To use ODI efficiently in a Data Warehouse project is it necessary to have a topology with an intermediate transformation server?

To use ODI efficiently in a Data Warehouse project is it necessary to have a topology with
an intermediate transformation server?

A.
It is mandatory to collect data directly from sources into a data warehouse server using
real time replication processes.

B.
Data transformation should be distributed across several scalable nodes.

C.
It is mandatory to have an intermediate server.

D.
It is better to load data directly from sources into a data warehouse server.