ท็อปปิก พาร์ทิชัน และออฟเซ็ต

Welcome! In this lesson, we'll explore three fundamental concepts in Kafka: Topics, Partitions, and Offsets.

These elements are crucial for understanding how Kafka organizes messages, scales, and ensures reliable processing.

Think of a Topic as a category or feed name where messages are published and stored.

Producers send messages to specific topics, and consumers subscribe to topics to receive messages.

For example, you might have a user-signups topic or an order-updates topic.

Imagine a TV broadcasting station. Each "channel" (like News, Sports, or Movies) is a Topic.

• Producers are the broadcasters sending programs to specific channels.
• Consumers are viewers tuning into their preferred channels.

Messages within a topic are ordered and immutable.

To handle large volumes of messages and enable parallel processing, Kafka divides a topic into multiple Partitions.

Each partition is an ordered, immutable sequence of messages within a topic.

When you create a topic, you specify how many partitions it should have.

Partitions are key to Kafka's scalability. Multiple consumers can read from different partitions of the same topic in parallel.

This means if a topic has 3 partitions, up to 3 consumers (in the same consumer group) can process messages simultaneously.

When a producer sends a message to a topic, Kafka decides which partition it goes into.

This is often based on a message key. Messages with the same key usually go to the same partition, ensuring order for related events.

If no key is provided, messages are typically distributed in a round-robin fashion.

Within each partition, every message is assigned a unique, sequential ID called an Offset.

The offset is like an index number for messages within that specific partition.

It starts from 0 for the first message and increments for each new message.

Consumers use offsets to keep track of which messages they have already processed in a partition.

When a consumer group processes messages, it "commits" its offset, indicating the last message it successfully handled.

This ensures that upon restart or failure, the consumer can resume from the correct position.

Kafka automatically manages offset commits by default, saving the last processed offset for each partition in a special Kafka topic (__consumer_offsets).

This "saving your spot" mechanism prevents re-processing already handled messages and ensures reliable delivery.

Let's test your understanding of Kafka's fundamental building blocks.

Great job! You've grasped three core Kafka concepts:

• Topics: Categories for messages, like distinct data feeds.
• Partitions: Divisions of a topic, enabling scalability and parallel consumption.
• Offsets: Unique, sequential IDs for messages within a partition, used by consumers to track progress.

These elements work together to provide Kafka's robust and scalable messaging capabilities.