If you're running on AWS: Databricks deploys Spark nodes in an Amazon Virtual Private Cloud (VPC) running in the customerâs own AWS account, giving the customer full control over their data and instances. VPCs enable customers to isolate the network that is used by their Databricks deployment from the rest of the networks they use on AWS, control access to it via Security Groups, and revoke permissions at any time. Data does not leave the customerâs storage infrastructure.
You can read the platform security primer and for more advanced topics watch the Data and AI summit video
