A cluster is the compute that runs your queries. In an In-VPC deployment, cluster nodes are provisioned by Karpenter on the workspace NodePool you created during deployment, and scale to zero when idle.
Create a cluster
- Open the workspace’s Compute Plane and go to the Clusters section.
- Select Create Cluster and enter a Cluster Name (must start with a letter and end with a letter or digit; letters (upper or lower case), numbers, and hyphens, 3–30 characters; case-sensitive).
- Select a Version - the Cluster Runtime (CRT) - and a Cluster Size.
- Configure autoscaling, QPS, and caching as needed, then create the cluster.
The cluster’s pods schedule onto the workspace NodePool (tainted with workspace-name=<workspace>), and Karpenter launches nodes to match. For the full set of cluster options, see Create and manage clusters and Cluster sizing and scaling.
CloudPrem-specific operations
Operating clusters in an In-VPC deployment - inspecting pods, NodePools, and scaling behavior with kubectl - is covered in CloudPrem cluster operations.
Next
Continue to Run your first query. Last modified on September 29, 2026
