If you have a Pseudo Distributed mode setup and you are able to run your program in that mode, you can be assured that the scaling in a multi node environment will happen automatically.

That is the beauty and commitment of hadoop, your can achieve greater throughput and performance by just adding nodes to your cluster. Your programming paradigm does not change if there are more or lesser number of nodes.

If you have to still look at it, amazon has a feature called Elastic Map Reduce (EMR) that allows you to quickly setup your cluster of hadoop/mapreduce systems.


Sarath.

Ok, let me rephrase my question?

Is there a document or sample to setup in multiple nodes? I see most of the samples in the LMS are single node and document related to mad reduce.
I am looking for step by step approach to configure multiple node cluster and to run the map reduce.

Keerthi

Hi Keerthi,

Here is the weblink for hadoop's fully-distributed mode deployment[multi-node cluster]
http://brokrishrocking.blogspot.in/2012/11/hadoop-installation-in-fully.html

Further references:
1. Tom White's "Hadoop : The Definitive Guide, Chapter#9" gives a complete step-by-step approach to deploy hadoop cluster in fully-distributed mode[multi-node cluster].
2. Other popular book is "Hadoop Operations and cluster management" which is specially designed for hadoop administrators.