> ## Documentation Index
> Fetch the complete documentation index at: https://docs.planasonix.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Semantic layer

> Define business metrics and a semantic layer for consistent analytics.

The **semantic layer** stores curated **metrics**, **dimensions**, and **entities** so every BI tool and SQL client asks the same questions the same way. You define grain, aggregation rules, and friendly names once, then expose them to analysts through Planasonix.

## Metrics catalog

Create entries such as `net_revenue`, `active_subscribers`, or `churn_rate` with:

* **Definition** – SQL or declarative expression referencing approved fact and dimension tables
* **Grain** – Per day, per account, per subscription, etc.
* **Constraints** – Filters that must always apply (exclude test tenants, internal accounts)
* **Certification** – Steward approval for executive use

Version metrics when definitions change; dashboards pin to versions until owners accept updates.

<Tip>
  Keep metric names business-stable even when underlying table names churn; update mappings in the semantic layer instead of renaming metrics constantly.
</Tip>

## Semantic definitions

**Dimensions** describe `who`, `what`, `where`, and `when` (product line, region, campaign). **Entities** map keys across facts (customer\_id bridges orders and support tickets). Document slowly changing dimensions explicitly so analysts know which history version they query.

<AccordionGroup>
  <Accordion title="Drill paths">
    Define allowed hierarchies (region → country → city) so self-serve tools expose consistent drill-downs.
  </Accordion>

  <Accordion title="Access control">
    Row-level security rules in the semantic layer should mirror warehouse entitlements; never widen access accidentally through friendly views.
  </Accordion>
</AccordionGroup>

## JDBC access

BI tools that speak **JDBC** can connect to Planasonix’s semantic JDBC endpoint (when enabled for your tenant) and issue SQL against published models. The driver translates requests to governed queries against your warehouse connection profile.

<Tabs>
  <Tab title="Tool compatibility">
    Tableau, Power BI, Looker (via database connections), and Excel Power Query commonly use JDBC/ODBC bridges—verify exact support with your CS engineer during onboarding.
  </Tab>

  <Tab title="Performance">
    Push heavy computation to the warehouse; the semantic layer should add predicates and joins, not materialize massive intermediate sets unless you design aggregates for that purpose.
  </Tab>
</Tabs>

<Warning>
  JDBC credentials are as sensitive as warehouse passwords. Store them in your vault and rotate on the same schedule as database keys.
</Warning>

## Related topics

<CardGroup cols={2}>
  <Card title="Data catalog" icon="books" href="/governance/data-catalog">
    Discover physical tables behind semantic models.
  </Card>

  <Card title="Impact analysis" icon="chart-network" href="/governance/impact-analysis">
    See downstream BI assets when metrics change.
  </Card>
</CardGroup>