Deploy a Multicluster Ingress on Google Kubernetes Engine

In the previous part of this tutorial for Federated Kubernetes, we have configured a geographically distributed Kubernetes cluster through federation. Now, we will deploy an application in the federated cluster, and expose it via a global ingress endpoint.

While the federated control plane orchestrates the deployments across clusters, we need to configure an ingress that can route the traffic to the nearest cluster based on user’s location. In case one of the clusters suffers from an outage, the ingress must be intelligent enough to route the request to another cluster that’s healthy.

On the Google Kubernetes Engine (GKE), the global ingress is configured as an HTTP load balancer that talks to the backend NodePort on each cluster.

The Google Cloud Platform (GCP) load balancing provides global load balancing for HTTP(S) requests. Customers can configure URL rules that route some requests to one set of instances and route other to other instances. Requests are always routed to the instance group that is closest to the user, provided that group has enough capacity and is appropriate for the request. If the closest group does not have enough capacity, the request is sent to the closest group that does have the capacity.

Google took advantage of the features exclusively available in GCP load balancer. Even though it is not natively integrated with gcloud SDK — the official command line interface for GCP — Google’s engineers published kubemci, the tool that can do global, multi-cluster load balancing.

Read the entire article at Forbes

Janakiram MSV is an analyst, advisor, and architect. Follow him on Twitter,  Facebook and LinkedIn.