Punctuators for time-based triggers in Kafka - Step-by-Step Execution

Process Flow - Punctuators for time-based triggers

Start Stream Processing

↓

Schedule Punctuator

↓

Wait for Time Interval

↓

Punctuator Triggered

↓

Execute Punctuator Logic

↓

Continue Processing Records

↩Back to Wait for Time Interval

The flow shows how Kafka Streams schedules a punctuator to run repeatedly at fixed time intervals, triggering custom logic while processing stream records.

Execution Sample

Kafka

streamProcessorContext.schedule(Duration.ofSeconds(10), PunctuationType.WALL_CLOCK_TIME, timestamp -> {
    System.out.println("Punctuator triggered at " + timestamp);
});

This code schedules a punctuator to run every 10 seconds, printing the current timestamp each time it triggers.

Process Table

Step	Action	Time (seconds)	Punctuator Triggered?	Output
1	Start stream processing	0	No
2	Schedule punctuator every 10 seconds	0	No
3	Wait until 10 seconds elapsed	10	Yes	Punctuator triggered at 10
4	Execute punctuator logic	10	Yes	Printed timestamp 10
5	Wait until 20 seconds elapsed	20	Yes	Punctuator triggered at 20
6	Execute punctuator logic	20	Yes	Printed timestamp 20
7	Wait until 30 seconds elapsed	30	Yes	Punctuator triggered at 30
8	Execute punctuator logic	30	Yes	Printed timestamp 30
9	Stop stream processing	30+	No	Stream closed
10	Exit		No	Punctuator stops as stream ends

💡 Stream processing stopped, so punctuator no longer triggers.

Status Tracker

Variable	Start	After 10s	After 20s	After 30s	Final
timestamp	0	10	20	30	Stream ended

Key Moments - 3 Insights

Why does the punctuator trigger at exact intervals even if no records arrive?

Because the punctuator is scheduled with WALL_CLOCK_TIME, it triggers based on real time passing, independent of record arrival, as shown in execution_table rows 3, 5, and 7.

What happens if the stream processing stops before the next punctuator time?

The punctuator stops triggering because the stream is closed, as shown in execution_table row 9 and exit_note.

Can the punctuator modify state or produce output?

Yes, the punctuator logic runs inside the scheduled callback and can update state or produce output, demonstrated by the printed timestamp in execution_table outputs.

Visual Quiz - 3 Questions

Look at the execution_table, what is the timestamp value when the punctuator triggers for the second time?

A 10

B 30

C 20

D 0

💡 Hint

Check the 'Time (seconds)' column at the second punctuator trigger in execution_table row 5.

At which step does the stream processing stop, causing the punctuator to stop?

A Step 8

B Step 9

C Step 7

D Step 10

💡 Hint

Look at execution_table row 9 where the stream is stopped.

If the punctuator was scheduled every 5 seconds instead of 10, how would the output change by step 6?

A It would trigger twice as often, so output at 5 and 10 seconds.

B It would trigger less often, only once at 10 seconds.

C It would not trigger at all.

D It would trigger only once at 20 seconds.

💡 Hint

Refer to the 'Time (seconds)' and 'Punctuator Triggered?' columns in execution_table to understand timing.

Concept Snapshot

Punctuators in Kafka Streams run code at scheduled time intervals.
Use streamProcessorContext.schedule() with a Duration and PunctuationType.
WALL_CLOCK_TIME triggers based on real time, independent of record flow.
Punctuator callback receives the current timestamp.
Useful for periodic tasks like flushing or state cleanup.

Full Transcript

This visual execution shows how Kafka Streams uses punctuators to run code at fixed time intervals. The stream starts and schedules a punctuator every 10 seconds. The punctuator triggers at 10, 20, and 30 seconds, printing the timestamp each time. The variable 'timestamp' updates accordingly. When the stream stops, the punctuator stops triggering. Key points include that punctuators run on real time, can run even without new records, and stop when the stream ends.