In cDOT, you must use svm peer relationships to replicate between svms and cluster peer relationships to replicate between clusters. You should set up node-based intercluster LIFs for replication between clusters. For data staying local (intra-cluster, svm to svm on the same cluster) it will be able to use the backend cluster network.
All replication on cDOT is at the VOLUME level.(NOT qtree, not at all!) qtrees in cDOT really allow for quotas and that's about it. all backup functions are flexvol based