Custom SerDes in Kafka - Step-by-Step Execution

Process Flow - Custom SerDes

Start

↓

Serialize Object

↓

Send Bytes to Kafka

↓

Receive Bytes from Kafka

↓

Deserialize Bytes to Object

↓

Use Object in Application

↓

End

This flow shows how a custom serializer converts an object to bytes before sending to Kafka, and a custom deserializer converts bytes back to the object after receiving.

Execution Sample

Kafka

public class UserSerializer implements Serializer<User> {
  @Override
  public byte[] serialize(String topic, User data) {
    return (data.getName() + "," + data.getAge()).getBytes();
  }
}

public class UserDeserializer implements Deserializer<User> {
  @Override
  public User deserialize(String topic, byte[] data) {
    String[] parts = new String(data).split(",");
    return new User(parts[0], Integer.parseInt(parts[1]));
  }
}

This code serializes a User object to bytes by joining name and age with a comma, and deserializes bytes back to a User object by splitting the string.

Process Table

Step	Action	Input	Output	Notes
1	Serialize User object	User(name=Alice, age=30)	byte[] representing "Alice,30"	Convert object fields to string, then to bytes
2	Send bytes to Kafka	byte[] "Alice,30"	Sent to Kafka topic	Bytes are transmitted over Kafka
3	Receive bytes from Kafka	byte[] "Alice,30"	byte[] "Alice,30"	Bytes received unchanged
4	Deserialize bytes	byte[] "Alice,30"	User(name=Alice, age=30)	Split string and parse age to int
5	Use User object	User(name=Alice, age=30)	User object ready	Application can now use the User object
6	End	-	-	Process complete

💡 Deserialization completes successfully when bytes are converted back to the original User object.

Status Tracker

Variable	Start	After Serialize	After Send	After Receive	After Deserialize	Final
user	User(name=Alice, age=30)	User(name=Alice, age=30)	User(name=Alice, age=30)	User(name=Alice, age=30)	User(name=Alice, age=30)	User(name=Alice, age=30)
bytes	null	byte[] "Alice,30"	byte[] "Alice,30"	byte[] "Alice,30"	null	null
deserializedUser	null	null	null	null	User(name=Alice, age=30)	User(name=Alice, age=30)

Key Moments - 3 Insights

Why do we convert the User object to a string before bytes?

Because Kafka sends data as bytes, we must convert the object to a byte array. Converting to a string first makes it easy to join fields and then get bytes, as shown in step 1 of the execution_table.

What happens if the deserializer receives corrupted bytes?

If bytes are corrupted, splitting or parsing may fail or produce wrong data. The deserializer expects the format shown in step 4. If format differs, errors or wrong objects occur.

Why do we need both serializer and deserializer?

Serializer converts objects to bytes for Kafka to send. Deserializer converts bytes back to objects for the application to use. Both must agree on the format, as seen in steps 1 and 4.

Visual Quiz - 3 Questions

Look at the execution_table, what is the output of step 1 (Serialize User object)?

A String "Alice,30"

B User object with name and age

C byte[] representing "Alice,30"

D null

💡 Hint

Check the Output column in step 1 of the execution_table.

At which step does the data change from bytes back to a User object?

A Step 4

B Step 2

C Step 3

D Step 5

💡 Hint

Look for the step where deserialize is called in the execution_table.

If the serializer joined fields with a semicolon instead of a comma, what would change in the execution_table?

A Step 5 would produce a different User object

B Step 1 output would have semicolon instead of comma

C Step 4 deserialization would fail without changes

D No changes needed anywhere

💡 Hint

Check the string format in step 1 output and how step 4 splits the string.

Concept Snapshot

Custom SerDes in Kafka:
- Serializer converts objects to byte arrays before sending.
- Deserializer converts byte arrays back to objects after receiving.
- Both must agree on the data format (e.g., CSV string).
- Implement Serializer<T> and Deserializer<T> interfaces.
- Used to send complex objects through Kafka topics.

Full Transcript

This visual execution shows how custom serialization and deserialization work in Kafka. First, a User object is converted to a byte array by joining its fields as a string and getting bytes. These bytes are sent through Kafka. On the receiving side, the bytes are converted back to a string, split by comma, and parsed to recreate the User object. Variables like user, bytes, and deserializedUser change as the program runs. Key moments include why conversion to string is needed, what happens if bytes are corrupted, and why both serializer and deserializer are required. The quizzes test understanding of each step's input and output and the importance of matching formats. This process allows Kafka to handle complex data types by converting them to bytes and back.