OBIEE to Power BI on Microsoft Fabric: Moving Oracle BI and Analytics Server Reporting
The RPD is rebuilt as a semantic model, not converted.
OBIEE to Power BI is not a dashboard redraw. The logic your reporting relies on sits in the RPD, and Power BI on Microsoft Fabric expects it as a governed semantic model, so the work is to rebuild that model, move the data and recreate reports on top.
This guide compares your options, maps each RPD layer to its Fabric equivalent, shows how Oracle data reaches OneLake and covers security and validation. Check Oracle's current documents for your release before you set a timeline, because the support position is the first thing to confirm.
OBIEE to Power BI: Confirm Your Oracle Support Position First
Support dates are set per release, and sources disagree. Oracle's public Lifetime Support Policy for middleware says support for Release 12c of Oracle BI follows the Fusion Middleware 12c dates, which that document lists as Premier Support to December 2026 and Extended Support to December 2027. Third-party advisers report that OBIEE 12c Extended Support closed earlier, in 2025, and point to the OBIEE note in My Oracle Support, Doc ID 1664916.1, as the governing source.
We cannot resolve that for your release, and a wrong date changes your plan. Read the Oracle documents for your exact version, and for Oracle Analytics Server use Doc ID 2775561.1. Then work back from the date you confirm. Our legacy BI vs Microsoft Fabric TCO comparison helps price staying put against moving.
Options: Oracle Analytics Server, Oracle Analytics Cloud or Power BI
You have four realistic routes. They differ in how much of your content survives untouched and how much you rebuild.
| Option | What changes | Best when | Watch for |
|---|---|---|---|
| Stay on OBIEE | Nothing changes in tooling | A short bridge while you plan | Support position of your release; patches and certification |
| Upgrade to Oracle Analytics Server | Same RPD-based tooling; Oracle documents an upgrade from OBIEE 12.2.1.4 using a new Oracle home and the Upgrade Assistant | You want to stay on Oracle with minimal content change | Licensing, Fusion Middleware stack and desupported features |
| Oracle Analytics Cloud | Oracle-managed cloud service | You want to stay on Oracle and move to its cloud | Feature differences from your current estate; Oracle publishes a comparison |
| Power BI on Microsoft Fabric | Re-platform: new semantic models and reports | You want one analytics platform with the rest of your data | A rebuild, not a conversion; plan triage and parity testing |
Oracle Analytics Server offers the same features as OBIEE 12.2.1.4 with a few exceptions, and Oracle documents the desupported ones, so check that list against your estate. Moving to Power BI instead is a choice about platform direction. If your finance and operational data is already heading into Fabric, a single platform removes a second set of skills, licences and servers. Our Microsoft Fabric business case sets out how to frame that decision.
RPD to Semantic Model: How Each Layer Maps
The RPD has three layers. The physical layer describes sources and joins, the business model and mapping layer defines logical tables, hierarchies, measures and aggregation rules, and the presentation layer exposes subject areas. In OBIEE to Power BI migration the business model layer is the one you rebuild most carefully, because that is where the numbers are defined.
The RPD layers and their Fabric equivalents.
| In OBIEE | In Power BI on Fabric |
|---|---|
| Physical layer: tables and connection pools | Source tables as Delta tables in OneLake, plus Fabric connections and an on-premises data gateway |
| Business model and mapping: logical tables and joins | Star-schema semantic model with relationships |
| Business model: hierarchies and time dimension | Power BI hierarchies and a marked date table |
| Business model: measures and aggregation rules | DAX measures; level-based rules become CALCULATE patterns |
| Presentation layer: subject areas | Display folders and hidden fields in the model |
| Variables and initialization blocks | Measures, parameters or a user mapping table |
| Application roles and data filters | Row-level security roles |
| Answers, dashboards and prompts | Power BI reports, report pages and slicers |
| BI Publisher reports | Power BI paginated reports |
| Agents and scheduled delivery | Subscriptions and alerts |
Rebuild on a star schema rather than copying the logical model table by table. Several subject areas often carry overlapping definitions of revenue or headcount. Define each metric once in a shared model and let reports reuse it. Choose Import, DirectQuery or Direct Lake per subject area based on data size and freshness. Our BI tool consolidation guide covers the model-first approach.
Rebuilding Calculations When You Move OBIEE to Power BI
Logical SQL functions do not translate one for one. Time-series functions such as AGO and TODATE become time intelligence over a complete, marked date table. A FILTER expression with USING becomes a CALCULATE with a column filter. Level-based measures become CALCULATE patterns that keep only the filters the original respected. Parent-child hierarchies are flattened with PATH and PATHITEM.
The key difference is that a DAX measure responds to the filter context of the report. Test each translated measure against the Oracle result at the same grain, with the same filters, before any report is built on it.
