Do we need to as many capacities as workspaces in an embedded workspace-based isolation scenario?

[dpiret]

Hi there.

We are evaluating the best scenario for a PBI embedded multi-tenant SaaS application. Because content customization and authoring per tenant is a requirement and the document Multi-tenancy solutions with Power BI embedded analytics recommends "Workspace separation is the recommended embedded analytics multi-tenancy solution. ", we are considering using workspace-based isolation.

But also the document Move your embedded app to production states, "Each workspace that contains a Power BI item related to the embedded content, must be assigned to capacities. For example, if an embedded report and the dataset bound to it reside in different workspaces, both workspaces must be assigned to capacities."

My question is whether all workspaces (aka tenants) must be assigned to different capacities, or can all be assigned to one (or more if required). If the case is the former, the solution would not be economically viable, hence the question.

Thanks for your help!

[DesNaidoo · Accepted solution]

Hi dpiret
All your workspaces reside within a tenant. You can have one capacity that all your workspaces can share or you can choose to split up your capacity based on your requirements
e.g. a P1 comes with 8 vCores. You can choose to create a 2 core capacity and assign that to your test workspaces and a 6 core for prod. Or just simply create 1 with 8 cores and assign to all your workspaces. 
Its down to what would work for you and your use-case