AdaptiveExecutionContext¶

AdaptiveExecutionContext is the execution context to share the following between the main query and all subqueries:

• SparkSession
• QueryExecution
• Subquery Cache
• Stage Cache

Creating Instance¶

AdaptiveExecutionContext takes the following to be created:

• SparkSession
• QueryExecution

AdaptiveExecutionContext is created when:

• QueryExecution is requested for the physical preparations rules (and creates an InsertAdaptiveSparkPlan)

Subquery Cache¶

subqueryCache: TrieMap[SparkPlan, BaseSubqueryExec]

AdaptiveExecutionContext creates a TrieMap (Scala) of SparkPlans (canonicalized ExecSubqueryExpressions to be precise) and associated BaseSubqueryExecs.

subqueryCache is used when:

• AdaptiveSparkPlanExec physical operator is requested for the adaptive optimizations (and creates a ReuseAdaptiveSubquery)

Stage Cache¶

stageCache: TrieMap[SparkPlan, QueryStageExec]

AdaptiveExecutionContext creates a TrieMap (Scala) of SparkPlans (canonicalized Exchanges to be precise) and associated QueryStageExecs.

stageCache is used when:

• AdaptiveSparkPlanExec physical operator is requested to createQueryStages