Forum Discussion
Datasize too big for one semantic model - best practise
Hi Community!
I have a couple of questions in regards to "best practise" when developing a golden / one source of truth semantic model 🙂
We are trying to build one semantic datamodel where all reports are supposed to have a live connection to the semantic model in order to minimize the maintenance of different datasources and business logic. We have Pro licenses and from my understanding 1GB of semantic model size. We want to have both sales, purchase and inventory data in our combined semantic model. However, we can already see some issues due to dataset size being larger than 1GB. Therefore we have looked into Premium pr. user licenses.
Now onto my questions:
1. What is the maximum size of a semantic model when you have a premium pr. user license? 100TB?
2. What is best practise when developing a semantic model where you want to have different datasources / business areas combined but the datasize is too big?
- Is it to make more than one semantic model and then try to make some composite models?
- And would that not be the opposite of trying to make one golden semantic model?
4 Replies
- Idrissshatila
Super User
Hello Isildur13 ,
1- premium per user will allow you to store up to 100 GB.
https://community.fabric.microsoft.com/t5/Community-Blog/Power-BI-Premium-VS-Power-BI-Premium-Per-User/ba-p/2202692
2- you can have different fact tables that represents different busiess areas in one semantic model, just make sure to build it as a star schema data model https://learn.microsoft.com/en-us/power-bi/guidance/star-schema
also just have the relavent columns you need, not to get all the data. - 123abc
Community Champion
The maximum size of a semantic model with a Premium Per User (PPU) license in Power BI is 100 GB. This is significantly larger than the 1GB limit with a Pro license, allowing for more extensive data models.
When developing a semantic model that combines different data sources and business areas, and the data size is too large, best practices include:
- Enabling large semantic models: If you’re using Power BI Premium, you can enable the Large semantic model storage format to allow your semantic models to grow beyond 10.
- Incremental refresh: Configure incremental refresh policies to load only the data that has changed, reducing the overall size and improving refresh performance.
- Composite models: If the data size still exceeds limits, consider creating more than one semantic model and using composite models to combine them. This approach allows you to maintain a single source of truth while managing large datasets effectively.
Creating multiple semantic models and using composite models can seem contrary to the idea of a single golden semantic model. However, it’s a practical solution to handle large datasets while preserving the benefits of centralized business logic and data sources. It’s about finding the right balance between a unified semantic layer and the technical limitations of the platform.
Beow Link may help you to solve your issues:
https://learn.microsoft.com/en-us/power-bi/enterprise/service-premium-large-models
If this post helps, then please consider Accepting it as the solution to help the other members find it more quickly.
In case there is still a problem, please feel free and explain your issue in detail, It will be my pleasure to assist you in any way I can.
Hi 123abc
Thanks for the links. I have som follow up questions 🙂
1: If we choose to upgrade our licenses from Pro to Premium pr. users - do when then need to upgrade all users who only needs to be able to consume the reports from Pro to Premium as well or is it possible to create a premium workspace where the semantic model is located and then a pro workspace where all the reports is located with a live connection to the semantic model in the premium workspace?
2: Also, if we choose to enable incremental refresh or large semantic model format would we then need to use external tools like tabular editor 2 or is it possible to store a "backup" of the semantic model report with a subset of data and then when changes has been applied or new business logic added then just upload the semantic model afterwards, then do a full load and lastly enable incremental or large dataset on the semantic model again?3: If we choose to make more than one semantic model and then create som composite models how do you then handle ex. a Calendar dimension. Would that be added to all 3 semantic models (sales, purchase, inventory) so you in reality would have three calendar dimensions to maintain or is there a smarter way of doing that sort of thing?
- Brunner_BI
Impactful Individual
A cheap way to reduce the size of your model is to run Measure Killer (external tool) to kick out everything that you do not need.
It can even analyze a shared dataset so you can include all reports that you need to support with your centralized model.
