DAX CrossJoin between heavy facts

Question

Dear Gurus,
I am using SSAS Tabular and creating SSRS reports on top of it. My datasets are pure DAX code, no MDX.
My issue is that I am unable to find out how to make my query fast when the dataset involves many facts.
Currently I am using the following query:
EVALUATE
ADDCOLUMNS(
FILTER(
SUMMARIZE(
CROSSJOIN(
FactFallRegistration,
FactPermissions
, Resident[Firstname]
, Resident[Lastname]
, Resident[NationalNumber]
, Room[RoomDescription]
, Room[RoomTypeEN]
, CareProfile[CareProfileCode]
), --Summarize
FactPermissions[# Active Permissions] > 0
)--Filter
, "ResidentFirstname", Resident[Firstname]
, "ResidentLastName", Resident[Lastname]
, "ResidentNationalNumber", Resident[NationalNumber]
, "ResidentRoomDescription", Room[RoomDescription]
, "ResidentRoomTypeEN", Room[RoomTypeEN]
, "RizivCareProfile", CareProfile[CareProfileCode]
, "# Falls", FactFallRegistration[# Falls]
)--ADDCOLUMNS
The following model represents my cube:
The starting point is the Fact Falls, all attributes I need to read are linked to the factFalls except DimCareProfile, and in order to get a care profile a measure called
# Active Permissions must be positive for one resident on one day.
Currently I am using a plain crossjoin embedded in a Filter that excludes the null/0 permissions; factPermissions is significant in term of size, hence the query as-is (without SSRS filters) takes 2:30minutes to execute.
How can I tune this query to propagate the context from falls to the careprofile without a crossjoin ?
What is the querying best practice when multiple facts are to be used in the same query (2 facts or more)?
Thanks so much for any insights
Miloud

Answer 1

Marco thanks for the tip !
But I am a bit surprised because the query still takes a very long time :-/
I don't get it, if I am at a certain row in Fact Fall, if then I want to get the CareProfile, going through the Fact Permissions should be ultra fast since the link (Resident + Date) is known...
Or is my query construct non accurate ?
Marco moreover, let's imagine the starting point is Fact Falls, how could you still get the value of the CareProfile ? I have troubles to understand how can I navigate from Fact Falls to RizivCareProfile
 Edit:
I made some progress, now I have a super fast query. The idea i followed is as below:
EVALUATE
CALCULATETABLE(
FILTER(
ADDCOLUMNS(
SUMMARIZE(
GENERATE(
Filter(
FactFallRegistration,
[# Falls Permitted Long Stay] = 1
, CareProfile
, Resident[Firstname]
, Resident[Lastname]
, Resident[NationalNumber]
, OperationalUnit[OperationalUnitNumber]
, 'Date'[BK_Calendar] , CareProfile[CareProfileCode]
, "CareProfile", CareProfile[CareProfileCode]
) -- summarize
, "# Falls", [# Falls Permitted Long Stay]
,[# Falls Permitted Long Stay] > 0
, Resident[NationalNumber] = "XXX"
ORDER BY
'Date'[BK_Calendar]
I embedded the many to many relationship in a measure to have a cleaner approach.
Now this data set gives me something like:
Now what I am unable to achieve is to get something like:
Indeed, I need to group on resident operational unit...etc then give the total number of falls and the last value of CareProfile (in the 1st table above it's C)
I tried using the LastNonBlank but no luck, how can I do a sort of tail or whatever DAX oriented call to get the last version of the careprofile ?
Thanks very much