Forum Discussion
Composite / Hybrid Tables
Hello,
i was just wondering about what are the best practices when creating a composite/hybrid table? splitting the data between hot and cold. take into consideration that we have a table in a fabric warehouse that contains well over 200M transactions, and we need to extract the data to Power BI with the best achievable performance, what are the steps that we should follow? should we split the data between hot and cold in fabric, then extract that data to power bi and combine the two tables with a union or how is it exactly done?
Hi Hussein_charif ,
Below article may help in deciding the strategy.
https://dax.tips/2021/09/06/intro-to-power-bi-aggregations/
This article covers the type of aggregates.Phil Seamark has covered in detail.
this article covers the hot cold partitioning mechanism
Please give kudos or mark it as solution once confirmed.
Thanks and Regards,
praful
5 Replies
composite/hybrid is not the right term here. What you are looking for are aggregations.
Hi Hussein_charif ,
Try below things.
1. Split the data in Fabric not inside Power BI.
2. Use Import for Hot and DirectQuery for Cold.
3. Don’t UNION inside Power BI. Composite model takes care of merging both layers.
4. Ensure both tables have identical schemas, identical keys and same relationships to dimensions.
5. Keep Import small and DirectQuery narrow.
If my response as resolved your issue please mark it as solution and give kudos.
Hi Hussein_charif ,
Below article may help in deciding the strategy.
https://dax.tips/2021/09/06/intro-to-power-bi-aggregations/
This article covers the type of aggregates.Phil Seamark has covered in detail.
this article covers the hot cold partitioning mechanism
Please give kudos or mark it as solution once confirmed.
Thanks and Regards,
praful
Hi Hussein_charif
Following up to confirm if the earlier responses addressed your query. If not, please share your questions and we’ll assist further.Hi Hussein_charif
Have you had a chance to look through the responses shared earlier? If anything is still unclear, we’ll be happy to provide additional support.
