Apache Kafka cluster which will be:

• Running on OpenShift
• Secure
• Accessible from outside of OpenShift

cannot be easily deployed only using template. A controller is required to take care of creating, managing and deleting the clusters. The controller should work on the "operator" concept used for managing Kubenretes resources. The clusters should be created and managed through Config Maps or Custom Resource Definitions. It works the following way:

• When ConfigMap with Cluster configuration is created, the controller / oeprator should deploy the cluster (both Kafka and Zookeeper)
• When the config map is modified, the runnign cluster should be updated
• When the config map is deleted, the cluster should be deleted as well

The cluster should be able to work both accross namespaces as well as within a single namespace (to be able to handle possible security constraints).