Skip to Content

Core Concepts

Datamarts

Choose the tables, columns and joins people can use, and how they are scoped per customer.

What is a datamart?

A datamart is a curated view of one datasource. It holds the tables and columns people can use, how those tables join, and which customer each row belongs to. Metrics, dashboards and Bryn read from it instead of the raw warehouse.

A datasource is the connection. A datamart is what you choose to expose from it. A workspace can read a datasource directly, but a datamart gives you control and tenancy in one place.

Datamart editor with the Tables & Columns, Relationships and Tenancy tabs, the schema tree and the column list of one table
A datamart in the editor. Tables & Columns lists the columns, types and display labels of each table.

What a datamart holds

PartWhat it decidesWhere you set it
Tables and columnsWhat people can query, with labels, aliases and default sortingTables & Columns
RelationshipsHow tables joinRelationships
TenancyWhich rows belong to which customer: Table-level (a client column) or Schema / database-levelTenancy
DescriptionsWhat tables and columns mean, so Bryn and your team pick the right fieldsThe description page. See Describe tables and columns

Datamarts and semantic models

Databrain shows one of two names in Data Studio, depending on your company:

  • Datamarts and Model status. You build a datamart in four steps: pick tables, choose columns, connect tables, scope data per customer. The datamart shows as Classic model.
  • Semantic Models. New companies see this list. You build a model in the semantic studio, alone or with Bryn, and publish it when its checks pass. The list shows Never published or Published.

Both are the same kind of object. Both feed workspaces and embeds, and both hold tenancy.

Semantic studio for a new model with the Build a model from card, the Build my model, Pick tables and Paste YAML options, and the Bryn panel
A new semantic model opens on its start card. Nothing goes live until you publish.

When to use one

  • You embed dashboards for customers. The datamart’s tenancy scopes every embedded query to one customer. Its tenancy wins over the datasource’s.
  • Your warehouse has many tables. Expose only the ones people need, with clear names.
  • You use Bryn. Descriptions and synonyms tell Bryn which fields to use.

Create one

Before you start, you need a connected datasource, and you must be an Admin or a Developer, or have the datamart Create permission.

New datamart dialog with Datamart name, Data source set to Product analytics, Tenancy level Table, and Create datamart button
The New datamart dialog. Name it, pick the datasource and check the tenancy level.
  1. Go to Data Studio > Datamarts and select New datamart.
  2. Enter a Datamart name, select the Data source and check the Tenancy level.
  3. Select Create datamart, then follow the four steps in the editor.

Full steps: Create a datamart. With Semantic Models: Create a semantic model.

Next steps