Answers, Dashboards and BI Publisher to Power BI Reports
Rebuild reports from the business question they answer, not from the layout. Microsoft's own migration guidance says conversion cannot be completely automated and advises focusing on the questions a report answers. Triage first: reports with no recent usage are retired, near-duplicates are merged, and the rest are rebuilt in waves.
Answers analyses and dashboards become Power BI reports, prompts become slicers, and drill paths become drill-through and bookmarks. Pixel-perfect BI Publisher output is rebuilt as paginated reports. Agents and scheduled delivery are replaced by subscriptions and alerts, planned per report.
- Confirm your Oracle release and its support dates from Oracle before planning.
- The RPD business model is rebuilt as a star-schema semantic model.
- Logical SQL becomes DAX, tested against Oracle results at the same grain.
- Oracle data reaches OneLake by mirroring, pipelines or copy jobs.
- Application roles become row-level security, tested user by user.
Oracle Data into OneLake
Your reports need the Oracle sources the RPD reads. Fabric can mirror an Oracle database into OneLake, which gives a read-only copy that updates in near real time, plus a SQL analytics endpoint. Microsoft lists Oracle versions 10 and above with LogMiner enabled, on-premises, OCI, Database@Azure and Exadata, through an on-premises data gateway.
Check the requirements early. The database must be in archive log mode with supplemental logging, and in write mode. Mirrored tables need a primary key or a unique index, a mirrored database supports up to 1,000 tables, and table names of 30 characters or more are not supported. Microsoft states that replication compute and the OneLake storage used by mirroring are not charged, and you pay for the compute used when you query the data.
When mirroring does not fit, use a pipeline, copy job or Dataflows Gen2 with the Oracle connector. Through a gateway the pipeline connector needs the Oracle Client for Microsoft Tools on the gateway machine. Copy job change data capture for Oracle is in preview, so check its status first. Our Oracle EBS and Fusion to Microsoft Fabric guide covers the ERP side of the same data. If you run Oracle BI Applications, treat its prebuilt warehouse and ETL as a separate decision.
Mapping OBIEE Security to Power BI
OBIEE controls access with application roles, object permissions and data filters, and often session variables populated by initialization blocks. In Power BI, map them as follows.
- Application roles become workspace and app access groups, aligned with Microsoft Entra ID groups.
- Data filters become row-level security roles with a DAX filter on the semantic model.
- Session variables that carried a user's region or entity become a user mapping table, matched to the signed-in user.
- Object-level restrictions become object-level security or separate models, decided per case.
Row-level security applies to the Viewer role, not to Admin, Member or Contributor, so test with a viewer account. Paginated reports use the built-in user field for their own filtering. Test every role user by user against what that person saw in OBIEE, and note where Direct Lake models need a fixed identity connection.
Validate Reporting Before You Switch
Run both estates in parallel and prove parity report by report. Compare row counts, key measures within an agreed tolerance and filtered views, and have each report owner sign off. Retire the Oracle report only after sign-off. Bad source data shows up faster in a new platform, so our guide to the cost of bad data across Oracle, SQL Server and Fabric and the piece on data quality across platforms explain what to monitor.
OBIEE to Power BI FAQ
Is OBIEE end of life?
Support dates are set by Oracle per release, so check yours. Oracle's public Lifetime Support Policy and the OBIEE support note in My Oracle Support (Doc ID 1664916.1) are the sources, and third-party advisers report different dates for the 12c line, which is why you should confirm before planning around a deadline.
Can I convert an RPD to Power BI automatically?
Not completely. Microsoft's migration guidance says a conversion to Power BI cannot be fully automated end to end. Tools can produce a starting draft, but the RPD's business model is best rebuilt as a star-schema semantic model and tested against the Oracle results.
Should I move to Oracle Analytics Server or to Power BI?
Oracle Analytics Server is the lower-change path for the same tooling, and Oracle documents an upgrade route from OBIEE 12.2.1.4. Power BI on Fabric is a re-platform and a rebuild, which makes sense when you want one analytics platform with the rest of your data and are ready to retire Oracle content you no longer use.
What replaces Answers, dashboards and BI Publisher?
Answers analyses and dashboards become Power BI reports, prompts become slicers, and BI Publisher output becomes Power BI paginated reports. Templates do not convert automatically, so each needed report is rebuilt.
How do I get Oracle data into Fabric?
Where the source database meets Microsoft's requirements, mirroring can replicate Oracle tables into OneLake. Otherwise use pipelines, copy jobs or Dataflows Gen2 with the Oracle connector, through an on-premises data gateway where the database is not publicly reachable.
How are application roles and data filters handled?
Application roles and data filters become row-level security roles on the semantic model. Session variables that held a user's region or entity are usually replaced by a user mapping table and the signed-in user, and each role is tested user by user.
For the full service, see OBIEE to Fabric migration, or request a free Oracle BI assessment. You can also read about our Power BI consulting.