We have several Power BI Report Server based reports/dashboards deployed that connect to SSAS data models. Most models until now have been reasonably sized model with fact tables that range from say 100K to about 5M records so performance has not been an issue.
However, we now have a model that has 40M records and growing, with somewhat complicated measures as well, so performance is starting to become a concern i.e. on say filtering or drilling down, the visuls could take 15-20 secs to refresh as opposed to say 2-5 secs. What we've also noticed is that the very first time a user runs a report it takes quite a while but the 2nd time it will be much better. This however applies to every new query getting generated upon using say a new set of filters or something.
So I have 2 questions:-
1. Does SSAS have some kind of caching concept where once a new query is run for the first time it will kind of keep it cached so when it is run again it will be faster?
2. If above is true, will the "query" differ for each user if row level security is in place e.g. Let's say user A runs a report and then user B runs the same report with same UI fitlers. User B however is not benefitted by the caching of User A's query because User B's query is slightly different given there is an additional row level security (where clause of sorts) being applied making the query slightly different?
