...

...

[This FLIP proposal is a joint work between of Ran Jinhao and , Dong Lin  and Xuannan Su ]

Motivation

As shown in the figure below, we might have a job that pre-processes records from a bounded source (i.e. inputA) using an operator (i.e. operatorA) which only emits results after its input has ended. We will call such operator full-dam operator. Then operatorB needs to join records emitted by operatorA with records from an unbounded source, and emit results with low processing latency in real-time.

Currently, supporting the above use-case requires all operators to be deployed at the start of the job. This approach wastes slot and memory resources because operatorB can not do any useful work until operatorA's input has ended. Even worse, operatorB might use a lot of disk space only to cache and spill records received from the unbounded source to its local disk while it is waiting for operatorA's output.

In this FLIP, we propose to optimize task deployment and resource utilization for the above use-case by allowing an operator to explicitly specify whether it only emits records after all its inputs have ended. JM will leverage this information to optimize job scheduling such that the partition type of the results emitted by this operator , as well as the results emitted by its upstream operators, will all be blocking, which effectively let Flink schedule and execute this operator as well as its upstream operators in batch mode. Hybrid shuffle mode(FLIP-235: Hybrid Shuffle Mode) can be used in batch mode part to further improve the performance when there are sufficient slot resources.

In addition, this FLIP also allows creating a GlobalWindows that the window will be triggered only at the end of its inputs. DataStream programs In addition, this FLIP also changes the defualt trigger of GlobalWindows, which is NerverTrigger that nerver fires before, to EndOfStreamTrigger. GlobalWindows with new default trigger can be used with the DataStream API to specify whether the window will end only at the end of its inputs. DataStream program (e.g. coGroup and aggregate) can take advantage of this information to significantly increase throughput and reduce resource utilization.

Objectives

The APIs provided in this FLIP achieves achieve the following objectives / benefits:

1) Improve throughputReduce resource usage.  For use-case that involves a mixture of bounded and unbounded workloads (e.g. the use-case specified in the example), the user can start a single Flink job to address such use-case where the performance of the bounded workload can be optimal (e.g. similar/higher than the corresponding performance in batch mode).

2) Reduce resource usage.  For use-case that involves an operatorB (with unbounded input) depending on the output of another operatorA, where operatorA only emits results at the end of its input, Flink will deploy operatorB after operatorA is finished. This approach reduces the unnecessary resource usage when operatorA is still processing its inputs.

3) Improve usability. For use-case that needs to invoke DataStream APIs (e.g. KeyedStream#window) with a window assigner that covers all input data, users can use the off-the-shelf GlobalWindows with new default trigger provided in this FLIP, instead of writing tens of lines of code to define this WindowAssigner subclass. 

Public Interfaces

an operatorB (with unbounded input) depending on the output of another operatorA, where operatorA only emits results at the end of its input, Flink will deploy operatorB after operatorA is finished. This approach reduces unnecessary resource usage when operatorA is still processing its inputs.

2) Improve usability. For use-case that needs to invoke DataStream APIs (e.g. KeyedStream#window) with a window assigner that covers all input data, users can use the off-the-shelf GlobalWindows to create a GlobalWindow that is only triggered at the end of its inputs, instead of writing tens of lines of code to define a WindowAssigner subclass. 

3) Introduce attributes to the operator to let Flink know if the operator only outputs results when the input has ended. All the operators defined in Flink should set the attributes accordingly to achieve the benefit above. User-defined operators should be able to set the attributes accordingly so that they can achieve the benefit.

Public Interfaces

1) Add the createWithEndOfStreamTrigger method in GlobalWindows which allows users of the DataStream API to create GlobalWindows with a window trigger that only fires when the input ends1) Change a default trigger, GlobalWindows#NeverTrigger, to GlobalWindows#EndOfStreamTrigger. This window assigner with new default trigger allows users of the DataStream API to specify whether the computation (e.g. co-group, aggregate) should emit data only after end-of-input.

