Windowed operations in Kafka - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of tumbling window aggregation

Given the following Kafka Streams code snippet that counts events per 5-second tumbling window, what is the output for the input events with timestamps in seconds: 1, 2, 3, 6, 7?

Kafka

KStream<String, String> input = builder.stream("input-topic");

KTable<Windowed<String>, Long> counts = input
  .groupByKey()
  .windowedBy(TimeWindows.ofSizeWithNoGrace(Duration.ofSeconds(5)))
  .count();

counts.toStream().foreach((windowedKey, count) -> {
  System.out.println("Window start: " + windowedKey.window().start() + ", count: " + count);
});

A

Window start: 0, count: 2
Window start: 5, count: 3

B

Window start: 0, count: 3
Window start: 5, count: 2

C

Window start: 1, count: 3
Window start: 6, count: 2

D Window start: 0, count: 5

💡 Hint

Remember that tumbling windows are fixed-size, non-overlapping intervals starting at multiples of the window size.

The 5-second tumbling windows start at 0 and 5 seconds. Events at timestamps 1, 2, 3 fall into the window starting at 0, so count is 3. Events at 6 and 7 fall into the window starting at 5, so count is 2.

❓ Predict Output

intermediate

Sliding window count output

What is the output of the following Kafka Streams sliding window count for events with timestamps 1, 2, 3, 6, 7, if the sliding window size is 5 seconds and advance interval is 1 second?

Kafka

KStream<String, String> input = builder.stream("input-topic");

KTable<Windowed<String>, Long> counts = input
  .groupByKey()
  .windowedBy(SlidingWindows.ofTimeDifferenceWithNoGrace(Duration.ofSeconds(5)))
  .count();

counts.toStream().foreach((windowedKey, count) -> {
  System.out.println("Window start: " + windowedKey.window().start() + ", count: " + count);
});

A

Window start: 1, count: 3
Window start: 2, count: 3
Window start: 3, count: 3
Window start: 6, count: 2
Window start: 7, count: 2

B

Window start: 0, count: 3
Window start: 1, count: 3
Window start: 2, count: 3
Window start: 5, count: 2
Window start: 6, count: 2

C

Window start: 1, count: 1
Window start: 2, count: 1
Window start: 3, count: 1
Window start: 6, count: 1
Window start: 7, count: 1

D Window start: 0, count: 5

💡 Hint

Sliding windows overlap and advance by 1 second, so events can be counted in multiple windows.

Sliding windows of 5 seconds with 1-second advance create overlapping windows starting at each second. Events at 1,2,3 fall into windows starting at 1,2,3 with count 3 each. Events at 6,7 fall into windows starting at 6,7 with count 2 each.

🔧 Debug

advanced

Identify the error in hopping window definition

The following Kafka Streams code attempts to create a hopping window of size 10 seconds that advances every 5 seconds. What error will this code produce?

Kafka

KStream<String, String> input = builder.stream("input-topic");

KTable<Windowed<String>, Long> counts = input
  .groupByKey()
  .windowedBy(HoppingWindows.ofSizeWithNoGrace(Duration.ofSeconds(10)).advanceBy(Duration.ofSeconds(5)))
  .count();

A No error, code runs correctly and produces hopping windows

B Runtime error: NullPointerException due to missing grace period

C Compilation error: method advanceBy() does not exist for HoppingWindows.ofSizeWithNoGrace()

D Compilation error: missing import for Duration class

💡 Hint

Check the Kafka Streams API for the correct way to specify advance interval for hopping windows.

The method advanceBy() is not available on the object returned by ofSizeWithNoGrace(). The correct way is to use HoppingWindows.of(Duration size, Duration advance) or chain advanceBy() before ofSizeWithNoGrace().

📝 Syntax

advanced

Correct syntax for session window with inactivity gap

Which of the following Kafka Streams code snippets correctly creates a session window with an inactivity gap of 10 seconds?

A SessionWindows.with(Duration.ofSeconds(10))

B SessionWindows.ofInactivityGapWithNoGrace(Duration.ofSeconds(10))

C SessionWindows.withInactivityGap(Duration.ofSeconds(10))

D SessionWindows.withGap(Duration.ofSeconds(10))

💡 Hint

Check the official Kafka Streams API for the correct method to specify session window gap.

The correct method to create a session window with a gap is SessionWindows.with(Duration gap). Other options are invalid or do not exist.

🚀 Application

expert

Number of windows created by hopping window

If you have a hopping window of size 12 seconds that advances every 4 seconds, how many overlapping windows will an event at timestamp 20 seconds belong to?

A 1

B 2

C 3

D 4

💡 Hint

Calculate how many windows of size 12 seconds, advancing every 4 seconds, cover the timestamp 20.

Hopping windows of size 12 seconds with 4 seconds advance overlap. Each event belongs to windows starting at timestamps: 8, 12, 16, and 20 seconds (all windows covering 20). So 4 windows include the event at 20 seconds.