org.apache.flink.api.common.eventtime

Interface WatermarkStrategy<T>

All Superinterfaces:

Serializable, TimestampAssignerSupplier<T>, WatermarkGeneratorSupplier<T>

All Known Implementing Classes:

AssignerWithPeriodicWatermarksAdapter.Strategy, AssignerWithPunctuatedWatermarksAdapter.Strategy

@Public
public interface WatermarkStrategy<T>
extends TimestampAssignerSupplier<T>, WatermarkGeneratorSupplier<T>

The WatermarkStrategy defines how to generate Watermarks in the stream sources. The WatermarkStrategy is a builder/factory for the WatermarkGenerator that generates the watermarks and the TimestampAssigner which assigns the internal timestamp of a record.

This interface is split into three parts: 1) methods that an implementor of this interface needs to implement, 2) builder methods for building a WatermarkStrategy on a base strategy, 3) convenience methods for constructing a WatermarkStrategy for common built-in strategies or based on a WatermarkGeneratorSupplier

Implementors of this interface need only implement createWatermarkGenerator(WatermarkGeneratorSupplier.Context). Optionally, you can implement createTimestampAssigner(TimestampAssignerSupplier.Context).

The builder methods, like withIdleness(Duration) or createTimestampAssigner(TimestampAssignerSupplier.Context) create a new WatermarkStrategy that wraps and enriches a base strategy. The strategy on which the method is called is the base strategy.

The convenience methods, for example forBoundedOutOfOrderness(Duration), create a WatermarkStrategy for common built in strategies.

This interface is Serializable because watermark strategies may be shipped to workers during distributed execution.

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.flink.api.common.eventtime.TimestampAssignerSupplier

TimestampAssignerSupplier.Context, TimestampAssignerSupplier.SupplierFromSerializableTimestampAssigner<T>

Nested classes/interfaces inherited from interface org.apache.flink.api.common.eventtime.WatermarkGeneratorSupplier

WatermarkGeneratorSupplier.Context

Method Summary

Modifier and Type	Method and Description
default TimestampAssigner<T>	createTimestampAssigner(TimestampAssignerSupplier.Context context) Instantiates a TimestampAssigner for assigning timestamps according to this strategy.
WatermarkGenerator<T>	createWatermarkGenerator(WatermarkGeneratorSupplier.Context context) Instantiates a WatermarkGenerator that generates watermarks according to this strategy.
static <T> WatermarkStrategy<T>	forBoundedOutOfOrderness(Duration maxOutOfOrderness) Creates a watermark strategy for situations where records are out of order, but you can place an upper bound on how far the events are out of order.
static <T> WatermarkStrategy<T>	forGenerator(WatermarkGeneratorSupplier<T> generatorSupplier) Creates a watermark strategy based on an existing WatermarkGeneratorSupplier.
static <T> WatermarkStrategy<T>	forMonotonousTimestamps() Creates a watermark strategy for situations with monotonously ascending timestamps.
default WatermarkAlignmentParams	getAlignmentParameters() Provides configuration for watermark alignment of a maximum watermark of multiple sources/tasks/partitions in the same watermark group.
static <T> WatermarkStrategy<T>	noWatermarks() Creates a watermark strategy that generates no watermarks at all.
default WatermarkStrategy<T>	withIdleness(Duration idleTimeout) Creates a new enriched WatermarkStrategy that also does idleness detection in the created WatermarkGenerator.
default WatermarkStrategy<T>	withTimestampAssigner(SerializableTimestampAssigner<T> timestampAssigner) Creates a new WatermarkStrategy that wraps this strategy but instead uses the given SerializableTimestampAssigner.
default WatermarkStrategy<T>	withTimestampAssigner(TimestampAssignerSupplier<T> timestampAssigner) Creates a new WatermarkStrategy that wraps this strategy but instead uses the given TimestampAssigner (via a TimestampAssignerSupplier).
default WatermarkStrategy<T>	withWatermarkAlignment(String watermarkGroup, Duration maxAllowedWatermarkDrift) Creates a new WatermarkStrategy that configures the maximum watermark drift from other sources/tasks/partitions in the same watermark group.
default WatermarkStrategy<T>	withWatermarkAlignment(String watermarkGroup, Duration maxAllowedWatermarkDrift, Duration updateInterval) Creates a new WatermarkStrategy that configures the maximum watermark drift from other sources/tasks/partitions in the same watermark group.

Methods inherited from interface org.apache.flink.api.common.eventtime.TimestampAssignerSupplier

of

Method Detail

createWatermarkGenerator

WatermarkGenerator<T> createWatermarkGenerator(WatermarkGeneratorSupplier.Context context)

Instantiates a WatermarkGenerator that generates watermarks according to this strategy.

Specified by:

createWatermarkGenerator in interface WatermarkGeneratorSupplier<T>

createTimestampAssigner

default TimestampAssigner<T> createTimestampAssigner(TimestampAssignerSupplier.Context context)

Instantiates a TimestampAssigner for assigning timestamps according to this strategy.