/**@PublicEvolving
public * Aclass GlobalWindows extends WindowAssigner<Object, GlobalWindow> {
   ···                   

    /**
     * Creates a {@link WindowAssigner} that assigns all elements to the same {@link GlobalWindow}.
 *
 * <p>Use this if* The youwindow wantis toonly useuseful aif {@linkyou Trigger}also andspecify {@link
a *custom org.apache.flink.streaming.api.windowing.evictors.Evictor} to do flexible, policy based windows.
 */
@PublicEvolving
public class GlobalWindows extends WindowAssigner<Object, GlobalWindow> {
	···
	/** A trigger that fires when input ends, as default Trigger for GlobalWindows. */
    @Internaltrigger. Otherwise, the window will
     * never be triggered and no computation will be performed.
     */
	public static GlobalWindows create() {
    public static class EndOfStreamTrigger extends Trigger<Object, GlobalWindow> { ...
    }     private static

 final long serialVersionUID = 1L;
 /**
     * Creates a @Override
{@link WindowAssigner} that assigns all elements to the publicsame TriggerResult onElement({@link GlobalWindow}
     * and the window is triggered if and only if the Objectinput element,stream long timestamp, GlobalWindow window, TriggerContext ctx) {
    is ended.
     */
	public static GlobalWindows createWithEndOfStreamTrigger() {
        return TriggerResult.CONTINUE;...
        }

        @Override
        public TriggerResult onEventTime(long time, GlobalWindow window, TriggerContext ctx) {
            return time == window.maxTimestamp() ? TriggerResult.FIRE : TriggerResult.CONTINUE;
        }

        @Override}

2) Add OperatorAttributesBuilder and OperatorAttributes for operator developers to specify operator attributes that Flink runtime can use to optimize the job performance.

package org.apache.flink.streaming.api.operators;
 
/** The builder class for {@link OperatorAttributes}. */
@Experimental
public class OperatorAttributesBuilder {
 
    public OperatorAttributesBuilder() {...}        

    /**
     * Set to true if and only if the operator only emits records after all its inputs have ended.
     * If it publicis TriggerResult onProcessingTime(long time, GlobalWindow window, TriggerContext ctx) {not set, the default value false is used.
     */   
     return TriggerResult.CONTINUE;public OperatorAttributesBuilder setOutputOnlyAfterEndOfStream(
        }

    boolean outputOnlyAfterEndOfStream) {...}   @Override
         

	/** If any operator publicattribute is voidnot clear(GlobalWindowset windowexplicitly, TriggerContextwe ctx)will throwslog Exception {}

        @Override
        public void onMerge(GlobalWindow window, OnMergeContext ctx) {}
    }
	···
}

2) Add OperatorAttributesBuilder and OperatorAttributes for operator developers to specify operator attributes that Flink runtime can use to optimize the job performance.

it at DEBUG level. */
	public OperatorAttributes build() {...}

}

package org.apache.flink.streaming.api.operators;
 
/**
 * OperatorAttributes element provides Job Manager with information that can be
 * used to optimize the job performance.
 */
@Experimental
public class OperatorAttributes {
  
    /**
     * Returns true if and only if the operator only emits records after all its inputs have ended.
     *
     * <p>Here are the implications when it is true:
     *

package org.apache.flink.streaming.api.operators;
 
/** The builder class for {@link OperatorAttributes}. */
@Experimental
public class OperatorAttributesBuilder {
    @Nullable private Boolean outputOnEOF = null;
 
    public OperatorAttributesBuilder() {...}
 
    public OperatorAttributesBuilder setOutputOnEOF(boolean outputOnEOF) {...}
   
    /**
     * If<ul>
 any operator attribute is null,* we will log<li> itThe atresults DEBUGof levelthis andoperator useas thewell following
as its chained operators have *blocking defaultpartition valuestype.
     * - outputOnEOF defaults<li> toThis false
operator as well as its */
chained operators will be executed in batch mode.
     * </ul>
     */     
	public OperatorAttributesboolean buildisOutputOnlyAfterEndOfStream() {...}
}

3) Add the getOperatorAttributes() API to the StreamOperator and StreamOperatorFactory interfaces.

package org.apache.flink.streaming.api.operators;
 
public interface StreamOperator<OUT> extends CheckpointListener, KeyContext, Serializable {
    ...
	
