How Azure Databricks pre-purchase discount is applied

You can use pre-purchased Azure Databricks commit units (DBCU) at any time during the purchase term. Any Azure Databricks usage is deducts from the pre-purchased DBCUs automatically.

Unlike VMs, pre-purchased units don't expire on an hourly basis. You can use them at any time during the term of the purchase. To get the pre-purchase discounts, you don't need to redeploy or assign a pre-purchased plan to your Azure Databricks workspaces for the usage.

The pre-purchase discount applies only to Azure Databricks unit (DBU) usage. Other charges such as compute, storage, and networking are charged separately.

Pre-purchase discount application

Databricks pre-purchase applies to all Databricks workloads and tiers. You can think of the pre-purchase as a pool of pre-paid Databricks commit units. Usage is deducted from the pool, regardless of the workload or tier. Usage is deducted in the following ratio:

Workload DBU application ratio — Standard tier DBU application ratio — Premium tier
Data Analytics 0.4 0.55
Data Engineering 0.15 0.30
Data Engineering Light 0.07 0.22

For example, when a quantity of Data Analytics – Standard tier is consumed, the pre-purchased Databricks commit units is deducted by 0.4 units. When a quantity of Data Engineering Light – Standard tier is used, the pre-purchased Databricks commit unit is deducted by 0.07 units

Determine plan use

To determine your DBCU plan use, go to the Azure portal > Reservations and click the purchased Databricks plan. Your utilization to-date is shown with any remaining units. For more information about determining your reservation use, see the See reservation usage article.

How discount application shows in usage data

When the pre-purchase discount applies to your Databricks usage, on-demand charges appear as zero in the usage data. For more information about reservation costs and usage, see Get Enterprise Agreement reservation costs and usage.

Need help? Contact us.

If you have questions or need help, create a support request.

Next steps