Tuning Hive queries that uses underlying HBase Table

[Manjula mohapatra]

I am querying Hive table ( mapped to HBase Table ) .

What are the techniques to tune the Hive query and to avoid HBase scans.

Query uses multiple SPLIT and SUBSTR functions and WHERE  condition
something like

select  col1, col2, ...,count(*)
from hiveTable

where split( col1)[0] > timestamp1  and split( col1)[0]<timestamp2
group by ....