[ https://issues.apache.org/jira/browse/HAWQ-1597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16501160#comment-16501160 ]
Hongxu Ma edited comment on HAWQ-1597 at 6/5/18 2:05 AM: --------------------------------------------------------- Good job, Thanks. [~wlin] was (Author: hongxu ma): Thanks! [~wlin] > Implement Runtime Filter for Hash Join > -------------------------------------- > > Key: HAWQ-1597 > URL: https://issues.apache.org/jira/browse/HAWQ-1597 > Project: Apache HAWQ > Issue Type: New Feature > Components: Query Execution > Reporter: Lin Wen > Assignee: Lin Wen > Priority: Major > Fix For: 2.4.0.0-incubating > > Attachments: 111BA854-7318-46A7-8338-5F2993D60FA3.png, HAWQ Runtime > Filter Design.pdf, HAWQ Runtime Filter Design.pdf, q17_modified_hawq.gif > > > Bloom filter is a space-efficient probabilistic data structure invented in > 1970, which is used to test whether an element is a member of a set. > Nowdays, bloom filter is widely used in OLAP or data-intensive applications > to quickly filter data. It is usually implemented in OLAP systems for hash > join. The basic idea is, when hash join two tables, during the build phase, > build a bloomfilter information for the inner table, then push down this > bloomfilter information to the scan of the outer table, so that, less tuples > from the outer table will be returned to hash join node and joined with hash > table. It can greatly improment the hash join performance if the selectivity > is high. -- This message was sent by Atlassian JIRA (v7.6.3#76005)