1

On Hadoop Cluster, Where we need to install Hive ,on DataNode or Namenode? On which factor we need to decide the installation node ( Datanode ot Namenode)

Thanks !!!

AshHadoopExpert
  • 11
  • 3

1 Answers1

0

Installation of hive is independent of the fact that it resides on the namenode or the datanode. Hive configuration file needs to know where is hadoop installed so that it can access the job tracker.

Once it has the knowledge of where job tracker is running, whenever you execute a query in Hive, it will convert your query into one or more mapreduce program and then it will submit this program to hadoop's jobtracker. Jobtracker then executes this map reduce program and show/store the output.

gonephishing
  • 1,388
  • 3
  • 18
  • 45
  • Thanks a lot for your answer. Do we need to install hive on each datanode. What is the best practice. – AshHadoopExpert Jan 21 '16 at 01:34
  • No, the best will be to install hive on the node from where you are going to submit jobs https://stackoverflow.com/questions/8456141/in-a-hadoop-cluster-should-hive-be-installed-on-all-nodes – gonephishing Jan 21 '16 at 10:41