The operator itself simply watches for instances of the Kong CRD and deploys a Kong instance for each. You’ll typically only have a single instance of the operator per cluster, and it can manage multiple Kong instances.
The individual Kong custom resources represent configurations of our Helm chart, which supports toggling various components on or off. Basic configurations will deploy all components in the same namespace and configure them to interact with other components created by the Kong CR, but you have the option to, for instance, configure one CR that only deploys the proxy and another that deploys the admin API and controller in a separate namespace, among other configurations:
Note that the operator/chart can’t automatically configure the correct endpoints for other instances. If you’re using split configurations, you’ll need to configure the appropriate service addresses for the other instances manually.