Hadoop MapReduce over multiple inputs

Question

I would like to make use of multiple input formats in a single Job. I have used org.apache.hadoop.mapreduce.lib.input.MultipleInputs however this utility seems to only be designed for inputs that exist on HDFS (have a Path).

Is there a way to use multiple input formats from disparate sources?

My specific need is as follows...

I would like to have a single job that performs a reduce side join from an existing elastic search index (utilizing the ESInputFormat provided by https://github.com/elasticsearch/elasticsearch-hadoop ) with a set of sequence files that contain information to be indexed. I would like to read from these multiple inputs merge to the reduce phase and insert into another index (with some additional logic) for later use.

Suggestions?

Answer 1

You can still use MultipleInputs and just pass in a non null Path. It doesn't need to point to a valid location to still work, it just can't be null.

This is ok I suppose.