Skip to content

AI-driven Kubernetes experimentation platform

Iter8 makes it easy to optimize business metrics and validate SLOs when you release new versions of Kubernetes apps/ML models.

Quick start in 5 mins

Experiments

Automate A/B, A/B/n, Canary, and Conformance experiments. Compare multiple versions. Find and safely promote the winning version based on business metrics and SLOs.

K8s stacks

Experiment over Knative, KFserving, or Istio. Or extend Iter8 over any K8s stack of your choice by adding just a few RBAC rules.

Metrics

Use metrics from any provider in your experiments, including Prometheus, New Relic, Sysdig and Elastic.

AI-driven

Under the hood, Iter8 uses rigorous Bayesian analysis for version assessments and multi-armed bandit algorithms for traffic shifting.

Progressive rollout

Progressively shift traffic towards the winning version or specify a fixed traffic split during an experiment.

Traffic engineering

Protect end-user experience with features such as mirroring, user segmentation, dark launch, and sticky sessions.

Resource config tools

Use plain YAML/JSON files, Helm, or Kustomize for configuring all the resources needed for an experiment.