Databricks Clean Room costing

[ShankarM]

hi,

Can you throw some light on how the compute, data sharing costing is done for various scenarios:

1. Collaborator 1 and Collaborator 2 are having Databricks accounts in the same region and same cloud. Is there a DBU cost and who will pay for it? I assume there is no egress cost since they are in same cloud account 

2. Collaborator 1 is non databricks source account ( e.g Oracle through Lakehouse federation ) and collaborator 2 is databricks account is in different region. Is there a DBU and egress  cost and who will pay for it 

3. Collaborator 1 and Collaborator 2 are having Databricks accounts in the same region but different cloud env. 

Is there a DBU and egress cost and who will pay for it?

4. As I understand the consumer can only see the metadata and not the actual data of the publisher. Can we provide access to row level data of publisher to consumer? If the data is masked by publisher will it show same to consumer as well

[Kaniz_Fatma · Accepted solution]

Hi @ShankarM, 

Collaborators in the same region and cloud:

• If both Collaborator 1 and Collaborator 2 have Databricks accounts in the same region and cloud, they will incur DBU (Databricks Unit) costs based on their usage. DBUs represent compute resources used by Databricks workloads.
• Since they're in the same cloud environment, there won't be any egress costs for data transfers between them within that cloud environment.
• Typically, each collaborator pays for their own DBU usage.

Collaborators in different regions:

• In this case, if Collaborator 1 is outside Databricks and Collaborator 2 is using Databricks in another region, Collaborator 2 will still face DBU costs.
• There might be egress costs associated with transferring data between different regions, depending on the data movement.
• Payment responsibilities depend on their agreement.

Collaborators in the same region but different cloud environments:

• When both collaborators have Databricks accounts in the same region but different cloud environments (e.g., AWS and Azure), they will each bear DBU costs.
• Egress costs may apply if data needs to move between different cloud providers.
• Payment arrangements are determined by their agreement.

Data visibility for consumers:

• By default, consumers can view metadata like table schemas shared by the publisher.
• Access to row-level data requires setting up appropriate permissions using Databricks’ access controls.
• Data masking applied by the publisher (e.g., column-level security) will also affect what consumers can see.

For more accurate cost estimates, you can use the Databricks Pricing Calculator.