[IMPALA-10412] ConvertToCNFRule can be apply to view table

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: Impala 4.0.0
• Fix Version/s: Impala 4.0.0
• Component/s: Frontend
• Epic Color: ghx-label-12
• Assignee: gaoxiaoqing
• Reporter: gaoxiaoqing
• Created: 29/Dec/20 09:23
• Updated: 07/Jan/21 17:32
• Resolved: 07/Jan/21 17:32

Description

ConvertToCNFRule can be apply to view table, if the predicate associated more than one tuples and the predicate can be pushed down to scan node.

e.g.

CREATE VIEW rawdata.test_cnf AS
SELECT
-- +straight_join
 events.*, users.p_channel p_channel_profile FROM rawdata.event_ros_p7 events LEFT OUTER JOIN
-- +BROADCAST
 rawdata.profile_wos_p7 users ON events.user_id = users.id

 

select event_id, p_query from test_cnf where p_query='a' or (p_package_id='b' and p_channel_profile='c');

 

the optimized plan

PLAN-ROOT SINK
|
04:EXCHANGE [UNPARTITIONED]
|
02:HASH JOIN [LEFT OUTER JOIN, BROADCAST]
|  hash predicates: events.user_id = users.id
|  other predicates: events.p_query = 'a' OR users.p_channel = 'c'
|  row-size=79B cardinality=6.31M
|
|--03:EXCHANGE [BROADCAST]
|  |
|  01:SCAN KUDU [rawdata.profile_wos_p7 users]
|     row-size=35B cardinality=151.23M
|
00:SCAN HDFS [rawdata.event_ros_p7 events]
   partitions=21/23 files=1410 size=10.99GB
   predicates: events.p_query = 'a' OR events.p_package_id = 'b'
   row-size=44B cardinality=6.31M