How can I achieve the elasticsearch and shark(spark) Integration

[mingle]

I have used the elasticsearch-hadoop successfully, mapreduce run over elasticsearch instead of hdfs so it is very fast loading the appropriate data, it avoids scan the whole data using elasticsearch index.

 but as we know ,mapreduce is very slow,so I want to use the shark(spark) which is faster than the mapreduce!!!

but I have no idea what to do Integrating the spark and elasticsearch.can anyone help me?

thanks