My dear friend Mario has finally got around to blogging and one of his latest posts shows, it is possible to run Hadoop (which if you are not familiar with, can be thought of as an open source version of Google’s MapReduce) in Azure. You need to setup a typical configuration of nodes (Name Nodes, Tracker and Slaves).
Sure, there are a number of dependencies some you would expect, others not (e.g. Cygwin – cringe!; but hopefully that will go away with Hadoop 0.22). I wonder what overheard the Cygwin runtime has?
It would be interesting to know if someone is (or planning) to use this at work.