HI All,
I had a question about creating non-native tables using HCat. The idea is to
have a storage driver for hbase
and use the "STORED BY" token to create the table. The storage driver would
contain a storage handler which
would use HCatInputFormat and HCatOutputFormat respectively.
The cli command would be like the following:
hcat -e "CREATE TABLE hbase_table(key int, value string)
STORED BY 'org.apache.hadoop.hcatalog.hbase.HBaseStorageHandler'
WITH SERDEPROPERTIES ('hbase.columns.mapping' = ':key,cf1:val')
TBLPROPERTIES ('hbase.table.name' = 'xyz');"
Currently, Hive also has a storage handler for HBase which it uses to create
HBase tables,
but it cannot be used with HCat as it uses it own custom input and output
formats.
The question is going forward, if Hive users want to perform DDl operations
using HCatalog for
non-native tables like HBase, then should they be creating a new storage driver
?
Thanks
Vandana Ayyalasomayajula
