What is actually inside a Discoverer estate?
Discoverer has two halves. The End User Layer is a metadata schema stored in the database: it groups tables and views into business areas, exposes them as folders and items with friendly names, defines the joins between folders, and holds calculated items, conditions, item classes and hierarchies. Workbooks sit on top of it: each workbook contains worksheets that pick items from one or more folders, apply conditions and parameters, and lay the result out as a table or crosstab. When a worksheet runs, Discoverer generates SQL from those choices and sends it to the database.
That shape drives the migration. There is no separate semantic server to replicate, but there is also no single governed model — the real business logic is scattered across custom folders full of hand-written SQL, calculated items defined in individual workbooks, and conditions that each author set slightly differently. Consolidating that into one model is where most of the value of the move comes from.
In most estates we see, the EUL sits on Oracle E-Business Suite, and many folders are built on EBS views rather than base tables. That matters twice: once for the data model, because the views carry joins and filters that have to be understood rather than copied, and again for security, because some EBS views only return the right rows when the session has been set up by an EBS login.
How do Discoverer objects map to Power BI?
This is the working map the plan is built from. Every line gets an owner and a decision before rebuild work begins, so nothing surprising appears halfway through the first business area.
Oracle Discoverer artefacts and their Power BI equivalents
| Oracle Discoverer |
Power BI equivalent |
What changes in practice |
| End User Layer (EUL) |
Power BI semantic model |
Metadata stops living in database tables only an administrator can edit and becomes a documented, refreshable model |
| Business area |
One semantic model, or a set of display folders within it |
The business area becomes the migration unit and usually the workspace boundary |
| Simple folder (one table or view) |
A Power Query table, or a view in a staging database |
Friendly item names survive as column names and descriptions |
| Complex folder (items from several folders) |
Relationships between fact and dimension tables |
The join is declared once in the model instead of being re-resolved for each worksheet |
| Custom folder (hand-written SQL) |
A native query, or better, a named view in the warehouse |
Buried SQL is reviewed, owned and tested rather than carried across unread |
| Folder joins |
Model relationships with explicit cardinality |
Fan-out joins that quietly double-counted in Discoverer are found and fixed |
| Calculated item |
DAX measure, or a Power Query column |
One definition of each calculation instead of a copy per workbook |
| Condition |
Report, page or visual filter |
Mandatory conditions that kept queries small become model or source-query filters |
| Parameter |
Slicer, or a parameter on a paginated report |
Users filter a loaded model rather than re-running a prompted query |
| Item and date hierarchies |
Model hierarchies and a proper date table |
Drill paths are rebuilt; fiscal calendars become explicit columns |
| Summary folders |
Import-mode model, or user-defined aggregations |
Usually unnecessary once data is imported; kept only where volume demands it |
| Workbook |
Power BI report |
Several near-duplicate workbooks are often merged into one report |
| Worksheet (table or crosstab) |
Report page with a table or matrix visual, or a paginated report |
Exploratory sheets go interactive; print-and-export sheets go paginated |
| Page items |
Slicers |
The same filtering habit, applied without re-running the query |
| Discoverer Plus (authoring) |
Power BI Desktop |
Authors build on the shared model instead of assembling their own joins |
| Discoverer Viewer (consumption) |
Power BI Service: workspaces and apps |
Readers get a browser and mobile experience with no Java dependency |
| Discoverer Administrator |
Model ownership in Power BI Desktop plus workspace administration |
Metadata changes go through a versioned model, not direct EUL edits |
| Business Area access for users and responsibilities |
Workspace roles and app audiences |
Redesigned from a documented access matrix; it is not migrated directly |
| EBS responsibility and operating-unit data security |
Row-level security from a user-to-value security table |
Rebuilt and tested per user, because the EBS login context does not exist in Power BI |
| Scheduled workbooks |
Scheduled refresh and subscriptions |
Refresh runs once for the model; delivery is planned per audience |
Want your own End User Layer mapped like this? Tell us which business area your team relies on most.
Book a free demo →
The folder and join rows carry most of the modelling risk, which is why we design against Microsoft's star schema guidance (opens in new tab) rather than recreating one table per folder. The measure work that follows is the data modeling and DAX half of the job. Microsoft's Power BI migration guidance (opens in new tab) follows the same order we do: assess and prioritise the existing content before rebuilding any of it.
