Forum Discussion
Advocate IMultidimensional or Tabular model?
JesperP Here is some food for thought, based on my opinion.
The Power BI Desktop, and all the modeling done in the tool is based on tabular. So it is a seemless transition between any development in the Desktop and what you may do in the model.
When using MD, Power BI is still using DAX to make calls to the MD cube, so there are some limitations and not all MD features may be supported.
Tabular (seems) to be where Microsoft is investing the vast majority of resources (based on the SQL 2016 release). It also has the potential to be much faster because it is (or can be) stored in memory, whereas MD is all on disk.
Another benefit (I'll be blogging on soon) is that you can directly connect your Desktop to the local in memory copy of the model when you are working on it. (looks like <modelname_YourName_GUID>) This is a huge development win for me, as I can test out my measures and calcs against the visuals right away prior to checking them in and deploying them.
If you are already well versed in DAX, my personal opinion is that you head down the tabular route.
Community ChampionJesperP i would suggest you go through this articles thoroughly to understand the difference based on your case whether you're going to use power bi service to connect or power bi desktop.
Multidimensional - https://powerbi.microsoft.com/en-us/documentation/powerbi-desktop-ssas-multidimensional/
Tabular - https://powerbi.microsoft.com/en-us/documentation/powerbi-desktop-analysis-services-tabular-data/
- JesperP
Thank you for your Reply, I also found those two links. Although they do provide valuable information, it stil isn't very clear about what Power BI features are supported. We found out by trying that, when using Multidimensional, some features were grayed-out.
We really want to create most of the logics/measures in our model itself, but it would be very convenient if we still have the ability to create some measures/calculated columns in the reports itself for instance.
I also don't really know what the main differences are of both models in General. I understood tabular could be slower when you have lots of data as opposed to Multidimensional.
Basically I want to know what would give us the most seamless-experience with Power BI. As we are still in an early stage, we can now build it however we want. Later, this could become an issue. :)JesperP Here is some food for thought, based on my opinion.
The Power BI Desktop, and all the modeling done in the tool is based on tabular. So it is a seemless transition between any development in the Desktop and what you may do in the model.
When using MD, Power BI is still using DAX to make calls to the MD cube, so there are some limitations and not all MD features may be supported.
Tabular (seems) to be where Microsoft is investing the vast majority of resources (based on the SQL 2016 release). It also has the potential to be much faster because it is (or can be) stored in memory, whereas MD is all on disk.
Another benefit (I'll be blogging on soon) is that you can directly connect your Desktop to the local in memory copy of the model when you are working on it. (looks like <modelname_YourName_GUID>) This is a huge development win for me, as I can test out my measures and calcs against the visuals right away prior to checking them in and deploying them.
If you are already well versed in DAX, my personal opinion is that you head down the tabular route.
- JesperP
Hi Enco1978,
Thank you for your reply. This is also what I assumed, I think tablular would give a more 'seamless' transition.
Also I think DAX works a bit easier than MD, but maybe thats just a matter of perspective.
Where can I find your blog? :smileyhappy:
