Why Message key and value in Kafka? - Purpose & Use Cases

The Big Idea

What if you could instantly find the right message without digging through a mess?

The Scenario

Imagine you have a huge stack of letters to send to different friends. You write their names on the envelopes but don't organize them. When you want to find a letter for a specific friend, you have to search through the whole pile every time.

The Problem

This manual way is slow and confusing. You might lose letters or send the wrong one. It's hard to keep track of who gets what, especially when the pile grows bigger and bigger.

The Solution

Using message keys and values in Kafka is like labeling each letter with a clear name (key) and the message inside (value). This helps Kafka organize and deliver messages efficiently, so you always know which message belongs to whom without searching through everything.

Before vs After

✗ Before

sendMessage(topic, message)

✓ After

sendMessage(topic, key, value)

What It Enables

It enables fast, reliable message delivery and easy grouping or ordering based on keys.

Real Life Example

Think of a bank sending transaction records. The account number is the key, and the transaction details are the value. This way, all transactions for one account go to the same place, making tracking simple and accurate.

Key Takeaways

Manual message handling is slow and error-prone.

Message keys help organize and route messages efficiently.

Keys and values together make data delivery reliable and easy to manage.

Practice

1. What is the primary role of the message key in Kafka?

easy

A. To decide which partition the message goes to

B. To store the actual data of the message

C. To encrypt the message content

D. To delete old messages automatically

Solution

1. Step 1: Understand message key purpose

   The message key in Kafka is used to determine the partition where the message will be stored.

2. Step 2: Differentiate key and value roles

   The message value holds the actual data, while the key helps with partitioning and ordering.

3. Final Answer:

   To decide which partition the message goes to -> Option A

4. Quick Check:

   Message key = Partition decision [OK]

Hint: Key decides partition, value holds data [OK]

Common Mistakes:

• Confusing key with message value
• Thinking key encrypts data
• Assuming key deletes messages

2. Which of the following is the correct way to send a Kafka message with a key and value in Java?

easy

A. producer.send(new ProducerRecord<>("topic", value));

B. producer.send(new ProducerRecord<>(key, value));

C. producer.send("topic", value);

D. producer.send(new ProducerRecord<>("topic", key, value));

Solution

1. Step 1: Review ProducerRecord constructor

   The correct constructor for sending a message with key and value includes topic, key, and value in that order.

2. Step 2: Check each option

   producer.send(new ProducerRecord<>("topic", key, value)); uses new ProducerRecord<>("topic", key, value), which is correct syntax. Others miss topic or key.

3. Final Answer:

   producer.send(new ProducerRecord<>("topic", key, value)); -> Option D

4. Quick Check:

   ProducerRecord(topic, key, value) = correct [OK]

Hint: ProducerRecord needs topic, key, then value [OK]

Common Mistakes:

• Omitting the topic parameter
• Swapping key and value order
• Using wrong constructor overload

3. Given the following Kafka producer code snippet, what will be the key of the message sent?

ProducerRecord record = new ProducerRecord<>("orders", "12345", "order details");
producer.send(record);

medium

A. "orders"

B. null

C. "12345"

D. "order details"

Solution

1. Step 1: Identify parameters in ProducerRecord

   The constructor is ProducerRecord(topic, key, value). Here, "orders" is topic, "12345" is key, and "order details" is value.

2. Step 2: Determine the key value

   The key is the second argument, which is "12345".

3. Final Answer:

   "12345" -> Option C

4. Quick Check:

   Second argument = key = "12345" [OK]

Hint: Second argument in ProducerRecord is key [OK]

Common Mistakes:

• Confusing topic with key
• Choosing message value as key
• Assuming key is null if not obvious

4. Identify the error in this Kafka producer code snippet:

ProducerRecord record = new ProducerRecord<>("topic", null, "data");
producer.send(record);

medium

A. Message key cannot be null

B. This code is valid; no error

C. ProducerRecord constructor parameters are in wrong order

D. Message value cannot be null

Solution

1. Step 1: Check if null key is allowed

   Kafka allows message keys to be null; this means messages can be sent without a key.

2. Step 2: Verify constructor usage

   The constructor ProducerRecord<>("topic", null, "data") is valid and sends a message with no key.

3. Final Answer:

   This code is valid; no error -> Option B

4. Quick Check:

   Null key allowed in Kafka messages [OK]

Hint: Null keys are allowed in Kafka messages [OK]

Common Mistakes:

• Assuming key must never be null
• Confusing parameter order
• Thinking value cannot be null without checking

5. You want to ensure all messages with the same user ID go to the same partition for ordering. Which Kafka message feature should you use and how?

hard

A. Use the user ID as the message key to ensure partitioning by key

B. Include the user ID only in the message value and leave key null

C. Set the user ID as the topic name

D. Send messages without keys and rely on default partitioning

Solution

1. Step 1: Understand partitioning by key

   Kafka uses the message key to decide the partition. Messages with the same key go to the same partition.

2. Step 2: Apply to user ID scenario

   Using the user ID as the message key ensures all messages for that user go to the same partition, preserving order.

3. Final Answer:

   Use the user ID as the message key to ensure partitioning by key -> Option A

4. Quick Check:

   Key = user ID for partitioning [OK]

Hint: Use key for grouping messages by user ID [OK]

Common Mistakes:

• Putting user ID only in value loses partition control
• Changing topic per user is inefficient
• Relying on default partitioning breaks ordering