So the big wins are when only a couple fields need to be read and the majority of the data can be ignored. They've even integrated this into spark. It'd be really nice to see the code released!
I'm surprised the conference accepted the paper without the source code being made public.