But even if we did it in this naive way, the total execution time is still a very reasonable (48 3 ms) + 260 ms = 404 ms*

I can't read things like this and avoid wondering why people keep reading him. He is so wrong so often I ... it's just beyond me.

Query execution time is just one part of performance. You still have to create the connection and transfer all the data across the wire. Sending N records in 1 big package is much more efficient than sending N+1 records in individual packages.

Run a performance measure on the actual code execution and I am almost positive you'll get worse performance than would be indicated by your query execution times alone.

From now on, I'm flagging every codinghorror post.