    /**
     * OperatorAttributesGet elementthe provides{@link JobOperatorAttributes} Managerof withthe informationoperator that canthe be
Flink *runtime usedcan use to optimize
     * the job performance.
 */
@Experimental
public class OperatorAttributes {    
   /**
     * Returns@return trueOperatorAttributes iffof the operator.
 can only emit records after*/
 inputs have reached EOF.@Experimental
    default OperatorAttributes getOperatorAttributes() *{
     *  <p>Here arereturn the implications when it is true:
     *new OperatorAttributesBuilder().build();
    }
}
 

public interface StreamOperatorFactory<OUT> extends Serializable {
    ...

    /* <ul>*
     * Get the <li>{@link TheOperatorAttributes} results of thisthe operator ascreated wellby asthe itsfactory. upstreamFlink operators have blocking partition type.runtime can
     * use it to <li>optimize Thisthe operatorjob asperformance.
 well as its upstream operators*
 will be executed in batch* mode.
@return OperatorAttributes of the operator * </ul>created by the factory.
     */
    @Experimental
    publicdefault booleanOperatorAttributes isOutputOnEOFgetOperatorAttributes() {...
        return new OperatorAttributesBuilder().build();
    }
}

Proposed Changes

13) Add the getOperatorAttributes() API to the StreamOperator and StreamOperatorFactory interfacesAPIs on PhysicalTransformation interface to get the corresponding operator attributes.

@Experimental@Internal 
public abstract interfaceclass StreamOperator<OUT>PhysicalTransformation<T> extends CheckpointListener, KeyContext, Serializable Transformation<T> {
    ...
 
    default OperatorAttributes getOperatorAttributespublic boolean isOutputOnlyAfterEndOfStream() {
        return new OperatorAttributesBuilder().build()false;
    }
}
 
@Experimental
public interface StreamOperatorFactory<OUT> extends Serializable {
    ...
 
    default OperatorAttributes getOperatorAttributes() {
        return new OperatorAttributesBuilder().build();
    }
}

Proposed Changes

1) Add the APIs on Transformation interface to get the corresponding operator attributes.

@Internal
public abstract class Transformation<T> {
    public boolean isOutputOnEOF() {
        return false;
    }
}

2) Update Transformation subclasses (e.g. OneInputTransformation and TwoInputTransformation) to override the newly added methods using the OperatorAttributes obtained from the underlying Operator.

3) Update JM to make use of the following operator attributes when compiling the Transformation graph into the JobGraph.

If a Transformation has isOutputOnEOF == true:

...

4) A blocking input edge with pending records is same as a source with isBacklog=true when an operator determines its RecordAttributes for downstream nodes.

This is needed in order for this FLIP to work with FLIP-327. More specifically, once both FLIP-327 and FLIP-331 are accepted, we need a way to determine the backlog status for input with blocking edge type.

5) When DataStream#coGroup is invoked with EndOfStreamWindows as the window assigner, Flink should generate an operator with isOutputOnEOF = true.

In addition, after FLIP-327 is accepted, suppose this API is called without any explicitly-assigned trigger or evictor, the program should also have isInternalSorterSupported = true to achieve higher throughput by using the optimizations currently done in batch mode.

The following optimization will be used to achieve higher throughput than the existing DataStream#coGroup in both stream and batch mode:

• It will instantiate two internal sorter to sorts records from its two inputs separately. Then it can pull the sorted records from these two sorters. This can be done without wrapping input records with TaggedUnion<...>. In comparison, the existing DataStream#coGroup needs to wrap input records with TaggedUnion<...> before sorting them using one external sorter, which introduces higher overhead.
• It will not invoke WindowAssigner#assignWindows or triggerContext#onElement for input records. In comparison, the existing WindowOperator#processElement invokes these methods for every records.

6) When DataStream#aggregate is invoked with GlobalWindows and its default trigger as the window assigner and trigger, Flink should generate an operator with isOutputOnEOF = true.

In addition, after FLIP-327 is accepted, suppose this API is called without any explicitly-assigned trigger or evictor, the program should also have isInternalSorterSupported = true to achieve higher throughput by using the optimizations currently done in batch mode.

More specifically, This operator will sort the input before aggregation, and avoid invoking window actions, which is similar to '5)'.

Analysis of APIs affected by this FLIP

This FLIP can potentially benefit all DataStream APIs that take WindowAssigner as the parameter.

More specifically, the following DataStream API can benefit from using GlobalWindows with default trigger.

