Forum Discussion
suggestion needed to optimize report size
Hello!
In our project, we have 20 different individual reports with nearly 50 tables to serve these reports all together to go in a single .pbix file. The file size is already nearly 2GB. We are doing in import mode(our priority). As the file size has become very big, we are unable to move ahead with the development of the report. The performance of the report is very horrible. If we want to switch to either direct query or composite mode, what will be the disadvatages? Any suggestions which mode is appropriate:Import/direct/dual?
FYI: Our reports are mostly table visuals.
Thanks
3 Replies
- amitchandak
Super User
power_user_123 , If you are a premium customer, try to reduce the size using incremental data settings of the development file and load data on the service
Deployment Pipeline, Load More Data on Test/Prod : https://youtu.be/l69cnWkoGX0
Other pro options
also, refer
https://www.youtube.com/watch?v=RnDdDlozcdo
https://www.youtube.com/watch?v=qZOEDBedATA
https://apexinsights.net/blog/top-5-tips-to-optimise-data-model
- ashishg
Advocate I
Hi,
You can go with Power Bi Dataflows so import your all table in Dataflows then after that import that dataflow in your pbix file so you get best dataset compression with increase in performance also and your dataset size will be nearly 300mb. I have used in my project it will help me alotHope its Helpful !!!
- KNP
Difficult to answer with the information provided. It is very much an, "it depends".
That said, things to look at.
- Do you actually need all 50 tables?
- Is it correctly modelled as a star schema? (this is critical)
- Have you removed ALL unnecessary/unused columns?
- Have you enabled incremental refresh on the large fact tables? (this, combined with setting the RangeStart and RangeEnd parameters to a smaller period should allow you to continue development and ensure quick refresh times once deployed to the service)
- Import mode is always preferred unless data is extremely large.
- Have you considered using Dataflows for the initial ETL?
- Have you separated the data model from the report files? (I assumed it is based on your comments)
- Do you have premium licenses?
Questions.
- What is the datasource(s)? Does it fold?
- How many rows in your largest fact tables?
- What about the performance is "horrible"?
Hopefully this will give you a few things to look at.
