SQL over the Vault
Views are written in standard SQL over the Vault and over each other, and grouped into Mart databases.
The Mart
Your BI team writes the reporting model as SQL views over the Vault. SchemaVortex checks each change before it goes live, keeps a record of every change, and serves the views to Power BI, Excel and any SQL client.
In short
The calculations your reports share are written once, as views in the Mart. Every report reads the same numbers, and every change to them is on record.
Views are written in standard SQL over the Vault and over each other, and grouped into Mart databases.
A change that does not compile is refused, and the live view keeps serving.
Each change is recorded with who made it and when. An earlier version of a view can be brought back.
A heavy view can store its result, and the result is rebuilt when the data behind it changes.
How it works
Mart views read the Vault and other Mart views. A view runs its SQL when it is queried, or it stores its result, which is rebuilt when the data behind it changes.
salescustomersproductsfinance
reporting.revenue_by_monthstored result, rebuilt when its data changesreporting.marginruns when queriedreporting.top_customersruns when queriedPower BIExcelany SQL client
While a stored result is being rebuilt, the view runs its SQL instead of serving the old copy. A Mart view never reads a Sandbox view, so personal logic stays out of the curated layer.
Changes and history
Views can change one at a time, or together in a Mart Plan that a Mart Publisher publishes. Either way, SchemaVortex keeps the history.
A Mart Plan collects new and changed views. If one of them does not compile, the whole publish stops before anything changes.
Each change is kept with who made it and when, and each version can be compared with the one before it.
Bring back an earlier version of one view, or every view as it stood at a chosen moment. The restore arrives as a Mart Plan, so it can be reviewed before it goes live.
Protection and scale
A view that runs when queried applies the Vault's masks for whoever queries it.
A view computed from protected columns stores its result only after a Data Warden approves them. If it later reaches new ones, it waits for approval again.
Another Query endpoint serves the same views over the same data, so a heavy Power BI estate does not slow down other users.
A short live demo: a view changes in a Mart Plan, the plan is checked and published, and the report reads the new view.