Subnet sizing for a new workspace
This page includes subnet sizing information for admins deploying new workspaces.
Configuring the size of your workspace's subnet is optional. If not explicitly set, Databricks will set the subnet's IP range automatically when deploying a new workspace. If you would like to set the subnet size, you can do so in the advanced network settings during workspace deployment. See Create a workspace using the account console.
If you deploy your workspace in a customer-managed VPC using the Account API, you must set the IP range yourself.
You cannot change the subnet IP range after your workspace is deployed. It's critical to choose a subnet size that provides enough IP addresses for your expected workload. If the subnet runs out of available IP addresses, Databricks compute might fail to start or jobs might fail to run.
Subnet requirements
Each Databricks compute node requires two IP addresses from the subnet. The total number of concurrent compute nodes across your workspace determines the number of IPs your workspace requires.
Subnet sizes are expressed in CIDR notation. For CIDR range requirements for network resources and attributes, see Network requirements.
Quick sizing guideline
Use the following table to estimate the maximum number of Databricks compute nodes your workspace can support based on the subnet size. The maximum nodes per workspace is the total number of concurrent nodes (compute instances) that can run in a workspace at any one point in time.
Nodes subnet size | Maximum Databricks nodes per workspace |
|---|---|
| 60 |
| 120 |
| 250 |
| 500 |
| 1000 |
| 2000 |
| 4000 |