Hi @h_h_ak  Thank you for your reply. Do you mean SQL warehouses (Serverless SQL)?

One reason we don't use it is because the price. In our region, the cheapest all-purpose compute and SQL warehouse are:

1. All-purpose compute: D4ads v5 - USD $0.81/hour (DBU+VMs)
2. Serverless SQL warehouse: 2X-Small (Standard_E8ds_v4) - USD $3.80/hour (DBU)

I know 2X-Small is faster than D4ads v5.

However, currently we import data from Databricks to Power BI in a daily schedule. Therefore, we don't care too much about the compute startup time and the querying performance.

Moreover, Serverless SQL warehouse does not like Snowflake, the Serverless SQL warehouse will auto stop after 10 minutes of inactivity. That means, we need to pay 10 more minutes for nothing.

Therefore, I think job compute is the best option to use. However, Power BI connect does not support using Databricks job compute.

Btw, whatever compute we use, we still need to use PAT of the Service Principal to authenticate the Power BI Service with Databricks workspace.

Thank you.

Regards,

Albert