Some confusions in the storage capacity associated with Pro

[TicketRaiser]

Hi

I have a ~2GB pbix file

license is power bi Pro with F32 fabric capacity

Now am i getting the below error due to the license or could this be because of something to do with the data/calculations?

Even though the pbix file is 2GB, i am still able to publish the dashboard into service with pro license, but this above error pops up when i try doing a manual refresh in the service after publishing the dashboard.
And inside workspace settings --> system storage the size of the semantic model is around 5 GB

And in this fabric metrics dashboard the file size is showing around 60 GB with CU usage going very high

I am so confused with all thsese figures and want to know if a pro license can solve this or is this an optimization issue 

some data metrics:

- there are around 17 tables we are bringing in using import, 2 of them are from bigquery where the combined physical storage and logical storage each are ~2gb

- other tables are bring imported from adls which are parquet files, organised in hierarchial namespace

[TicketRaiser]

  thanks for reaching out, below are the steps i have taken so far and still we are facing the refresh issue.

Many to Many Relationships

• There were 2 many to many relationships, solved it by introducing dim table

Date/Time tables

• There were many date/time table created by power bi inherently leading to memory bloat
• Removed 30 odd tables; reconfigured the date/time intelligence options and refreshed the desktop dashboard

Rewriting Measures

• The measures were checked for the time taken using DAX studio server timings
• Time heavy measures were optimized to be efficient and tested using 

  the logical size which was previously 60 gb has come down to ~7GB which apparently is still not good enough

Below are the next steps we are aiming for:

• Do the deduplication measures at the etl level
• removing columns 
• data resize  - last resort

[v-saisrao-msft]

Hi @TicketRaiser,

Have you had a chance to review the solution we shared by @R1k91 ? If the issue persists, feel free to reply so we can help further.

Thank you.

[R1k91 · Accepted solution]

I would suggest to use dax studio and its view metrics to reduce the semantic model size and do model optmiziation

- identify big columns/tables

- verify data types (datetime when only date is needed is a waste of space for example)

- verify if columns are used; if not remove

- verify if table are used; if not remove

- remove useless rows

[v-saisrao-msft]

Hi @TicketRaiser,

Have you had a chance to review the solution we shared by @rohit1991? If the issue persists, feel free to reply so we can help further.

Thank you.

[rohit1991]

Hii @TicketRaiser 

It’s a capacity and model optimization problem. Your 2 GB PBIX expands significantly in the service due to import mode, column encoding, BigQuery data, and Large Semantic Model (LSM) behavior, which explains the 5 GB semantic model size and high in-memory (CU) usage during refresh. The 60 GB you see reflects memory consumption during refresh, not stored size. A Pro license alone won’t fix this; you need to optimize the model (reduce columns/cardinality, review BigQuery imports) or scale the Fabric capacity to handle the refresh workload.

[TicketRaiser]

Data source error: Resource Governance: This operation was canceled because there wasn't enough memory to finish running it. Either reduce the memory footprint of your dataset by doing things such as limiting the amount of imported data, or if using Power BI Premium, increase the memory of the Premium capacity where this dataset is hosted. More details: consumed memory 9671 MB, memory limit 9671 MB, database size before command execution 568 MB. See https://go.microsoft.com/fwlink/?linkid=2159753 to learn more. Cluster URI: WABI-WEST-US-E-PRIMARY-redirect.analysis.windows.net Activity ID: 2665768e-fd35-4c15-b96f-f5177e5d7bcb Request ID: cb6296a9-39b9-10d8-fbd0-c6adfd3691d1 Time: 2026-02-11 10:14:24Z?

hi @rohit1991 after all the optimizations we did like reducing columns and optimizing dax, this is the error its throwing.