2

Data Integrations

Last Update 7 個月前

How to integrate Birdie + Gong Birdie + Databricks Integration

Birdie + Databricks Integration

Willian Macedo

Last Update 7 個月前

This guide explains how Birdie integrates with Databricks using read-only SQL queries. Birdie supports different integration models depending on how Databricks is configured in your organization.

Supported Integration Models

1. OAuth Integration (Service Principal) (recommended)

Birdie authenticates using a Databricks service principal and short-lived OAuth tokens issued by the workspace OIDC endpoint.

This model is recommended when:

  • You want OAuth-based, non-interactive access

  • You run automated or production ingestion pipelines

  • You want to avoid long-lived credentials



2. Personal Access Token (PAT)

Birdie authenticates using a technical user and a Personal Access Token.

This model is commonly used when:

  • OAuth is not yet enabled


  • You are running a proof-of-concept or non-production setup

Create the Identity Used by Birdie

You may use a Service Principal (recommended) or a technical user.

Service Principal

  1. Go to Settings - Workspace admin - Identity and access - Service principals

  2. Click Add service principal

  3. Name it birdie

  4. Enable:

    • Workspace Access

    • Databricks SQL Access

  5. Generate a client secret

Technical User + PAT

  1. Create a user named birdie

  2. Grant Workspace Access and Databricks SQL Access

  3. Generate a Personal Access Token (PAT)

Grant Read-Only SQL Permissions

Unity Catalog environments

Legacy / Hive Metastore environments

Repeat SELECT grants for each table or view Birdie should ingest.

Validate Authentication (OAuth example)

Using your workspace hostname:

You should receive an access_token.

Validate Databricks API Access

Expected response:

Validate SQL Execution

If this returns 1, Birdie can execute SQL queries inside your warehouse.

Share Connection Details with Birdie

Provide the following information securely:

  • Workspace URL


  • Authentication method (OAuth or PAT)

  • Client ID + Client Secret or PAT

  • SQL Warehouse ID

  • Catalog, schema, and table names

  • Partition column used for incremental ingestion

Data Types Supported

Birdie can ingest structured datasets exposed in Databricks tables or views, including:

  • Conversations and messages

  • Feedback datasets:

    • nps
    • csat
    • review
    • survey

  • Operational or reference tables (accounts, users, metadata)

Each dataset must be exposed as:

  • A table, or

  • A view

One table or view per dataset type.

 

Was this article helpful?

0 out of 0 liked this article