hive for processing and hbase for low latency read

creating table which can be accessed both by hive and hbase, this is done in cases where we require quick (low latency) searches and faster processing of data. In this case we can use hive for processing and hbase for quick searches with same underlying table.

steps:

1. create table in hive

   create table integration (a int, b string);
   

2. load data in hive table

   load data local inpath '/path/to/data/data.csv'
   overwrite into table integration;
   

3. create hive-hbase table

   create table hive_table(key int, value string) stored by
   'org.apache.hadoop.hive.hbase.HBaseStorageHandler' with
   SERDEPROPERTIES ("hbase.table.name"="hbase_table");
   

4. insert data in hive-hbase table

   insert overwrite table hive_table select * from integration;
   

5. verify data from hive

   select * from hive_table;
   

6. verify data from hbase

   scan 'hbase_table'
   

Note: in hive we can have duplicate keys but in hbase we can have only unique keys so when we load data in the table which is hive-hbase then it updates the same key with new timestamp and by default latest one will be shown.