Cloud infrastructure

Umii comes with a complete cloud infrastructure that deploys container images as individual services in a highly available backend infrastructure.

• All individual components of the cluster can be scaled up and down as you see fit, depending on your availability/cost requirements.

• Applications may be deployed redundantly within and across nodes in the cluster, so that end users do not notice service failures because they will automatically be redirected to an identical copy of the crashed service.

• Container startup times in the Umii infrastructure are roughly twice as fast as in similar solutions.

• Multiple monitoring tools make it easy to understand the state of the cluster at all times.

• The system never has to go offline for any updates to the application code. This is achieved by progressively shifting traffic between the old and new versions of a service.

Technical Details

We prebuilt a complete Kubernetes distribution that uses CRI-O and crun as CRI and runtime, respectively.

Traffic inside the cluster is routed and monitored through Envoy proxies which communicate via TLS-encrypted gRPC channels. These proxies are automatically deployed as sidecar-proxies by a an Istio service mesh deployment.

The monitoring tools that are enabled per default are Kiali and Graphana.