You can deploy MongoDB Search and Vector Search with the Kubernetes Operator against a range of MongoDB deployment topologies. The topology that you choose determines how the Kubernetes Operator deploys and connects the mongot process to your MongoDB deployment.
Use this page to confirm that your planned topology is supported before you deploy. To confirm the MongoDB Server, mongot, and Kubernetes Operator versions that you need, see Search and Vector Search Compatibility and Requirements.
Deployment Dimensions
A supported topology is a combination of the following dimensions:
MongoDB Cluster topology. Whether the
mongotprocess serves a MongoDB replica set or a sharded cluster. For sharded clusters, the Kubernetes Operator deploys a separatemongotgroup for each shard.Deployment type. Whether your MongoDB deployment runs in a single Kubernetes cluster (single-cluster) or spans multiple Kubernetes clusters (multi-cluster). In a multi-cluster topology, a single Kubernetes Operator in a central cluster manages the
mongotprocesses in all member clusters.Load balancer mode. When you run multiple
mongotinstances, you must place an L7 load balancer betweenmongodand themongotpods. You can use one of the following options:Managed load balancer. The Kubernetes Operator deploys and configures an Envoy proxy.
Unmanaged load balancer. You provide and configure your own L7 load balancer.
A single
mongotinstance connects directly tomongodand does not require a load balancer.MongoDB source mode. Whether the MongoDBSearch resource uses
spec.source.mongodbResourceRefto reference a MongoDB resource that the Kubernetes Operator manages (managed) or usesspec.source.externalto configure the connection (external).
Supported Topology Matrix
The following table lists the support status for each combination of deployment dimensions:
Important
External MongoDB sources and multi-cluster topologies require MongoDB Enterprise. MongoDB Community supports only replica set deployments that the Kubernetes Operator manages inside Kubernetes.
Cluster Topology | Deployment Type | Load Balancer Mode | MongoDB Source Mode | Support Status |
|---|---|---|---|---|
Replica set | Single-cluster | Managed | Managed | Supported |
Replica set | Single-cluster | Managed | External | Supported |
Replica set | Single-cluster | Unmanaged | Managed | Supported |
Replica set | Single-cluster | Unmanaged | External | Supported |
Replica set | Multi-cluster | Managed | Managed | Not supported |
Replica set | Multi-cluster | Managed | External | Supported |
Replica set | Multi-cluster | Unmanaged | Managed | Not supported |
Replica set | Multi-cluster | Unmanaged | External | Not supported |
Sharded cluster | Single-cluster | Managed | Managed | Supported |
Sharded cluster | Single-cluster | Managed | External | Supported |
Sharded cluster | Single-cluster | Unmanaged | Managed | Supported |
Sharded cluster | Single-cluster | Unmanaged | External | Supported |
Sharded cluster | Multi-cluster | Managed | Managed | Not supported |
Sharded cluster | Multi-cluster | Managed | External | Supported |
Sharded cluster | Multi-cluster | Unmanaged | Managed | Not supported |
Sharded cluster | Multi-cluster | Unmanaged | External | Not supported |
Topology Considerations
Review the following considerations when you choose a topology:
External MongoDB sources and multi-cluster topologies are available only for MongoDB Enterprise. MongoDB Community supports replica set deployments that the Kubernetes Operator manages inside Kubernetes.
In a multi-cluster topology, configure the MongoDB source using
spec.source.external, even if the Kubernetes Operator manages the MongoDB deployment. Multi-cluster topologies don't supportspec.source.mongodbResourceRef.Multi-cluster topologies require the managed load balancer. The Kubernetes Operator supports unmanaged load balancers only in single-cluster topologies.
Multi-cluster topologies require network connectivity between the central cluster and the Kubernetes API server of each member cluster.
For other deployment limitations, see mongot Deployment Limitations.