Skip to main content
Once a data source is connected, its detail view opens onto more than a connection summary. Two specialist work areas — Data Modelling and Schema Drift — let an analyst turn raw tables into reusable, well-described models and keep watch for unexpected changes to the source’s structure over time.

Data Modelling (Beta)

Data Modelling lets you explore a source’s tables and columns, then build curated “models” on top of them — saved, named definitions of which tables matter, how they relate to each other, and what each column means.

Schema Explorer

The default view of a source’s structure is the Schema Explorer: a list of tables on the left and a column detail pane on the right. It shows row counts and column counts per table and lets you browse columns and sample data. This is read-only inspection.

Exploring and navigating models

Control / ButtonWhat it does
Table list (Schema Explorer)Browse every table on the source; click a table to see its columns, row count, and column count
Open Data Modelling UISwitches from the Schema Explorer into the model builder
Back to Schema ExplorerReturns from the model builder to the plain browser
Model gridShows every model already built as cards, each with name, description, a materialization label, and when it was last updated
New modelOpens a dialog to create a model (name, optional description, and materialization choice)
Model card (click)Opens that model in the editor
Card menu → EditOpens the model in the editor
Card menu → DeleteMoves the model to Trash

Inside the model editor

Control / ButtonWhat it does
Entities list (left)The physical tables on this source; add one to the model with a click
Suggestions list (left)AI-proposed relationships between tables, each with a confidence score, plus accept (✓) and dismiss (✗) buttons
CanvasThe working area where tables appear as boxes and relationships as lines between them
Zoom in / Zoom out / Fit viewNavigate the canvas
Properties panel (right)Edit the selected model’s details, relationships, the columns each table contributes, and preview the generated query
MaterializePersists the model so it can be queried directly; available only for non-ephemeral models
Save ModelSaves your current changes
Models (back button)Returns to the model list; prompts you to confirm if you have unsaved changes
Materialization choices: Ephemeral (defined but not stored), View, Table, and Incremental. Only non-ephemeral models can be materialized.

Schema Drift

A data source’s structure can change underneath you. The Schema Drift area watches for these changes, records them as alerts, and lets reviewers triage and resolve them. Each alert carries a severity (Info, Warning, or Critical), when it was first detected, how many times it has been seen, and a structured before/after view.
Control / ButtonWhat it does
Open alerts tabThe active triage list — drift changes still needing attention
History tabAll alerts regardless of status, for review and audit
Snapshots tabThe timeline of captured structure snapshots
Policy tabConfigure who is notified about drift and when
Alert row (click)Opens the structured detail: tables and columns added, removed, or changed
ViewOpens the same alert detail
AcknowledgeMarks an open alert as seen, with an optional note; recorded in the audit trail
ResolveCloses out an alert once handled; logged in the audit trail
View snapshot diffCompares previous and new captured structures side by side
Status indicatorsEach alert shows Open, Acknowledged, Resolved, or Auto-resolved

Drift alert policy

Control / FieldWhat it does
RecipientsChoose which roles receive the alert email — data source owner, original creator, department admins, org admins
Custom emailsAdd extra recipients beyond the role list
Severity thresholdSet the minimum severity that triggers a notification
Quiet hoursDefer notifications during a chosen time window and time zone
Save / ResetSave your policy changes, or discard them
If no recipients are selected, drift is still recorded on screen but no alert email is sent.

Who can use it

  • Owners and editors of a data source build and curate its models and triage drift.
  • Viewers can browse the Schema Explorer, open models, and review drift alerts, but cannot change models or resolve alerts.
  • Department admins and org admins are the natural default recipients for drift alerts.