Hadoop – 100x Faster. How we did it

It's much easier to make the first version 100 times slower and then comment the offending line in the new version :-)

That phrase "in-memory computing" always gets under my skin. What other sorts of computing do you generally do?