How is the SQL behind each workbook extracted?
Every Discoverer worksheet resolves to a SQL statement, and capturing it is the most reliable specification a migration can have. For a modest estate it can be done by hand, worksheet by worksheet, from the authoring tools. For a large one, extraction tools read the EUL and workbook definitions in bulk; the consultancy ennVee, for example, has published a Discoverer to Power BI case study for a US municipal government (opens in new tab) in which its tool extracted the SQL behind 99 workbooks and 112 worksheets from a much larger pool of reports identified during assessment.
That example also shows the point of the inventory: the reports worth rebuilding were a small share of what existed. Extracted SQL tells us what each surviving worksheet asks the database for. It does not tell us whether the logic is right, so each query is read for its grain, its joins, and any hard-coded values before it becomes part of the model. Where a custom folder's SQL is genuinely the clearest expression of the rule, it can run as a native database query (opens in new tab) in Power Query, though we usually prefer a named view the database team owns.
Connectivity needs a check up front. Microsoft's Oracle database connector documentation (opens in new tab) lists its supported Oracle Database versions and the on-premises data gateway it needs for scheduled refresh. Older EBS databases that fall outside that list are better served by landing the required tables in a supported staging database first.
How is E-Business Suite security rebuilt?
This is the part of a Discoverer migration that should never be assumed. In an EBS-mode EUL, users sign in with an EBS username and responsibility, Business Area access is granted to those users and responsibilities, and many EBS views filter rows using the operating unit or ledger context the responsibility sets up at login. Power BI has none of that machinery, and ennVee's own write-up on migrating from Discoverer to Power BI (opens in new tab) notes that Business Area access security cannot be migrated directly.
So we treat security as a design task. First we document what each responsibility actually allows — which operating units, ledgers, inventory organisations or cost centres — and confirm it with the system owner. That becomes a security table of users and the values they may see, and a row-level security (opens in new tab) role filters the model against the signed-in user. Content-level access becomes workspace roles and app audiences. Views that depend on the EBS session context are queried through an account and filter we have verified, because run outside that context they can return no rows or rows nobody should see.
How is this different from an OBIEE migration?
Discoverer and OBIEE are different Oracle products, and the projects differ accordingly. Discoverer's metadata is the EUL inside the database and its logic is spread across folders, custom SQL and workbooks; the work is consolidation. OBIEE centralises its logic in a repository and a BI Server, so the work there is reading that repository — covered on our OBIEE to Power BI migration page. Some Oracle estates run both; when they do, we scope them as related but separate workstreams.
What does the first phase cost?
We scope the first phase as one business area and the workbooks people still use — typically a single department such as payables, procurement or general ledger. That is small enough to scope in hours before work starts, and it is typically sized to a Growth ($4,600) or Power ($8,000) block depending on how much consolidation and security work the area needs. The pilot and the work after it run on prepaid Flex hours that never expire — Starter ($2,500), Growth ($4,600), or Power ($8,000), with hourly work from $100/hour — so the next business area is a scoped task, not a new negotiation.
Cost follows the shape of the EUL more than the count of workbooks. Forty worksheets reading one well-built folder are a smaller job than ten worksheets on custom SQL folders with operating-unit security. You get an estimate in hours per deliverable, approve only what you want built, and unused hours stay banked. See Flex pricing.
How long does it take?
Individual reports typically ship within 3–5 business days of approval once the model is standing, and the migration is a sequence of those tasks rather than one big-bang cutover. The model and the security table come first, because every report after them gets cheaper. We recommend running Discoverer alongside Power BI for one full period close so users can compare both outputs on live data before the business area is switched off.
Who this is for
Finance, operations and IT teams on Oracle E-Business Suite whose day-to-day reporting still runs through Discoverer: an EUL only one administrator understands, Java and browser workarounds to keep Viewer running, an EBS or database upgrade blocked by reporting, or a backlog of Excel exports re-keyed every month. Published examples range widely in size: IT Convergence describes moving an aircraft parts supplier from Discoverer to Power BI (opens in new tab) — eleven Finance and Procurement reports — while the municipal example above began from thousands. It is a poor fit if you want every worksheet copied unchanged with no decisions made; that recreates the sprawl in a new tool.