• DataStream#windowAll
• KeyedStream#window
• CoGroupedStreams#Where#EqualTo#window
• JoinedStreams#Where#EqualTo#window

Benchmark results

To demonstrate our optimization improvements，we run each benchmark in different execution modes and configurations. Each result is measured by taking the average execution time across 5 runs with the given configuration.

We run benchmarks on a MacBook with the latest Flink 1.18-snapshot and parallelism=1. Default standalone cluster configuration is used except:

jobmanager.memory.process.size: 6400m
taskmanager.memory.process.size: 6912m

1) Execute DataStream#CoGroup

This benchmark uses DataStream#coGroup to process records from two bounded inputs. Each input will generate records with (key = i, value = i) for i from 1 to 8*10^7.

Below is the DataStream program code snippet.

DataStream<Tuple2<Integer, Double>> source1  = env.fromCollection(
                new DataGenerator(dataNum), Types.TUPLE(Types.INT, Types.DOUBLE));
DataStream<Tuple2<Integer, Double>> source2  = env.fromCollection(
                new DataGenerator(dataNum), Types.TUPLE(Types.INT, Types.DOUBLE));

source1.coGroup(source2)
    .where(tuple -> tuple.f0)
    .equalTo(tuple -> tuple.f0)
    .window(GlobalWindows.create())
    .apply(new CustomCoGroupFunction())
    .addSink(...);

private static class CustomCoGroupFunction
            extends RichCoGroupFunction<Tuple2<Integer, Double>, Tuple2<Integer, Double>, Integer> {
        @Override
        public void coGroup(
                Iterable<Tuple2<Integer, Double>> iterableA,
                Iterable<Tuple2<Integer, Double>> iterableB,
                Collector<Integer> collector) {
            collector.collect(1);
        }
    }

The following result shows the throughput (records/sec) when the benchmark is executed in streaming mode, batch mode, optimized streaming mode after this PR, and optimized streaming mode with hybrid shuffle after this PR.

The result shows that DataStream#coGroup in optimized streaming mode can be 22X as fast as streaming mode and 3X as fast as batch mode. Hybrid shuffle can further improve throughput by 11%.

2) Execute DataStream#Aggregate

This benchmark uses DataStream#aggregate to process 8*10^7 records. These records are evenly distributed across 8*10^5 keys. More specifically, the source will generate records with (key = i, value = i) for i from 1 to 8*10^5, and repeat this process 100 times.

Below is the DataStream program code snippet.

DataStreamSource<Tuple2<Long, Double>> source  = env.fromCollection(
                        new DataGenerator(dataNum, keyNum), Types.TUPLE(Types.LONG, Types.DOUBLE));
source.keyBy(value -> value.f0)
      .window(GlobalWindows.create())
      .aggregate(new Aggregator())
      .addSink(...);

public static class Aggregator implements AggregateFunction<Tuple2<Long, Double>, Tuple2<Long, Double>, Double> {
        @Override
        public Tuple2<Long, Double> createAccumulator() {
            return new Tuple2<Long, Double>(0L, 0.0);
        }

        @Override
        public Tuple2<Long, Double> add(Tuple2<Long, Double> myData, Tuple2<Long, Double> accData) {
            accData.f1 = accData.f1 + myData.f1;
            return accData;
        }

        @Override
        public Double getResult(Tuple2<Long, Double> result) {
            return result.f1;
        }

        @Override
        public Tuple2<Long, Double> merge(Tuple2<Long, Double> acc1, Tuple2<Long, Double> acc2) {
            acc1.f1 = acc1.f1 + acc2.f1;
            return acc1;
        }
    }

The following result shows the throughput (records/sec) when the benchmark is executed in streaming mode, batch mode, optimized streaming mode after this PR, and optimized streaming mode with hybrid shuffle after this PR.

The result shows that DataStream#aggregate in optimized streaming mode can be 10X as fast as streaming mode and 11% faster than batch mode. Hybrid shuffle can further improve throughput by 15%.

3) Execute a program that needs to fully process data from a bounded source before processing data from another unbounded source.

