What This Guide Covers

Kafka is powerful. But how you consume messages can either throttle or turbocharge your system.
In this deep-dive, we’ll cover:

• 3 Kafka consumption patterns: @KafkaListener, manual polling, and reactive Kafka
• Advanced considerations: throughput, backpressure, retries, and error recovery
• Real-world metaphors with Bob (our everyman hero)
• Illustrated cartoon visuals to solidify each mental model
• Retry and Dead Letter logic for production-grade reliability

---

1. Meet Bob

Bob isn’t a developer. He’s just a regular guy who navigates life like we all do:

• Doctor’s appointments
• Airports and travel
• Mobile check-ins
• Burrito emergencies

We’ll use Bob’s everyday scenarios as mental models to understand Kafka consumer behavior.
Because, believe it or not, Kafka is everywhere — even in Bob’s daily life.

2. @KafkaListener: Bob at the Doctor’s Office

Bob walks into his local clinic. He:

• Takes a number
• Waits in a crowded room
• Gets called one by one
• Sees the doctor only when it’s his turn

This is exactly how @KafkaListener works.

Code Example

@KafkaListener(topics = “invoices”, groupId = “billing”) public void listen(String message) { processMessage(message); // Simulates Bob’s checkup }

Technical Breakdown

Bob’s Verdict: “It works, but man… that waiting room was painful.”

4. Reactive Kafka: Bob Checks In Online

Now Bob’s just flexing.

From his phone, he:

• Checks in online
• Gets a mobile boarding pass
• Arrives, scans, and walks right in

No threads. No blocking. Everything is event-driven and backpressure-aware.

Code Example

KafkaReceiver.create(receiverOptions) .receive() .concatMap(record -> Mono.fromRunnable(() -> { process(record.value()); record.receiverOffset().acknowledge(); })) .subscribe();

Technical Breakdown

Bob’s Verdict: “I checked in while brushing my teeth. Welcome to the future.”

Retry & Dead Letter Queues: Bob’s Burrito Catastrophe

Bob orders a burrito. But:

• It’s too spicy
• It’s delivered to the wrong address
• His card gets declined

Each failure is retried — but after 3 tries, the burrito is trashed and he gets a refund.

In Kafka, we need the same logic: retry a few times, then move the message to a Dead Letter Topic (DLT).

Retry Logic in Spring Kafka

@Bean public DefaultErrorHandler errorHandler(KafkaTemplate