Package org.apache.flink.runtime.executiongraph

Class VertexInputInfoComputationUtils

java.lang.Object

org.apache.flink.runtime.executiongraph.VertexInputInfoComputationUtils

public class VertexInputInfoComputationUtils
extends Object

Util to compute JobVertexInputInfos for execution job vertex.

Method Summary

Modifier and Type	Method	Description
static JobVertexInputInfo	computeVertexInputInfoForAllToAll(int sourceCount, int targetCount, Function<Integer,Integer> numOfSubpartitionsRetriever, boolean isDynamicGraph, boolean isBroadcast, boolean isSingleSubpartitionContainsAllData)	Compute the JobVertexInputInfo for a DistributionPattern.ALL_TO_ALL edge.
static JobVertexInputInfo	computeVertexInputInfoForPointwise(int sourceCount, int targetCount, Function<Integer,Integer> numOfSubpartitionsRetriever, boolean isDynamicGraph)	Compute the JobVertexInputInfo for a DistributionPattern.POINTWISE edge.
static Map<IntermediateDataSetID,JobVertexInputInfo>	computeVertexInputInfos(int parallelism, List<? extends IntermediateResultInfo> inputs, boolean isDynamicGraph)
static Map<IntermediateDataSetID,JobVertexInputInfo>	computeVertexInputInfos(ExecutionJobVertex ejv, Function<IntermediateDataSetID,IntermediateResult> intermediateResultRetriever)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

computeVertexInputInfos

public static Map<IntermediateDataSetID,JobVertexInputInfo> computeVertexInputInfos(ExecutionJobVertex ejv,
 Function<IntermediateDataSetID,IntermediateResult> intermediateResultRetriever)
                                                                             throws JobException

Throws:

JobException

computeVertexInputInfos

public static Map<IntermediateDataSetID,JobVertexInputInfo> computeVertexInputInfos(int parallelism,
 List<? extends IntermediateResultInfo> inputs,
 boolean isDynamicGraph)

computeVertexInputInfoForPointwise

public static JobVertexInputInfo computeVertexInputInfoForPointwise(int sourceCount,
 int targetCount,
 Function<Integer,Integer> numOfSubpartitionsRetriever,
 boolean isDynamicGraph)

Compute the JobVertexInputInfo for a DistributionPattern.POINTWISE edge. This computation algorithm will evenly distribute subpartitions to downstream subtasks according to the number of subpartitions. Different downstream subtasks consume roughly the same number of subpartitions.

Parameters:

sourceCount - the parallelism of upstream

targetCount - the parallelism of downstream

numOfSubpartitionsRetriever - a retriever to get the number of subpartitions

isDynamicGraph - whether is dynamic graph

Returns:

the computed JobVertexInputInfo

computeVertexInputInfoForAllToAll

public static JobVertexInputInfo computeVertexInputInfoForAllToAll(int sourceCount,
 int targetCount,
 Function<Integer,Integer> numOfSubpartitionsRetriever,
 boolean isDynamicGraph,
 boolean isBroadcast,
 boolean isSingleSubpartitionContainsAllData)

Compute the JobVertexInputInfo for a DistributionPattern.ALL_TO_ALL edge. This computation algorithm will evenly distribute subpartitions to downstream subtasks according to the number of subpartitions. Different downstream subtasks consume roughly the same number of subpartitions.

Parameters:

sourceCount - the parallelism of upstream

targetCount - the parallelism of downstream

numOfSubpartitionsRetriever - a retriever to get the number of subpartitions

isDynamicGraph - whether is dynamic graph

isBroadcast - whether the edge is broadcast

isSingleSubpartitionContainsAllData - whether single subpartition contains all data

Returns:

the computed JobVertexInputInfo