Preserves the order in which actions are defined in the flow.
Preserves the order in which actions are defined in the flow.
same as input, no modifications
Default strategy, at the moment it will not do anything.
Preserves the order of the actions in which they are defined, but at first will give preference to loaders.
Preserves the order of the actions in which they are defined, but at first will give preference to loaders. If there are no loaders, keeps the order.
With Spark, waimak writers would usually force execution of the DAG and will produce outputs, while other waimak actions could be preparing the steps of the DAG.
With Spark, waimak writers would usually force execution of the DAG and will produce outputs, while other waimak actions could be preparing the steps of the DAG. In order to load the Spark Executors with work this strategy: 1) will first choose only writers, as those are most likely to execute the DAG 2) if there were no writers, it will choose only actions with inputs, as those will to DAG execution faster 3) if there were no writers and actions with inputs leaves them as they are
In order to race to actions that execute Spark DAG faster, it is needed to schedule certain actions earlier, regardless in which order they are defined.
In order to race to actions that execute Spark DAG faster, it is needed to schedule certain actions earlier, regardless in which order they are defined. This function will first apply rules from raceToOutputs and than will sort the selected actions in the order of labels defined by the 'orderedLabels' argument.
Defines various priority strategies for DataFlowExecutor to use.
Created by Alexei Perelighin on 24/08/2018.