Skip to main content

​
Overview

Build a connector when you need to query Unify data on a schedule and load it into a warehouse or another destination. A connector coordinates the query job lifecycle, writes each result row to the destination, and saves a checkpoint so the next sync only requests new or changed data. Unify maintains example connectors that you can run without modification or adapt to another platform. Each example implements job polling, pagination, retries, and incremental syncs.
unifygtm/bulk-api-connector-examples
Loading repository data...

​
Example connectors

Fivetran Connector SDK

Start with a Python connector that queries data incrementally and upserts rows into a Fivetran destination.

Airbyte

Start with a declarative source that models query jobs with Airbyte’s asynchronous stream support.
The Fivetran example is useful when you want to customize connector behavior in Python. The Airbyte example is useful when you prefer a declarative manifest that you can inspect and test in Connector Builder.

​
How the examples work

1

Create a query job

The connector creates a job for each resource being synced. After the first sync, it filters the query using the checkpoint from the previous run.
2

Wait for the job to finish

The connector polls the job with backoff until it reaches FINISHED. It retries temporary failures and respects the Retry-After header when rate limited.
3

Load the results

The connector retrieves each result page and writes the rows to the destination. Rows are keyed by stable IDs so overlapping incremental queries can be processed idempotently.
4

Save a checkpoint

After the rows are loaded successfully, the connector saves the newest update time. The next run uses that checkpoint to query new or changed data.
See Using query jobs for more information about creating jobs, polling their status, and retrieving results.

​
Prerequisites

Before using an example, make sure you have:
  • A Unify API key associated with a user.
  • Access to the destination where you want to load data.
  • The local development tools required by the example’s README.
  • A list of the resources and fields that you want to sync.
Store credentials in a local, ignored configuration file. Do not commit API keys or destination credentials to your repository.

​
Getting started

1

Choose an example

Select the Fivetran or Airbyte implementation based on your destination and preferred development model. Open its README for the current installation and runtime requirements.
2

Configure the connector

Add your Unify API key and choose which resources to query. For object records, also select the attributes that should be included in each result.
3

Test a focused sync

Start with one resource and a recent checkpoint. Run the connector locally, inspect the destination rows, and verify that a second run only processes new or changed data.
4

Customize the data model

Adjust resource selection, field mappings, destination table names, and concurrency for your use case. Keep query job creation rate limits in mind when syncing multiple resources in parallel.
5

Deploy and monitor

Follow the example’s deployment instructions for your platform. Monitor job failures, expired results, retries, and checkpoint advancement as part of normal connector operations.