One more post on MapReduce and parallel SQL, this time for the folks at O’Reilly Radar.
Just for the record, I think MapReduce is fine, but not especially interesting technology. The thing is, the “teachable moment” it presents is really great stuff, because it is bringing people toward data-centric parallel programming. So it’s good for the data-centric research business in general, and especially for data-centric approaches to parallelism.
I.e. chum in the water for our research on Lincoln…
