Materialization
Materialization is a performance optimization for Custom Models with complex or heavily queried data. When enabled, Lakehousecat pre-computes the model's central analytical views and stores the results — so chart and dashboard queries read from pre-built data instead of re-running joins on every request.
When to Enable
Materialization is beneficial when:
- Charts and dashboards on this model load slowly.
- The model combines data from many data sources with complex relationships.
- The model is used frequently by many users simultaneously.
For simple models or infrequently used models, the default (non-materialized) behavior is sufficient.
How to Enable
Open the Custom Model editor, go to the General tab, and switch on Enable Materialization.
Click Save to apply the setting. The change takes effect automatically — no semantic rebuild is required.
Materialization can only be enabled by Administrators and Builders.
What Happens When Materialization is Active
Faster chart queries
The analytical views that power charts are stored as pre-computed results instead of recalculated live. Chart and dashboard load times improve, especially for models that join many data sources.
Automatic refresh after data loads
Whenever data in a connected data source is refreshed (full load or incremental), the materialized views for this model are updated automatically. Chart data stays current without any manual action.
Storage overhead
Pre-computing the views requires additional storage proportional to the size of the model's analytical data. The extra storage is released automatically if materialization is disabled.
Enabling and Disabling
Materialization can be toggled at any time from the General tab without rebuilding the semantic model.
| Action | Effect |
|---|---|
| Enable | Converts the model's analytical views to pre-computed form. Takes effect immediately after saving. |
| Disable | Reverts to live-computed views. Pre-computed data is released. Takes effect immediately after saving. |
Chart names and connections are preserved in both states — existing charts and dashboards continue to work without changes.
Important Notes
- Deleting the semantic model removes the materialized views along with the rest of the model's data layer. When you rebuild the semantic model, materialization is re-applied automatically if the setting is still enabled.
- Updating the semantic model (not deleting) does not affect materialization — the views remain in place and are refreshed on the next data load.
- Materialization is applied per model. Enabling it for one Custom Model does not affect other models, even if they share the same data sources.
Best Practices
- Enable materialization for models that drive production dashboards or are shared with many users.
- Leave materialization off during initial setup and testing — enable it once the model is stable and the data layer is finalized.
- If storage cost is a concern, limit materialization to the models where query performance matters most.
See General for the full list of settings in the General tab.