It does, which one can argue it’s an implementation detail. The differences I mentioned above eg sharding, transaction boundary, secondary indices, etc. should

It does, which one can argue it’s an implementation detail. The differences I mentioned above eg sharding, transaction boundary, secondary indices, etc. should be generally applicable to a non-relational db and not unique to TAO. I could be wrong though; as I know little about other graph dbs. It would probably be more productive to compare RDB and Graph with certain workload examples (OLTP, OLAP, joins, scans, etc.).