Hadoop??streaming

[zhxfl]

Hello??we know Hadoop-streaming is use for Hadoop to run native program. 
Hadoop-streaming supports  Map and Reduce logic. Reduce logic means Hadoop 
collect all values with same key and give the stream for the native 
application.Spark has PipeRDD too, but PipeRDD doesn't support Reduce logic. So 
it's difficulty for us to transplant our application from Hadoop to Spark.
Anyone can give me advise, thanks!