[HIVE-7207] support partial rowkey scan in HBase filter pushdown

• Type: Improvement
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Labels: Hbase HbaseStorageHandler
• Assignee: Unassigned
• Reporter: Ning Zhang
• Created: 10/Jun/14 17:25
• Updated: 10/Jun/14 17:28

Description

One of our Hive tables is backed up by Hbase (HBaseStorageHandler), to simulate the partitioned Hive Table by "DataDate", we use composite rowkey in Hbase, e.g. DataDate_Userid_Actionid_Timestamp. The example rowkey is as follow.

rowkey:
20140601_784353454593233274_20123282_1401632522132
20140601_784353454_20123282_1401632522132
20140601_784470763593179377_20485247_1401632520825
20140601_784470763593233227_20485222_1401632520821

However, it seems Hive does not support "partial rowkey scan". For example I want to get all data that were generated on 06/01/2014, so I issue the following Hive query, but Hive returns nothing.

select * from table where DataDate="20140601";

After several attempts, I found that I have to give exact row key (e.g. 20140601_784353454_20123282_1401632522132) so that Hive can find that record.

The reason I want to see the "partial rowkey scan" feature is because: in Hbase, partial table scan should have better performance than full table scan.

Is there any plan in Hive community to support "partial rowkey scan" in near future?

Issue Links

• relates to: HIVE-1643 support range scans and non-key columns in HBase filter pushdown
• relates to: HIVE-1226 support filter pushdown against non-native tables