On Tue, Jan 27, 2009 at 8:44 AM, Russel Winder <russel.win...@concertant.com > wrote:
> > There is already an Apache project to "avoid shared memory to make > > application that exploit parallelism easier". That is Hadoop > (especially > > with Pig). > > Certainly Hadoop has a place -- it implements a scatter/gather > framework. However that is just one architectural tool, not all > problems are amenable to a straightforward scatter/gather. > Amen to that!