ShuffledJoin Physical Operators¶

ShuffledJoin is an extension of the JoinCodegenSupport abstraction for join operators that shuffle two child relations using some join keys.

Contract¶

isSkewJoin Flag¶

isSkewJoin: Boolean

Used when:

• ShuffledJoin is requested for node name and requiredChildDistribution

Implementations¶

• ShuffledHashJoinExec
• SortMergeJoinExec

Node Name¶

nodeName: String

nodeName adds (skew=true) suffix to the default node name when the isSkewJoin flag is enabled.

nodeName is part of the TreeNode abstraction.

Required Child Output Distribution¶

requiredChildDistribution: Seq[Distribution]

requiredChildDistribution are HashClusteredDistributions for the left and right keys.

requiredChildDistribution is part of the SparkPlan abstraction.