spark技术分享SparkStrategies — Container of Execution Planning Strategies
SparkStrategies is an abstract Catalyst query planner that merely serves as a “container” (or a namespace) of the concrete execution planning strategies (for SparkPlanner):
-
FlatMapGroupsWithStateStrategy -
SpecialLimits -
StatefulAggregationStrategy -
StreamingDeduplicationStrategy -
StreamingRelationStrategy
SparkStrategies has a single lazily-instantiated singleRowRdd value that is an RDD of internal binary rows that BasicOperators execution planning strategy uses when resolving OneRowRelation (to RDDScanExec leaf physical operator).
|
Note
|
OneRowRelation logical operator represents SQL’s SELECT clause without FROM clause or EXPLAIN DESCRIBE TABLE.
|
