HIve on tez or Hive on Spark?

Hello everybody,

I'm trying to play with Hive and Elastic. I've been able to deploy a basic Hive config and adding the elastic-hadoop-hive jars I can create an external table mapped on Hadoop. Now, I have some messages telling that using mr as execution engine will be deprecated. Then I understand that I should use tez or Spark.

Which one should I use? Tez seems a little bit complex to install (need to build from source using Maven), but people says that performances are better.

Does anyone have some opinion on this? What is the optimal setup for Hive on Elasticsearch?

Thanks a lot,


Elastic doesn't take an official standpoint on whether to use MR, Tez, or Spark for Hive's execution layer. We primarily test our Hive solution on a local distribution that is based on MR.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.