oracle-coherence-operator Certified

The Coherence Operator is a Kubernetes Operator that is used to manage Oracle Coherence clusters in Kubernetes. The Coherence Operator takes on the tasks of that human DevOps resource might carry out when managing Coherence clusters, such as configuration, installation, safe scaling, management and metrics.

The Coherence Operator is a Go based application built using the Operator SDK. It is distributed as a Docker image and Helm chart for easy installation and configuration.

Using the Coherence Operator to manage Coherence clusters running in Kubernetes has many advantages over just deploying and running clusters with the resources provided by Kubernetes. Coherence can be treated as just another library that your application depends on and uses and hence, a Coherence application can run in Kubernetes without requiring the Operator, but in this case there are a number of things that the DevOps team for an application would need to build or do manually.

JVMs that run as Coherence cluster members need to discover the other members of the cluster. This is discussed in the Coherence Well Known Addressing section of the documentation. When using the Operator the well known addressing configuration for clusters is managed automatically to allow a Coherence deployment to create its own cluster or to join with other deployments to form larger clusters.

The Operator configures the Coherence site and rack properties for cluster members based on Kubernetes Node topology labels. This allows Coherence to better distribute data across sites when a Kubernetes cluster spans availability domains.

When scaling down a Coherence cluster, care must be taken to ensure that there will be no data loss. This typically means scaling down by a single Pod at a time and waiting for the cluster to become "safe" before scaling down the next Pod. The Operator has built in functionality to do this, so scaling a Coherence cluster is as simple as scaling any other Kubernetes Deployment or StatefulSet.

Alongside safe scaling, because the Coherence CRD supports the Kubernetes scale sub-resource it is possible to configure the Kubernetes Horizontal Pod Autoscaler to scale Coherence clusters based on metrics.