How to Configure Databricks Backup User
You set up a databricks backup user by creating a service identity with read-only table permissions. Next, you bind that identity to your export job and restrict its workspace rights. Many teams take a shortcut and give this account full workspace admin roles. Broad roles let one stolen token read, change, or wipe all workspace assets. How to set up a databricks backup user in your workspace Start by creating a dedicated machine identity instead of a human account. Human accounts break automa
Your coding agents run with your access.
Claude Code reaches prod with a developer's credentials, and approval prompts get clicked through. hoop.dev checks every query, masks sensitive data, and holds risky changes for a human.
You set up a databricks backup user by creating a service identity with read-only table permissions. Next, you bind that identity to your export job and restrict its workspace rights. Many teams take a shortcut and give this account full workspace admin roles. Broad roles let one stolen token read, change, or wipe all workspace assets.
How to set up a databricks backup user in your workspace
Start by creating a dedicated machine identity instead of a human account. Human accounts break automated jobs when someone leaves the company or resets a password. Machine identities stay active and keep human logins out of scheduled tasks.
Next, grant that identity read rights on the specific catalogs and tables you back up. Do not grant modify, drop, or admin rights at the catalog root. The backup account only needs to read table rows and metadata during export runs. You also need to grant permission to start or use a compute cluster.
Finally, set up short-lived credentials for the identity rather than static keys. Federated tokens expire fast and lower the risk of leaked secrets in scripts. Store any required secret values in a managed secret store outside the code.
Where native backup roles fall short
Native catalog roles decide which tables an identity can open when a job starts. Once the session starts, the platform trusts every query that identity sends. This setup solves basic login control, but it leaves a runtime gap.
A backup account must read wide ranges of production tables to copy them. If an engineer or script uses those credentials interactive-style, they can pull raw records. Native roles cannot tell a scheduled dump apart from a manual query.
Static grants also fail to scrub sensitive columns on the fly. Production tables often hold emails, phone numbers, card numbers, and national IDs. When the backup job or a support user runs a query, raw values flow out. Without runtime checks, sensitive fields land in secondary storage without field-level masking.
User Provisioning (SCIM) + End-to-End Encryption: architecture patterns and best practices
Continue reading? Get the full guide.
Free. No spam. Unsubscribe anytime.
Audit logs show that a query ran, not how to block bad statements first. Teams need controls that inspect each statement in flight and mask matched fields in results.
Best practices for ongoing backup account hygiene
Review the table grants on your backup identity at the end of every month. Data teams add new schemas often, and old wildcard grants can expose fresh tables. Remove read rights from archive schemas that no longer need daily snapshots.
Separate the identity that writes backups from the identity that restores data. A restore account needs write permissions, while a daily export account only needs read permissions. Keeping these two roles apart stops a compromised export script from overwriting live tables.
Limit where the backup identity can log in from by setting network rules. Restrict logins to the exact IP ranges of your scheduled job runners. If someone copies a token to a laptop, the network rule blocks the login.
Check database and API queries on the wire
Identity tools, roles, and network rules decide who opens a session. Setup alone cannot check what each query does or mask sensitive values in a response once a session starts. When workflows touch Postgres, MySQL, SQL Server, MongoDB, ClickHouse, Kubernetes APIs, SSH, or internal HTTP APIs, hoop.dev sits inline in the wire protocol beside those services as an open-source proxy, a sidecar for your Postgres or API. On every query an AI agent or engineer runs against production, hoop.dev decides to pass, mask, block, or hold it for human review.
Your login details and client tools stay as they are while hoop.dev guardrails refuse bad deletes or drops before they run, and hoop.dev data masking rewrites matched sensitive values in responses in memory. Furthermore, hoop.dev holds SOC 2 Type II, is MIT licensed, and writes every query and decision to an audit trail that generates evidence for HIPAA, GDPR, and PCI DSS. See the source on GitHub.
Frequently asked questions
Should a databricks backup user be a human account or a service identity?
Always use a non-human service identity for automated backup jobs. Human accounts tie jobs to one person and often bypass strict automation rules.
What minimum permissions does a backup identity need?
The identity needs read access on target catalogs, schemas, and tables. It also needs permission to run queries on a designated SQL warehouse or cluster.
How do we protect sensitive fields during data exports?
Apply dynamic data masking to rewrite emails, phone numbers, card numbers, and national IDs. This step replaces raw values in query results before data leaves the workspace.
Related guides
In this post