Filter on child types

I see, so generally if the dataset is in the millions, DQL would be a better fit than graphQL?