Does SQL Server Tabular Data Stream transfer all values of every row, even the values that don't change? I read the spec a while ago and it appears the answer is "yes."
It seems like that would be an extreme bandwidth optimization and make possible a lot more flexible queries: SQL could detect that I did a join of a parent and child table, returned repeated rows of the parent for each child, but only pass the parent "once":
SELECT p.*, c.* FROM PARENT p INNER JOIN CHILD c ON ....
--p.* values are only shipped once -> saves bandwidth
Currently, I do separate queries for parent and child, which can result in an expensive repeated WHERE clause being executed twice or necessitate saving parent ids in a table variable...
View Complete Post