[ https://issues.apache.org/jira/browse/HIVE-7158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14030340#comment-14030340 ]
Lefty Leverenz commented on HIVE-7158: -------------------------------------- Does the design doc need guidance about this (or is it time to add Tez documentation to the user docs)? * [Hive on Tez | https://cwiki.apache.org/confluence/display/Hive/Hive+on+Tez] At a minimum, Configuration Properties needs to document these parameters: * new parameter: hive.tez.auto.reducer.parallelism * new parameter: hive.tez.max.partition.factor * new parameter: hive.tez.min.partition.factor * new default for [hive.exec.reducers.bytes.per.reducer | https://cwiki.apache.org/confluence/display/Hive/Configuration+Properties#ConfigurationProperties-hive.exec.reducers.bytes.per.reducer] (with version information) * new default for [hive.exec.reducers.max | https://cwiki.apache.org/confluence/display/Hive/Configuration+Properties#ConfigurationProperties-hive.exec.reducers.max] (with version information) > Use Tez auto-parallelism in Hive > -------------------------------- > > Key: HIVE-7158 > URL: https://issues.apache.org/jira/browse/HIVE-7158 > Project: Hive > Issue Type: Bug > Reporter: Gunther Hagleitner > Assignee: Gunther Hagleitner > Labels: TODOC14 > Fix For: 0.14.0 > > Attachments: HIVE-7158.1.patch, HIVE-7158.2.patch, HIVE-7158.3.patch, > HIVE-7158.4.patch, HIVE-7158.5.patch > > > Tez can optionally sample data from a fraction of the tasks of a vertex and > use that information to choose the number of downstream tasks for any given > scatter gather edge. > Hive estimates the count of reducers by looking at stats and estimates for > each operator in the operator pipeline leading up to the reducer. However, if > this estimate turns out to be too large, Tez can reign in the resources used > to compute the reducer. > It does so by combining partitions of the upstream vertex. It cannot, > however, add reducers at this stage. > I'm proposing to let users specify whether they want to use auto-parallelism > or not. If they do there will be scaling factors to determine max and min > reducers Tez can choose from. We will then partition by max reducers, letting > Tez sample and reign in the count up until the specified min. -- This message was sent by Atlassian JIRA (v6.2#6252)