Earn the coveted Fabric Analytics Engineer certification. 100% off your exam for a limited time only!
Hi!
I'm testing the composite model for a business issue for a customer of ours.
Situation:
DQ:
1 Fact table with 174 rows
In memory:
dimension Debtor: 7750 rows
dimension Part: 60000 rows
the relations are made with integer keys. Both are Many to One (Single).The measure is a simple sum of a amount collumn.
Everyting out of the Part dimension i throw against this measure is fast and snappy! But when is use the Debtor dimension (much smaller) it sometimes takes more then 20 seconds for a simpe sort. I can't get my head around this, why is de larger dimension much faster then the smaller one?
Anyone any idea?
Hi @WVonk,
If I understand your sceanrio correctly that the dimension Part runs fast than the dimension Debtor?
If it is, I'm afraid that the performance will be affected by many factors. If you have more columns of dimension Debtor then the dimension Part, the performance of dimension Debtor will be bad than the dimension Part.
In addtion, the complex of the measure will also affect the performance.
Best Regards,
Cherry
Hi @v-piga-msft
Thats correct, the Part dimension is always much faster than the debtor. Even when i only load 1 collumn from the Debtor dimension.
I found out that when switch all three tables to DirectQeury, it's all fast. When i switch the Fact to DQ everything with the Debtor dimension is slow.
In Dax studio, a simple sum measure. With the PartID around 60 milliiseconds and with the Debtor ID around 800 ms tot 33 seconds.
Is there anything i can futher investigate regarding this?
User | Count |
---|---|
127 | |
108 | |
99 | |
63 | |
62 |
User | Count |
---|---|
136 | |
113 | |
102 | |
71 | |
60 |