Connect a database and register a dataset

This guide is for administrators. A connection opens the door to a database; a dataset defines which table the team can work with.

1. Prepare a read-only account

Before opening CORE FIELD, request a dedicated database account with access only to the required schemas and tables.

It should be able to run SELECT. Avoid INSERT, UPDATE, DELETE, table creation, or administrative privileges unless they are truly necessary.

Never document the real connection URI

Passwords, tokens, and connection URIs must not appear in screenshots, queries, Markdown files, or Git.

2. Connect and test

1. Open Data → Connect database.
2. Choose the correct engine.
3. Enter the details in the protected form.
4. Select Test connection.
5. Save only after the test succeeds.

A successful test confirms connectivity and authentication. You still need to limit which datasets and which people can query it.

3. Register robotic.sensor_data

1. Open Assets → Datasets.
2. Select + Dataset.
3. Choose the database, the robotic schema, and the sensor_data table.
4. Confirm the registration.
5. Open the dataset and review its columns.

4. Prepare the meaning

• Mark timestamp as the time column.
• Add plain-language descriptions for CPPW and THHN_1…THHN_4.
• Define which columns can filter or group data.
• Create metrics whose names explain their aggregation.
• Assign a responsible owner.

A well-prepared dataset makes time, columns, and metrics predictable in Explore.

5. Give access without opening everything

Grant the exact datasource access permission to the appropriate role. Database-wide access can expose more datasets and SQL Lab capabilities than intended.

Read Configure roles and data sources and test with a non-administrator account.

✓

Resultado esperadoExpected result

The connection uses least privilege, robotic.sensor_data appears as a dataset, and an authorized account can use it without receiving global access.