Compare two or more versions of your app/ML model and identify the winning version based on business reward metrics using rigorous statistical analysis. Iter8's A/B(/n) experiments help you grow your business with every release of the app.
Ensure that new app versions satisfy latency, error and other service-level objectives before they are released.
Load testing/built-in metrics¶
Use Iter8's built-in mechanisms for generating load and collecting latency and error related metrics for APIs.
Chaos injection reveals weaknesses in your apps by pushing your app/cluster to the brink. Iter8's chaos testing experiments help you determine if your apps continue to satisfy SLOs amid chaos.
Advanced traffic engineering¶
Use advanced traffic engineering features from your service mesh/ingress such as mirroring, traffic segmentation based on headers or cookies, and session affinity as part of Iter8 experiments.
Progressive traffic shifting¶
Use Iter8's multi-armed bandit based AI to progressively shift traffic towards the winning version of your app.
Combine any of the above testing and rollout patterns within an experiment.
Embed Iter8 experiments within any CI/CD/GitOps process. Iter8 can automatically create pull requests, trigger GitHub actions workflows, and emit HTTP notifications at various stages of the experiment.
Iter8 works with diverse app/ML/serverless frameworks. Experiment with Kubernetes deployments, Helm releases, Knative (serverless) services, KFServing (ML) inference services, Seldon (ML) deployments, and more.
Metrics from any database¶
Use metrics from any database such as Prometheus, Elastic, New Relic, and Sysdig as part of Iter8 experiments.