I had two synthetic keys in my data model and I hav removed same. But after removing I tried to compare execution time for both dashboards with and without synthetic keys. But I don't find any difference. Could you please let me how should I measure the performance change without synthetic key?
A synthetic key must not be wrong from a datamodel point of view and could be technically the same as if you used composite keys to resolve the synthetic keys and then you couldn't of course measure any differences. To have synthetic keys is more a kind of a bad habit and make it's difficult to read and understand the datamodel if you have many of them - so I suggest to remove them but there are of course other opnions, too - see:
