Right, but that has nothing to do with connection pooling and sync. And there is no structural reason that makes easier to implemented prepared statements for PostgreSQL than for Cassandra. It's anecdotal evidence.