Share your thoughts on DirectQuery for Power BI datasets and Azure Analysis Services (preview)

Hello,
We have build some Datasets with the Fact tables as import and the Dimensions as Direct Query from another PBI Dataset (DQ over AS).
The performance as very bad when we build data visualisation.
We would like to understand the performance of DQ over AS and we do not find a lot of documentation. Can we assume that the performance should be is equal to DQ "classic" or equal to Import (as it is also a pbi table).
For now it is more equal to a classic DQ and therefore bad performance for dimensions with more than a few hundreds of lines..
I am wondering if it will be possible when we use DQ over AS to have the option to get a "local" copy of meta data *AND* also data in the model to get better perfiormance .
something similar that broadcasting a table in databricks for exemple.
That would really be awesome for performance as the data of the table will be part of the model and the copy/synchronisation will be handle by power bi. In our hand we need the dimension to be updated only in the remote dataset.

Any thoughs on this ?
Any links for performance for DQ over AS will be great