Can develop MapReduce program by MATLAB run effectively with Hadoop?

Hadoop by default support Java, C++, and python. However, I developed an algorithm using Matlab to deal with normal size data, later I decide to scale my algorithm using MapReduce to deal with big data.

I want to continue my work in Matlab, but I don't know if Matlab fully and smoothly supports Hadoop or not, therefore I want to hear advice from experts before it's too late.

Answer 1

[Kojiro Saito]

MATLAB does support MapReduce on Hadoop with MATLAB Compiler OR Parallel Computing Toolbox / MATLAB Distributed Computing Server.

(1) MATLAB Compiler

With MATLAB Compiler, you can create compiled MapReduce applications. This is a compiled application, so you can execute it with MATLAB Runtime.

For detail, please see this doc: Compiled MapReduce Applications on Hadoop Clusters

(2) Parallel Computing Toolbox / MATLAB Distributed Computing Server With these parallel products, you can scale up your mapreduce algorithm to Hadoop.

Before executing mapreduce, it's needed to configure mapreducer environment, for example,

setenv('HADOOP_HOME', '/path/to/hadoop_home/');
cluster = parallel.cluster.Hadoop;
mr = mapreducer(cluster);

For detail, please see this doc: Run mapreduce on a Hadoop Cluster