The Google Kubernetes Engine (GKE) is a cluster manager and orchestration system built by Google on top of Kubernetes (https://kubernetes.io), their open source system for automated deployment, scaling, and management of containerized applications. GKE is an example of a CaaS platform and is consequently a deployment unit for an application. In this case, this should be a Docker image.
With this approach, we initially build a fat jar with an application and then wrap it into a Docker image. For that purpose, we may add Dockerfile and run the docker build command manually or by using Maven or Gradle plugins to do it as part of a standard build pipeline. After testing, we may deploy an image to the Google Cloud Container Registry (https://cloud.google.com/container-registry) or even use the Google Cloud Container Build service (https://cloud.google.com/cloud-build) to perform actual build and container deployment procedures.
Along with containers that hold our production service, we may deploy all the monitoring infrastructure just by referencing public Docker images. For example, we may reference Prometheus, Grafana, and Zipkin in the same cluster configuration file. With Kubernetes based platforms, it is easy to reach system elasticity with embedded auto-scaling mechanisms.
Similarly, we may deploy applications to any CaaS platform including, but not limited to, Amazon Elastic Container Service, Azure Container Service, Pivotal Container Service, or even on-premise solutions such as OpenShift (https://www.openshift.com) or Rancher (https://rancher.com).