Creating and maintaining a GKE cluster
Once you have a workload suitable and ready to deploy to GKE, you need a place to deploy it. As with most things GCP, you have a choice whether to perform the cluster creation via APIs or the portal. When repeatability, testability, and automation are a concern, you should always go with scripting your configuration and executing actions via GCP APIs or the CLI. When working in a prototyping or R&D environment, it's totally acceptable to perform actions, such as GKE cluster creation, using the portal since you are likely to be tweaking the settings.
When creating a GKE cluster, the bulk of the configuration is centered around the machine type supporting your cluster, the number of nodes, and what zones or regions the cluster will be deployed to. For instance, this is a three-node cluster deployed to the us-east-1b zone running on n1-standard-1 machine:
gcloud container clusters create my-gke-cluster --zone us-east-1b --machine-type n1-standard-1There...
