Forum Discussion
How to combine multiple existing Power BI datasets
We are likely targeting a solution that looks like below. Its not a strong solution as the lineage is a little messy but with the reports now in a common location, we can move forward with delivering the solution to the intended audience
-in workspace1 create a visual-a
-in workspace2 create a visual-b
-in workspace3 create a visual-c
-'save a copy' of visual-a to workspace4
-'save a copy' of visual-b to workspace4
-'save a copy' of visual-c to workspace4
---------------------------------------------
-deliver the 3 reports in the workspace4 app as tabs on left menu
OR
-deliver the 3 reports in a dashboard within workspace4
I am trying to work out is this functionaility has a name. It looks like a super powerful feature- but not shouted about. Any thoughts on risks and considerations if using in " in anger"
OUr use case is simple. I have large fact tables which will be applied to each domain model, and I also have a massive customer dimension (lucky me)
So having a model for customer, and a model for each domain allows me to toad the customer once and reuse. It also allows me to combine domains to compare "stuff"
This functionailty is super- but what is the risk?
I have been using this multi-dataset functionality for a couple years and it has always worked as it was suposed to.
I would say the only drawback or risk factor involved is the inherent complexity of enlarging your dataset with more tables and measures by adding more datasets to it: You have to be very consistent on which dimensions/tables of which particular sub-dataset in your "hub" dataset are used to filter the whole "hub" dataset. For example, you might find you have multiple date tables (one per subdataset) and any of them could play the role of filtering the other date tables, but you only want to stablish this role on one dates table (always the same), and I think its best to always define single oriented filter/directionality (even if that forces you to switch from 1to1 relationship to 1 to many, Im not sure you can have single oriented directionality on on 1 to 1 relationships) with the other dates tables so that at all times the filter flows from your dimensional/ruling dates table to the others and not the other way round, which led to some trouble in some occasions.