What Problem Does Kafka Solve?

When a customer places an order on Zomato, multiple things need to happen at the same time:

• Payment service needs to charge the customer
• Restaurant service needs to notify the kitchen
• Notification service needs to send a confirmation SMS

Without Kafka, the Zomato app would call each of these services directly over HTTP/REST. If the payment service is slow, everything waits. If the notification service is down, data is lost.

Kafka solves this by decoupling producers from consumers using a message broker.

Core Components

Producer

• The component that sends (writes) data into Kafka
• In Zomato's case: the Zomato App, Driver App, REST API, IoT Sensors
• Producers don't care who reads the data — they just publish to a topic

Broker (Kafka Server)

• The heart of Kafka — stores and manages messages
• Contains Topics, which are named channels for messages
• Messages are persisted on disk — even if consumers are slow or offline
• Can be clustered (multiple brokers) for high availability

Topic

• A named stream of messages