[KYLIN-3490] For single column queries, only dictionaries are enough

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: v2.5.0
• Assignee: Zhong Yanghong
• Reporter: Zhong Yanghong
• Created: 10/Aug/18 08:58
• Updated: 17/Sep/18 00:59
• Resolved: 20/Aug/18 10:49

Description

A common use case for BI tools is as follows:

1. Firstly, extract all of the values of a dimension column
2. Then, select part of the values as filter condition.

Previously query for the first step requires to hit all of the segments' cuboid data, which may not be efficient, especially when the segments occupy many regions.

To use dictionary rather than cuboid data to answer this kind of queries, will reduce the cost of many rpcs to hbase.

Sample queries are as follows:

select A
from T
group by A

select distinct A
from T

select max(A)
from T

Issue Links

• relates to: KYLIN-3489 Improve the efficiency of enumerating dictionary values