Multi-Cluster Reference Architecture Without a Service Mesh

The MongoDB Enterprise Kubernetes Operator is deprecated. The new MongoDB Controllers for Kubernetes Operator replaces the MongoDB Enterprise Kubernetes Operator. The first version of the Controllers for Kubernetes Operator is functionally equivalent to v1.33 of the Enterprise Kubernetes Operator. For more information about this change, and guidance on migrating to the new Operator, see the release notes for the first new version. There will be no future releases of the MongoDB Enterprise Kubernetes Operator. Each version will reach end of life according to the existing one year support policy. Please migrate to the Controllers for Kubernetes Operator for continued support.

The multi-cluster reference architecture includes multiple deployment strategies, each with the common foundation of MongoDB custom resources being deployed across multiple Kubernetes clusters without a service mesh.

Standard multi-cluster deployments consist of one or more elements from each of the following categories, including but not limited to Kubernetes clusters and a means of securely networking between them.

You can access all of the code examples included in the following guides from the architectures directory of the Kubernetes Operator official GitHub repository, or you can simply copy and paste the same examples directly from the guides included in this reference architecture.

Follow the procedures within the following categories to deploy a tested, working multiple-Kubernetes cluster implementation of the MongoDB Enterprise Kubernetes Operator.

Note

The examples included in this reference architecture are opinionated and include specific, third-party tools and platforms for illustration's sake. You can substitute any equivalent tool or platform. MongoDB does not support third-party tooling.

Kubernetes Infrastruture

To deploy any variety of Multi-Cluster MongoDB implementation, you must have multiple Kubernetes clusters. You can deploy your Kubernetes clusters in any way you like. To deploy them on GKE (Google Kubernetes Engine), you can follow the steps in this guide:

Deploy multiple Kubernetes clusters on GKE.

Cluster-to-Cluster Networking

All implementations of multiple Kubernetes cluster Kubernetes Operator deployments require each cluster to expose FQDN. To achieve this without deploying a service mesh you can follow the guides within this section of the reference architectures.

Configure External DNS.

MongoDB Kubernetes Operator

The MongoDB Enterprise Kubernetes Operator handles the deployment of MongoDB custom resources, such as the Ops Manager and MongoDB database intances in Kubernetes. For general information about the Kubernetes Operator for multi-cluster deployments, see Multi-Cluster Replics Sets. To deploy the Kubernetes Operator to your Kubernetes cluster, see:

Deploy the Operator.

Ops Manager

MongoDB Enterprise Kubernetes Operator deployments require either a Cloud Manager or an Ops Manager instance. These required resources facilitate workloads such as backing up data and monitoring database performance. To deploy Ops Manager across multiple Kubernetes clusters, see:

Deploy multi-cluster Ops Manager.

MongoDB Deployments

MongoDB Enterprise Kubernetes Operator facilitates the deployment of MongoDB and related resources in one or more Kubernetes clusters. To learn more, see Multi Cluster Replicasets and see Multi-Cluster Sharded Clusters.