Hadoop like solutions allow parallelism => i.e. speed of processing depends on number of nodes you are able to allocate for your task.
Key idea of the analyzing is based on two functions: Map & Reduce, defined by developer basing on business task criteria.
Each stage has key-value pair as input and output, types of those are defined by developer.
Function is usually written in java, but there is a number of solutions that allow using of other languages.

I would recommend to review Google Code University materials on the link below prior to starting the Hadoop studying - to become more familiar with concept and techniques.

Links:

• http://code.google.com/edu/parallel/mapreduce-tutorial.html

posted 2 months ago