Data Sources¶
The Datasources dashboard maps where your Power BI content gets its data: connector types, shadow-BI exposure, gateway binding, misconfigured links, dependency hot spots and connection security.
Open it with Select Analysis > Datasources. Every panel shows the current state of datasources, gateways and connections. The date filter does not apply; the environment and capacity filters do, except for orphans, gateways and connection security, which are environment-wide because they have no capacity link.
Image of the Datasources dashboard: Datasource Type Distribution table, Governance Risk Tiers donut, Cloud vs On-Prem Gateway Binding tiles and Misconfigured Bindings
../images/dashboards/datasources.png
Before you begin¶
- Datasources come from the workspace scan. Gateways and connections come from the Fabric API, which needs the tenant setting that lets service principals use Fabric APIs. See Tenant Settings.
- The Fabric API step of Validate the Connection passes.
Panels¶
| Panel | What it shows | Click |
|---|---|---|
| Datasource Type Distribution | Datasources by connector type: Type, Datasources, Dataset usages, Datasets touching | - |
| Governance Risk Tiers | Datasources by tier (below), with "Shadow-BI exposure x% - Governed y% of n scoped datasources." | - |
| Cloud vs On-Prem Gateway Binding | On-Prem Bound (a gateway object is behind the datasource), Cloud / Unbound (no resolvable gateway object), and the datasets' On-Prem Required flag (required, not required, unknown) | - |
| Misconfigured Bindings | Bindings Power BI flags as misconfigured at three levels: Datasource, Dataset usage, Artifact usage. Zero is the healthy result. | Each level opens its list |
| Datasource Fan-In (Blast Radius) | Datasources with the most dependent datasets and artifacts: Datasource, Type, Dependents, Risk (critical SPOF, high fan-in, normal) | A datasource opens Datasource Dependents |
| Orphaned Datasources | Dead Connectivity: share of datasources with no dataset or artifact usage, by type | Orphaned Datasources |
| Backend Server Concentration | Which servers and databases the estate touches: Server, Datasources, Databases, Case variants. Verbatim case shows server names as stored; Fold case merges case variants. | A server opens Server Datasources |
| Gateway Inventory & Health | Gateways: Gateway, Type (OnPremises, VirtualNetwork, Personal (unmanaged)), Version (flagged outdated when behind the latest member), Members, Cloud refresh | - |
| Connection Security Posture | Gateway connections with a flag: Skip test, Not encrypted, Public privacy, No SSO, and a breakdown by dimension | A flag opens Flagged Connections |
Governance risk tiers¶
| Tier | Meaning |
|---|---|
| Governed | Central, IT-managed backends |
| ShadowRisk | Ad-hoc, local or user-owned sources (shadow BI) |
| OpaqueCustom | Custom connectors that cannot be assessed from metadata |
| Other | Unclassified |
Read the panel notes
- "No gateway inventory captured for this environment; on-prem binding cannot be resolved. This is not evidence of a 100% cloud estate."
- "Connection metadata is sparsely captured - treat this posture as a lower bound."
- "Server names are shown verbatim (no case folding) - case fragmentation of the same backend is itself a governance finding."
- File, Web and SharePoint sources carry no server name, so Backend Server Concentration counts only datasources with a server.
Detail lists¶
| Drawer | Contents |
|---|---|
| Misconfigured Datasources / Misconfigured Dataset Links / Misconfigured Artifact Links | Flagged bindings at each level |
| Orphaned Datasources | Datasources with no dataset or artifact usage (environment-wide). Dataflow-only datasources count as orphans here. |
| Datasource Dependents | Every dataset and artifact that uses one datasource: its full blast radius |
| Server Datasources | The datasources on one backend server |
| Flagged Connections | Gateway connections matching one security flag |
Questions this page answers¶
- Which databases and services does Power BI depend on, and which ones are single points of failure?
- How much content runs on unmanaged, user-owned sources?
- Which datasources need a gateway, and are the gateways current?
- Which connections skip encryption or SSO, or use public privacy levels?
- Which datasources are unused and can have their credentials retired?