Batching and linger configuration in Kafka - Step-by-Step Execution

Process Flow - Batching and linger configuration

Message arrives to producer

↓

Add message to batch buffer

↓

Is batch size reached?

No→Wait linger.ms time

↓

Send batch

↓

Send batch

↓

Clear batch buffer

↓

Wait for next message

Messages are collected in a batch buffer. When the batch size is reached or linger time expires, the batch is sent.

Execution Sample

Kafka

producer_config = {
  'batch.size': 16384,  # bytes
  'linger.ms': 5       # milliseconds
}

# Messages arrive and are batched before sending

This config sets batch size and linger time to control when messages are sent in batches.

Process Table

Step	Message Arrived	Batch Size (bytes)	Batch Full?	Linger Timer (ms)	Action	Output
1	Msg1 (4000 bytes)	4000	No	Start 5ms timer	Add to batch	No send
2	Msg2 (5000 bytes)	9000	No	Timer running	Add to batch	No send
3	Msg3 (8000 bytes)	17000	Yes (17000 > 16384)	Cancel timer	Send batch	Batch sent with Msg1, Msg2, Msg3
4	Msg4 (3000 bytes)	3000	No	Start 5ms timer	Add to batch	No send
5	No new message	3000	No	Timer expires	Send batch	Batch sent with Msg4
6	-	0	-	-	Batch cleared	Ready for new messages

💡 Batch sent when size limit reached or linger timer expires

Status Tracker

Variable	Start	After 1	After 2	After 3	After 4	After 5	Final
Batch Size (bytes)	0	4000	9000	17000	3000	0	0
Linger Timer (ms)	Off	5ms started	5ms running	Cancelled	5ms started	Expired	Off
Batch Content	[]	[Msg1]	[Msg1, Msg2]	[Msg1, Msg2, Msg3]	[Msg4]	[]	[]

Key Moments - 3 Insights

Why does the batch send immediately at step 3 instead of waiting for the linger timer?

Because the batch size exceeded the configured limit (17000 > 16384 bytes), the batch sends immediately, cancelling the linger timer as shown in step 3 of the execution_table.

What happens if no new messages arrive before the linger timer expires?

The linger timer triggers sending the current batch even if it is not full, as shown at step 5 where the timer expires and the batch with Msg4 is sent.

Does the linger timer start again after sending a batch?

No, after sending a batch, the batch buffer clears and the linger timer is off until a new message arrives, starting a new timer as shown at step 6.

Visual Quiz - 3 Questions

Look at the execution_table, what is the batch size after step 2?

A 4000 bytes

B 17000 bytes

C 9000 bytes

D 3000 bytes

💡 Hint

Check the 'Batch Size (bytes)' column at step 2 in the execution_table.

At which step does the linger timer get cancelled due to batch size?

A Step 3

B Step 1

C Step 2

D Step 5

💡 Hint

Look for 'Cancel timer' action in the execution_table.

If the linger.ms was set to 0, what would happen to the batch sending behavior?

A Batches would send immediately when batch size is reached only.

B Batches would send immediately after every message without waiting.

C Batches would never send.

D Batches would send only when linger timer expires.

💡 Hint

Consider how linger.ms controls waiting time before sending batches.

Concept Snapshot

Batching collects messages up to batch.size bytes.
linger.ms sets max wait time before sending batch.
Batch sends when size reached or linger expires.
Helps improve throughput by sending fewer larger requests.
Configurable in Kafka producer settings.

Full Transcript

This visual execution shows how Kafka producer batches messages before sending. Messages add to a batch buffer. If the batch size exceeds the configured batch.size, the batch sends immediately, cancelling any linger timer. If the batch size is not reached, the linger.ms timer starts on the first message and waits that time before sending the batch. This ensures messages are sent timely even if batch size is small. Variables tracked include batch size in bytes, linger timer state, and batch content. Key moments clarify why batches send immediately on size limit and how linger timer triggers sending when no new messages arrive. The quiz tests understanding of batch size changes, timer cancellation, and effects of linger.ms settings. This helps beginners see how batching and linger configuration control message sending in Kafka producers.