private JobGraph createJobGraphV1(
String pointsPath, String centersPath, String resultPath, int numSubTasks)
throws JobGraphDefinitionException {
// -- init
// -------------------------------------------------------------------------------------------------
final TypeSerializerFactory<?> serializer = RecordSerializerFactory.get();
JobGraph jobGraph = new JobGraph("Distance Builder");
// -- vertices
// ---------------------------------------------------------------------------------------------
JobInputVertex points = createPointsInput(jobGraph, pointsPath, numSubTasks, serializer);
JobInputVertex models = createModelsInput(jobGraph, centersPath, numSubTasks, serializer);
JobTaskVertex mapper = createMapper(jobGraph, numSubTasks, serializer);
JobOutputVertex output = createOutput(jobGraph, resultPath, numSubTasks, serializer);
// -- edges
// ------------------------------------------------------------------------------------------------
JobGraphUtils.connect(points, mapper, ChannelType.NETWORK, DistributionPattern.POINTWISE);
JobGraphUtils.connect(models, mapper, ChannelType.NETWORK, DistributionPattern.BIPARTITE);
JobGraphUtils.connect(mapper, output, ChannelType.NETWORK, DistributionPattern.POINTWISE);
// -- instance sharing
// -------------------------------------------------------------------------------------
points.setVertexToShareInstancesWith(output);
models.setVertexToShareInstancesWith(output);
mapper.setVertexToShareInstancesWith(output);
return jobGraph;
}
private static JobGraph createJobGraph(
String pointsPath, String centersPath, String resultPath, int numSubTasks, int numIterations)
throws JobGraphDefinitionException {
// -- init
// -------------------------------------------------------------------------------------------------
final TypeSerializerFactory<?> serializer = RecordSerializerFactory.get();
@SuppressWarnings("unchecked")
final TypeComparatorFactory<?> int0Comparator =
new RecordComparatorFactory(new int[] {0}, new Class[] {IntValue.class});
JobGraph jobGraph = new JobGraph("KMeans Iterative");
// -- vertices
// ---------------------------------------------------------------------------------------------
JobInputVertex points = createPointsInput(jobGraph, pointsPath, numSubTasks, serializer);
JobInputVertex centers = createCentersInput(jobGraph, centersPath, numSubTasks, serializer);
JobTaskVertex head = createIterationHead(jobGraph, numSubTasks, serializer);
JobTaskVertex mapper =
createMapper(jobGraph, numSubTasks, serializer, serializer, serializer, int0Comparator);
JobTaskVertex reducer =
createReducer(jobGraph, numSubTasks, serializer, int0Comparator, serializer);
JobOutputVertex fakeTailOutput =
JobGraphUtils.createFakeOutput(jobGraph, "FakeTailOutput", numSubTasks, numSubTasks);
JobOutputVertex sync = createSync(jobGraph, numIterations, numSubTasks);
JobOutputVertex output = createOutput(jobGraph, resultPath, numSubTasks, serializer);
// -- edges
// ------------------------------------------------------------------------------------------------
JobGraphUtils.connect(points, mapper, ChannelType.NETWORK, DistributionPattern.POINTWISE);
JobGraphUtils.connect(centers, head, ChannelType.NETWORK, DistributionPattern.POINTWISE);
JobGraphUtils.connect(head, mapper, ChannelType.NETWORK, DistributionPattern.BIPARTITE);
new TaskConfig(mapper.getConfiguration())
.setBroadcastGateIterativeWithNumberOfEventsUntilInterrupt(0, numSubTasks);
new TaskConfig(mapper.getConfiguration()).setInputCached(0, true);
new TaskConfig(mapper.getConfiguration())
.setInputMaterializationMemory(0, MEMORY_PER_CONSUMER * JobGraphUtils.MEGABYTE);
JobGraphUtils.connect(mapper, reducer, ChannelType.NETWORK, DistributionPattern.BIPARTITE);
new TaskConfig(reducer.getConfiguration())
.setGateIterativeWithNumberOfEventsUntilInterrupt(0, numSubTasks);
JobGraphUtils.connect(
reducer, fakeTailOutput, ChannelType.NETWORK, DistributionPattern.POINTWISE);
JobGraphUtils.connect(head, output, ChannelType.NETWORK, DistributionPattern.POINTWISE);
JobGraphUtils.connect(head, sync, ChannelType.NETWORK, DistributionPattern.BIPARTITE);
// -- instance sharing
// -------------------------------------------------------------------------------------
points.setVertexToShareInstancesWith(output);
centers.setVertexToShareInstancesWith(output);
head.setVertexToShareInstancesWith(output);
mapper.setVertexToShareInstancesWith(output);
reducer.setVertexToShareInstancesWith(output);
fakeTailOutput.setVertexToShareInstancesWith(output);
sync.setVertexToShareInstancesWith(output);
return jobGraph;
}