1. Connect the account
Data Security reuses the cloud accounts already connected under Cloud Security. If the account is connected there, it appears under Data Security → Scan Configuration with no further setup. To connect a new one, follow the provider guide:- AWS — role assumption with an external ID that TigerGate mints. A client-supplied external ID is rejected.
- GCP — service account or workload identity.
- Azure — app registration with a directory role.
Data Security scans AWS, GCP and Azure. Cloud Security connects more providers
than this — an account of another kind stays visible in Scan Configuration but
is marked as one Data Security does not scan, rather than reported as an
account whose scan failed.
2. Choose what to scan
Open the account under Scan Configuration. Inventory needs nothing further. To classify content, set Scan depth to Read object content in the scope below and choose a scope. Both are required — a grant alone does not start a content read, and neither does a scope without the grant below. See Scan configuration for scope, sampling profiles and frequency.3. Grant object access
Reading object content requires a separate, verified grant per account. The posture role is deliberately not sufficient: inventory and content are different levels of access, and one should not silently imply the other. Until that grant exists, the scheduler says so plainly rather than appearing idle:4. Database access
Relational and NoSQL stores need a read-only database user. Set it under Database access in the same panel:- Create a read-only user in your database and grant it
SELECT. - Enter that username.
- Prefer the passwordless method your cloud offers — IAM database authentication on AWS, workload identity on GCP, Microsoft Entra on Azure.
What TigerGate stores
| Kind | Stored |
|---|---|
| Cloud credentials | Encrypted, per account. Role assumption is preferred so there is no long-lived key. |
| Database passwords | Never. Only a reference to a secret in your account. |
| Object contents | Never. Classification produces classifier IDs, counts, confidences and masked exemplars. |
| Row values | Never. Same as above. |