Context
A large international retail group ran a Kubernetes platform on IaaS, administered with Rancher: the virtual machines and the clusters were its own to maintain.
Challenge
Moving to a managed Kubernetes service, more resilient and lighter to maintain, and testing resilience against outages, up to the loss of a cloud zone.
What I did
- I validated the AKS architecture that replaced the IaaS platform administered with Rancher: a managed Kubernetes service whose control plane is operated by Azure.
- I organised several simulated outage exercises, including the loss of a cloud zone, and fault injections on AKS clusters with Azure Chaos Studio.
Outcome
A managed Kubernetes platform, more resilient and lighter to maintain than the former IaaS platform.
Stack
- AKS
- Kubernetes
- Azure Chaos Studio
- chaos engineering