> Our approach has clear semantics, i.e. our extended SQL returns subsets of all tables with only those tuples that would be part of the traditional (single-tab

> Our approach has clear semantics, i.e. our extended SQL returns subsets of all tables with only those tuples that would be part of the traditional (single-table) query result set, however without performing any denormalization through joins. Yes, that sounds just like a GraphQL connection result with associated nested connections. I believe that there's link deduplication for GraphQL which can also be enabled to avoid the copies. The main difference is avoiding multiple round-trips--one for each table--which within a datacenter network doesn't add a whole lot of latency compared to the time serving those queries and some can also be done in parallel.