Official documentation on resource allocation for shared capacities

[AMeyersen]

I am looking for documentation on how shared capacities allocate resources to users but have not been able to find enough details.

One of my clients is a medium-sized company. Neither the dataset size nor the user count justifies a dedicated premium (or fabric) capacity, so the solution runs on shared capacity using Power BI Pro licenses.

We are planning to increase the user count from approximately 10 test users to around 100 production users.

I would like to understand whether a performance drop for report consumers is to be expected. Does Power BI allocate a maximum amount of shared resources per tenant or workspace or report? Or does it grant resources dynamically per user request, regardless of the request source?

I was not able to find any reliable information except the basic restrictions on dataset size and refresh frequency. 

As far as I understand, resources are allocated per user request, so users should not experience a performance drop. However, there might still be some impact, somehow. Unfortunately, I have been unable to find reliable sources that explain the finer details of resource sharing and how it affects different usage scenarios.

PLEASE do not post a generic ChatGPT response - I wouldn't post a question here if ChatGPT had given me a helpful answer.

[ibarrau · Accepted solution]

Hi. I don't have the details and docs. I can say that shared capacity has the limits you have found for sizing reports at workspace level. As long as the development is not over the limits, it won't suffer a performance drop due to concurrency or capacity. If you have a performance issue, it's a modeling or dax issue, but the shared capacity is not the problem.

Just for the heads up. Nowadays Fabric has so many sku/plans that it's not that expensive as premium was before. Just saying, it's not for making you change. We have been working with only pro for the last 9 years (100 users) and no issues.

I hope that helps,