Setup unity catalog external location to minio

[gwilson]

We have a minio server running in Azure that we have connected to the spark clusters directly. As we move to unity catalog, we would like to make the data stored in our minio servers accessible as an external location in Azure Databricks account via unity catalog. Is it possible to connect a non-ADLS object storage as an external storage location in Azure Databricks?

[Kaniz]

Hi @gwilson,  Here are the steps to connect a Minio server to Azure Databricks as an external storage location:

1. Configure your Minio server to allow network access and obtain the endpoint URL, access key, and secret key.

2. Set the Spark configuration values in the spark.conf file of your Azure Databricks workspace to interface with Minio using the s3a connector.

3. Register the Minio storage location as an external table in Unity Catalog (or using the Databricks CLI) using the Delta format.

4. Access the data stored in Minio via the external table minio_data from your Spark scripts.

Following these steps, you can connect your Minio server to Azure Databricks as an external storage location and use it for your Spark workload.

[174817]

Hi @Kaniz ,

I have a server on Azure that supports the S3 protocol, and I am trying to follow these instructions in order to use Unity on Azure DataBricks with it.  I am not sure about this part of your reply:

---

Set the Spark configuration values in the spark.conf file of your Azure Databricks workspace

I know how to configure Spark on a DataBricks Compute Spark cluster, but I could not find where to do this on a Unity catalog.  I tried in the web GUI under "Catalog > Catalog Explorer > External Data" and also under "Compute / SQL Warehouses".

Is it possible to set a Unity Catalog external location to access an S3-compatible endpoint on Azure?

Thanks!

[gwilson]

Support for S3-compatible storage is not available with unity catalog.