The following program demonstrates the scenario described in the motivation section. The program needs to pre-processes records from a bounded source (Source1) using an operator (Process1) which only emits results after its input has ended. Then anther operator(Process2) needs to process records emitted by Process1 with records from an unbounded source, and emit results with low processing latency in real-time.

 source1.keyBy(value -> value.f0)
  .window(GlobalWindows.create())
  .aggregate(new MyAggregator()).name("Process1")
  .connect(source2.keyBy(value -> value.f0))
  .transform("Process2", Types.INT, new MyProcessOperator())
  .addSink(...); 

We can use this program to demonstrate that the program requires less slot resources. More specifically, suppose we configure the standalone cluster with taskmanager.numberOfTaskSlots = 2, and set the Source1，Process1, Source2 and Process2 in 4 different SlotSharingGroups, the program will fail to be deployed before this FLIP. And the program can be deployed successfully after this FLIP. This is because Source2 and Process2 can be deplyed after Source1，Process1 finished and released their slots.

Additionally, we can use this program to demonstrate that it can achieve higher performance because Process2 will not need to keep buffer records emitted by Source2 in its memory while Process1 has not reached EOF. More specifically, the program can fail with OOM before this FLIP when the number of records in inputs is high. And the program can finish successfully without OOM after this FLIP.

Compatibility, Deprecation, and Migration Plan

The changes made in this FLIP are backward compatible. No deprecation or migration plan is needed.

Future Work

...

2) Update the following Transformation subclasses to override the newly added methods using the OperatorAttributes obtained from the underlying Operator:

• OneInputTransformation
• TwoInputTransformation
• AbstractMultipleInputTransformation

3) Update JM to make use of the following operator attributes when compiling the Transformation graph into the JobGraph.

If a Transformation has isOutputOnlyAfterEndOfStream == true:

• The process of operator chaining can still be done. After that, the results of its operator chain will be set blocking.
• This operator will be executed in batch mode.
  
  More specifically, the checkpoint is disabled when these operators are running, such that these operators can do operations not compatible with checkpoints (e.g. sorting inputs). And operators should re-read the data from the upstream blocking edge or sources after failover. 

4) Enumerate all existing operators defined in Flink and set the isOutputOnlyAfterEndOfStream accordingly for those operators that may only output at the end of the input to enable the optimization mentioned in 3):

• WindowOperator: When used with GlobalWindows that is created with the method createWithEndOfStreamTrigger as the window assigner
• StreamSortOperator

5) When DataStream API like DataStream#CoGroup is invoked with GlobalWindows, the WindowOperator will check if the window assigner is an instance of GlobalWindow and only triggers at the end of the inputs. If true, the operator will have OperatorAttributes with isOutputOnlyAfterEndOfStream = true to achieve higher throughput by using the optimizations in batch mode.

Analysis of APIs affected by this FLIP

This FLIP can potentially benefit all DataStream APIs that take WindowAssigner as the parameter.

More specifically, the following DataStream API can benefit from using GlobalWindows which is created with the method createWithEndOfStreamTrigger.

• DataStream#windowAll
• KeyedStream#window
• CoGroupedStreams#Where#EqualTo#window
• JoinedStreams#Where#EqualTo#window

Compatibility, Deprecation, and Migration Plan

The changes made in this FLIP are backward compatible. No deprecation or migration plan is needed.

As mentioned in Proposed Change 4), we will enumerate all the existing operators defined in Flink and set the isOutputOnlyAfterEndOfStream accordingly. For user-defined operators, their behavior stays the same as before. If the user-defined operators only output at the end of input, they must set the isOutputOnlyAfterEndOfStream to true to enable the optimization.

Future Work

• It would be useful to add an ExecutionState (e.g. Finishing) to specify whether the task has reached the end for all its inputs. This allows JM to deploy its downstream tasks and possibly apply hybrid shuffle to increase job throughput.
• Setting the upstream operators of the full-dam operator in batch mode will also increase the job throughput, and Hybrid shuffle mode can also be used in batch mode part to further improve the performance when there are sufficient slot resources.
• Optimizing the implementation of frequently used full-dam operators,  such as aggregate/CoGroup/Join/..., can achieve higher throughput by using the optimizations currently done in batch mode. For example, we can instantiate an internal sorter in the operator, and it will not have to invoke WindowAssigner#assignWindows or triggerContext#onElement for each input record.