Cognite FDM 2 (on behalf of Celanese)

Hi @Jay Vansickle thanks for the inputs, 

We understand that the aggregation API is quite limited - partially this is influenced by the fact that aggregations potentially have to scan all instances of a type to compute a result, and graph traversal operations which are feasible when you consider a small part of the graph are too expensive when you need to scan all instances. Some limitations may be lifted in the future, such as more sophisticated date grouping/filtering. For many of the use cases where aggregations on large data volumes are appropriate, Industrial Log Analytics may be a better option. We are already testing this service on some Celanese data, though certainly not yet production grade. 

Best Regards

Elka