Specified by:

createTimestampAssigner in interface TimestampAssignerSupplier<T>

getAlignmentParameters

@PublicEvolving
default WatermarkAlignmentParams getAlignmentParameters()

Provides configuration for watermark alignment of a maximum watermark of multiple sources/tasks/partitions in the same watermark group. The group may contain completely independent sources (e.g. File and Kafka).

Once configured Flink will "pause" consuming from a source/task/partition that is ahead of the emitted watermark in the group by more than the maxAllowedWatermarkDrift.

withTimestampAssigner

default WatermarkStrategy<T> withTimestampAssigner(TimestampAssignerSupplier<T> timestampAssigner)

Creates a new WatermarkStrategy that wraps this strategy but instead uses the given TimestampAssigner (via a TimestampAssignerSupplier).

You can use this when a TimestampAssigner needs additional context, for example access to the metrics system.

  WatermarkStrategy<Object> wmStrategy = WatermarkStrategy
   .forMonotonousTimestamps()
   .withTimestampAssigner((ctx) -> new MetricsReportingAssigner(ctx));
 

withTimestampAssigner

default WatermarkStrategy<T> withTimestampAssigner(SerializableTimestampAssigner<T> timestampAssigner)

Creates a new WatermarkStrategy that wraps this strategy but instead uses the given SerializableTimestampAssigner.

You can use this in case you want to specify a TimestampAssigner via a lambda function.

  WatermarkStrategy<CustomObject> wmStrategy = WatermarkStrategy
   .<CustomObject>forMonotonousTimestamps()
   .withTimestampAssigner((event, timestamp) -> event.getTimestamp());
 

withIdleness

default WatermarkStrategy<T> withIdleness(Duration idleTimeout)

Creates a new enriched WatermarkStrategy that also does idleness detection in the created WatermarkGenerator.

Add an idle timeout to the watermark strategy. If no records flow in a partition of a stream for that amount of time, then that partition is considered "idle" and will not hold back the progress of watermarks in downstream operators.

Idleness can be important if some partitions have little data and might not have events during some periods. Without idleness, these streams can stall the overall event time progress of the application.

withWatermarkAlignment

@PublicEvolving
default WatermarkStrategy<T> withWatermarkAlignment(String watermarkGroup,
                                                                    Duration maxAllowedWatermarkDrift)

Creates a new WatermarkStrategy that configures the maximum watermark drift from other sources/tasks/partitions in the same watermark group. The group may contain completely independent sources (e.g. File and Kafka).

Once configured Flink will "pause" consuming from a source/task/partition that is ahead of the emitted watermark in the group by more than the maxAllowedWatermarkDrift.

Parameters:

watermarkGroup - A group of sources to align watermarks

maxAllowedWatermarkDrift - Maximal drift, before we pause consuming from the source/task/partition

withWatermarkAlignment

@PublicEvolving
default WatermarkStrategy<T> withWatermarkAlignment(String watermarkGroup,
                                                                    Duration maxAllowedWatermarkDrift,
                                                                    Duration updateInterval)

Creates a new WatermarkStrategy that configures the maximum watermark drift from other sources/tasks/partitions in the same watermark group. The group may contain completely independent sources (e.g. File and Kafka).

Once configured Flink will "pause" consuming from a source/task/partition that is ahead of the emitted watermark in the group by more than the maxAllowedWatermarkDrift.

Parameters:

watermarkGroup - A group of sources to align watermarks

maxAllowedWatermarkDrift - Maximal drift, before we pause consuming from the source/task/partition

updateInterval - How often tasks should notify coordinator about the current watermark and how often the coordinator should announce the maximal aligned watermark.

forMonotonousTimestamps

static <T> WatermarkStrategy<T> forMonotonousTimestamps()

Creates a watermark strategy for situations with monotonously ascending timestamps.

The watermarks are generated periodically and tightly follow the latest timestamp in the data. The delay introduced by this strategy is mainly the periodic interval in which the watermarks are generated.

See Also:

AscendingTimestampsWatermarks

forBoundedOutOfOrderness

static <T> WatermarkStrategy<T> forBoundedOutOfOrderness(Duration maxOutOfOrderness)

Creates a watermark strategy for situations where records are out of order, but you can place an upper bound on how far the events are out of order. An out-of-order bound B means that once the an event with timestamp T was encountered, no events older than T - B will follow any more.

The watermarks are generated periodically. The delay introduced by this watermark strategy is the periodic interval length, plus the out of orderness bound.

See Also:

BoundedOutOfOrdernessWatermarks

forGenerator

static <T> WatermarkStrategy<T> forGenerator(WatermarkGeneratorSupplier<T> generatorSupplier)

Creates a watermark strategy based on an existing WatermarkGeneratorSupplier.

noWatermarks

static <T> WatermarkStrategy<T> noWatermarks()

Creates a watermark strategy that generates no watermarks at all. This may be useful in scenarios that do pure processing-time based stream processing.