Jonah,
> Hmm, how did you come to that conclusion? From every definition of
> MPP I can find, GridSQL meets the requirements.
>
> GridSQL:
> - Acts as a single, large-scale system by means of shared-nothing
> clustering (as does Greenplum IIRC)
> - Partitions and executes independent units-of-work in parallel across
> multiple independent microprocessors
> - Is capable of scaling to hundreds (if not thousands) of nodes
>
> What definition are you using, because I can't seem to find it in my top 5?
I was looking at the inability to do unrestricted joins across nodes,
but I realize that there are other MPP systems that have the same
restriction. So I guess that's too fine a restriction.
What's the largest GridSQL cluster anyone's tested, so far? Could we
look at using it for TPC-H?
--Josh