Apache Druid

A native connector to Druid ships with Superset (behind the DRUID_IS_ACTIVE flag) but this is slowly getting deprecated in favor of the SQLAlchemy / DBAPI connector made available in the pydruid library.

The connection string looks like:

druid://<User>:<password>@<Host>:<Port-default-9088>/druid/v2/sql

Here’s a breakdown of the key components of this connection string:

• User: username portion of the credentials needed to connect to your database
• Password: password portion of the credentials needed to connect to your database
• Host: IP address (or URL) of the host machine that’s running your database
• Port: specific port that’s exposed on your host machine where your database is running

Customizing Druid Connection

When adding a connection to Druid, you can customize the connection a few different ways in the Add Database form.

Custom Certificate

You can add certificates in the Root Certificate field when configuring the new database connection to Druid:

When using a custom certificate, pydruid will automatically use https scheme.

Disable SSL Verification

To disable SSL verification, add the following to the Extras field:

engine_params:
{"connect_args":
    {"scheme": "https", "ssl_verify_cert": false}}

Aggregations

Common aggregations or Druid metrics can be defined and used in Superset. The first and simpler use case is to use the checkbox matrix exposed in your datasource’s edit view (Sources -> Druid Datasources -> [your datasource] -> Edit -> [tab] List Druid Column).

Clicking the GroupBy and Filterable checkboxes will make the column appear in the related dropdowns while in the Explore view. Checking Count Distinct, Min, Max or Sum will result in creating new metrics that will appear in the List Druid Metric tab upon saving the datasource.

By editing these metrics, you’ll notice that their JSON element corresponds to Druid aggregation definition. You can create your own aggregations manually from the List Druid Metric tab following Druid documentation.

Post-Aggregations

Druid supports post aggregation and this works in Superset. All you have to do is create a metric, much like you would create an aggregation manually, but specify postagg as a Metric Type. You then have to provide a valid json post-aggregation definition (as specified in the Druid docs) in the JSON field.