[ https://issues.apache.org/jira/browse/SPARK-4644?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14229512#comment-14229512 ]
Shixiong Zhu commented on SPARK-4644: ------------------------------------- I don't think solving things like `groupByKey` is valuable. If people want to cache values of a key by themselves in `groupByKey`, our optimization of `groupByKey` is useless, OOM happens in the user side. If they don't, they can always use `reduceByKey` to solve their problems. `join` is different from `groupByKey` because people have no alternative solution. {quote} we could provide an interface similar to ExternalAppendOnlyMap but which returns an Iterator[(K, Iterable[V])] pairs {quote} If I understand correctly, the iterator should be {noformat}Iterator[(K, Iterable[LEFT], Iterable[RIGHT])]{noformat}. It should collect the values of the same key from both LEFT and RIGHT. > Implement skewed join > --------------------- > > Key: SPARK-4644 > URL: https://issues.apache.org/jira/browse/SPARK-4644 > Project: Spark > Issue Type: Improvement > Components: Spark Core > Reporter: Shixiong Zhu > Attachments: Skewed Join Design Doc.pdf > > > Skewed data is not rare. For example, a book recommendation site may have > several books which are liked by most of the users. Running ALS on such > skewed data will raise a OutOfMemory error, if some book has too many users > which cannot be fit into memory. To solve it, we propose a skewed join > implementation. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org