Hive Installation

Pre-requistes: Hadoop should have been installed on your system. Also, check if hadoop is in the path and HADOOP_HOME environment variable is set.

Hive requires a metastore for storing metadata information. This installation of hive will use embedded derby database.

Download hive from apache website. I have downloaded hive-0.8.0-bin.tar.gz. 

Explode it /opt folder.

Create tmp folder in hdfs and set permissions (if one does not already exists):

hadoop fs -mkdir  /tmp

hadoop fs -chmod g+w   /tmp

Create a /user/hive/warehouse folder and set permission as follows:

hadoop fs -mkdir /user/hive/warehouse

hadoop fs -chmod g+w /user/hive/warehouse

To start hive, go to /opt/hive-0.8.0-bin/bin directory and run ./hive. This will show a hive prompt.

Here are some simple hive operations to verify if your instance of Hive is working correctly :

create table customer(id INT, name STRING, address STRING);

-> This creates a table by name customer with id, name and address fields.

show tables;

-> This lists all the tables that were created.

alter table customer add columns(zipcode INT);

--> This adds a column zipcode to the customer table.

describe customer;

--> shows details about customer table including fields and their types.