[ https://issues.apache.org/jira/browse/HIVE-15580?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15836414#comment-15836414 ]
Xuefu Zhang commented on HIVE-15580: ------------------------------------ Hi [~Ferd] and [~dapengsun], I'm wondering if you guys could help measure the performance impact of the patch here? We at Uber don't have a dedicated environment, so getting accurate measurement is challenging. It would be great if you guys can help. Based on the result, we may have some followup work to do. Thanks. > Eliminate unbounded memory usage for orderBy and groupBy in Hive on Spark > ------------------------------------------------------------------------- > > Key: HIVE-15580 > URL: https://issues.apache.org/jira/browse/HIVE-15580 > Project: Hive > Issue Type: Improvement > Components: Spark > Reporter: Xuefu Zhang > Assignee: Xuefu Zhang > Fix For: 2.2.0 > > Attachments: HIVE-15580.1.patch, HIVE-15580.1.patch, > HIVE-15580.2.patch, HIVE-15580.2.patch, HIVE-15580.3.patch, > HIVE-15580.4.patch, HIVE-15580.5.patch, HIVE-15580.patch > > > Currently, orderBy (sortBy) and groupBy in Hive on Spark uses unbounded > memory. For orderBy, Hive accumulates key groups using ArrayList (described > in HIVE-15527). For groupBy, Hive currently uses Spark's groupByKey operator, > which has a shortcoming of not being able to spill to disk within a key > group. Thus, for large key group, memory usage is also unbounded. > It's likely that this will impact performance. We will profile and optimize > afterwards. We could also make this change configurable. -- This message was sent by Atlassian JIRA (v6.3.4#6332)