradual rollout

The traffic engineering use-cases (4 - 7 above) are achieved by using Iter8 along with a Kubernetes service mesh or ingress.

Quick Start

1. Install Iter8

Install Iter8 using Go 1.16+ as follows.

go install github.com/iter8-tools/iter8@latest
# you can now run iter8 (from your gopath bin/ directory)

2. Download experiment

Download the load-test experiment folder from Iter8 hub as follows.

iter8 hub -e load-test

3. Run experiment

Iter8 experiments are specified using the experiment.yaml file. The iter8 run command reads this file, runs the specified experiment, and writes the results of the experiment into the result.yaml file.

Run the experiment you downloaded above as follows.

cd load-test
iter8 run

4. Assert outcomes

Assert that the experiment completed without any failures and SLOs are satisfied.

iter8 assert -c completed -c nofailure -c slos

5. Generate report

Generate a report of the experiment in HTML or text formats as follows.

HTML Report

iter8 report -o html > report.html
# open report.html with a browser. In MacOS, you can use the command:
# open report.html

The HTML report looks as follows.

Text Report

iter8 report -o text

Congratulations! 🎉 You completed your first Iter8 experiment.

Documentation

Contributing