com.pervasive.datarush.operators.partition

Class PartitionHint

java.lang.Object

com.pervasive.datarush.operators.AbstractLogicalOperator

com.pervasive.datarush.operators.StreamingOperator

com.pervasive.datarush.operators.DeferredCompositeOperator

com.pervasive.datarush.operators.AbstractDeferredRecordOperator

com.pervasive.datarush.operators.partition.PartitionHint

All Implemented Interfaces:

LogicalOperator, PipelineOperator<RecordPort>, RecordPipelineOperator

public final class PartitionHint
extends AbstractDeferredRecordOperator

Forces the input data to be partitioned into parallel streams of data for subsequent parallel operations. If requested, the output will also be sorted. It is also possible to force data to be staged to disk, breaking physical streaming; consumers of the output will run in a physical graph executed after the physical graph executing this operator. Even if not forced, the (re)partitioning may cause staging to occur.

Normally these actions happens automatically in a logical graph as required, but this operator provides a mechanism for explicit control.

Field Summary

Fields inherited from class com.pervasive.datarush.operators.AbstractDeferredRecordOperator

input, output

Constructor Summary

Constructors

Constructor and Description
PartitionHint() Forces a partitioning into parallel streams of data.
PartitionHint(DataDistribution partitioning) Forces a partitioning into parallel streams of data guaranteeing the specified distribution.

Method Summary

Modifier and Type	Method and Description
protected void	compose(DeferredCompositionContext ctx) Compose the body of this operator.
protected void	computeMetadata(StreamingMetadataContext ctx) Implementations must adhere to all of the contracts specified by StreamingOperator.computeMetadata(com.pervasive.datarush.operators.StreamingMetadataContext).
DataOrdering	getDataOrdering() Gets the data ordering of the output.
DataDistribution	getPartitioning() Gets the data distribution of the output.
boolean	getPreserveSortOrder() If true, data order will be preserved ( provided that the data order is known ).
boolean	isForceStaging() Indicates whether data will be forcibly staged to disk.
void	setDataOrdering(DataOrdering dataOrdering) Sets the data ordering of the output.
void	setForceStaging(boolean forceStaging) Sets whether data must be staged to disk.
void	setPartitioning(DataDistribution partitioning) Sets the data distribution of the output.
void	setPreserveSortOrder(boolean preserveSortOrder) Sets whether data order will be preserved ( provided that the data order is known ).

Methods inherited from class com.pervasive.datarush.operators.AbstractDeferredRecordOperator

getInput, getOutput

Methods inherited from class com.pervasive.datarush.operators.DeferredCompositeOperator

computeOutputTypes

Methods inherited from class com.pervasive.datarush.operators.AbstractLogicalOperator

disableParallelism, getInputPorts, getOutputPorts, newInput, newInput, newOutput, newRecordInput, newRecordInput, newRecordOutput, notifyError

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.pervasive.datarush.operators.LogicalOperator

disableParallelism, getInputPorts, getOutputPorts

Constructor Detail

PartitionHint

public PartitionHint()

Forces a partitioning into parallel streams of data. The resulting streams will be partitioned in an unspecified fashion and be unordered.

PartitionHint

public PartitionHint(DataDistribution partitioning)

Forces a partitioning into parallel streams of data guaranteeing the specified distribution. The resulting streams will be unordered.

Parameters:

partitioning - the requested data distribution

Method Detail

getPartitioning

public DataDistribution getPartitioning()

Gets the data distribution of the output.

Returns:

the requested data distribution

setPartitioning

public void setPartitioning(DataDistribution partitioning)

Sets the data distribution of the output. The output is guaranteed to be distributed in the specified manner.

Parameters:

partitioning - the requested data distribution

getDataOrdering

public DataOrdering getDataOrdering()

Gets the data ordering of the output.

Returns:

the requested data ordering

setDataOrdering

public void setDataOrdering(DataOrdering dataOrdering)

Sets the data ordering of the output. The output is guaranteed to be ordered in the specified manner.

Parameters:

dataOrdering - the requested data ordering

isForceStaging

public boolean isForceStaging()

Indicates whether data will be forcibly staged to disk.

Returns:

whether staging is required

setForceStaging

public void setForceStaging(boolean forceStaging)

Sets whether data must be staged to disk.

Parameters:

forceStaging - indicates whether the data must be staged

getPreserveSortOrder

public boolean getPreserveSortOrder()

If true, data order will be preserved ( provided that the data order is known ). Note that this can be a very expensive operation ( when running distributed it requires a re-sort following redistribution), so in-general this option should be avoided. If it is known that the downstream operations require data to be sorted, though, it may be advantageous to specify this option because, in the pseudo-distributed case, it is cheaper than performing a re-sort. By default, we do not preserve sort order.

Returns:

whether to preserve sort order.

setPreserveSortOrder

public void setPreserveSortOrder(boolean preserveSortOrder)

Sets whether data order will be preserved ( provided that the data order is known ). Note that this can be a very expensive operation ( when running distributed it requires a re-sort following redistribution), so in-general this option should be avoided. If it is known that the downstream operations require data to be sorted, though, it may be advantageous to specify this option because, in the pseudo-distributed case, it is cheaper than performing a re-sort. By default, we do not preserve sort order.

Parameters:

preserveSortOrder - whether to preserve sort order.

computeMetadata

protected void computeMetadata(StreamingMetadataContext ctx)

Description copied from class: DeferredCompositeOperator

Implementations must adhere to all of the contracts specified by StreamingOperator.computeMetadata(com.pervasive.datarush.operators.StreamingMetadataContext). In addition, DeferredCompositeOperators must declare required metadata so as to satisfy requirements of the operators that are added during DeferredCompositeOperator.compose(com.pervasive.datarush.operators.DeferredCompositionContext).

Specified by:

computeMetadata in class DeferredCompositeOperator

Parameters:

ctx - the context

compose

protected void compose(DeferredCompositionContext ctx)

Description copied from class: DeferredCompositeOperator

Compose the body of this operator. Implementations should do the following:

1. Instantiate and configure sub-operators, adding them to the provided context via the method OperatorComposable.add(O)
2. Create necessary connections via the method OperatorComposable.connect(P, P). This includes connections from the composite's input ports to sub-operators, connections between sub-operators, and connections from sub-operators output ports to the composite's output ports

Specified by:

compose in class DeferredCompositeOperator

Parameters:

ctx - the context