I think they have different use cases: fdb when you want transactions, cassandra when you want throughput.