Identity in Databricks is built on authentication. Every user must be known, verified, and tied to a profile. This identity links directly to permissions. Databricks supports identity federation through providers like Azure Active Directory, AWS IAM, and SCIM provisioning. Centralizing identity prevents duplicate accounts and shadow access.
Access control in Databricks defines the scope of power each identity holds. Workspaces contain notebooks, jobs, clusters, and data objects. Permissions work on a hierarchy: you can grant or limit access to compute resources, table data, or project artifacts. Role-Based Access Control (RBAC) and Attribute-Based Access Control (ABAC) give fine-grained management. RBAC maps groups and roles to predefined permissions. ABAC uses attributes like department, project tag, or environment to allow or deny actions dynamically.
Securing identities means scoping permissions tightly. Remove defaults. Deny until explicitly needed. Audit frequently. In multi-cloud or hybrid setups, sync your identity provider with Databricks so deactivated accounts lose access instantly.