AWS WorkSpaces can be configured with subnets that can be shared within an AWS account or across AWS accounts using resource sharing mechanisms, but this depends on the specific AWS service and context.
For Databricks workspaces on AWS, documentation states that you can choose to share one subnet across multiple workspaces or both subnets across workspaces, and you can also reuse existing security groups rather than creating new ones. However, in practice with the Databricks Account Console UI for network configuration, it says you cannot reuse subnets or security groups between workspaces. The key detail is that while you may share VPC subnets and security groups in theory for efficiency and cost reasons, Databricks recommends using unique subnets and security groups for each workspace to avoid IP exhaustion and scaling issues. Also, you cannot reuse the network configuration object itself across workspaces, meaning each workspace needs a distinct network configuration even if underlying subnets or security groups are shared or reused.
Summary for sharing subnets in AWS Databricks workspaces:
-
You can share subnets in the underlying VPC between multiple Databricks workspaces.
-
For best practice, it is recommended to use unique subnets and security groups per workspace to avoid conflicts.
-
Network configuration objects in the Databricks Account Console are workspace-specific and cannot be reused across workspaces.
-
If you share subnets across workspaces, your VPC/subnet sizing must be large enough to support the combined workspace usage.
-
Reusing security groups is allowed but also recommended to be unique per workspace for clarity and security.
This clarifies the apparent documentation vs UI message conflict: reusing subnets and security groups is possible, but Databricks enforces unique network configuration objects per workspace and recommends unique subnet/security groups for scalability and operational safety.
Sources: Databricks AWS customer-managed VPC documentation, Databricks network configuration docs, and related AWS resources.โ