Skip to content

Getting started

edited Jun 2 at 6:06

Kafka is a high throughput publish-subscribe messaging system implemented as distributed, partitioned, replicated commit log service.

Taken from official Kafka site


A single Kafka broker can handle hundreds of megabytes of reads and writes per second from thousands of clients.


Kafka is designed to allow a single cluster to serve as the central data backbone for a large organization. It can be elastically and transparently expanded without downtime. Data streams are partitioned and spread over a cluster of machines to allow data streams larger than the capability of any single machine and to allow clusters of co-ordinated consumers


Messages are persisted on disk and replicated within the cluster to prevent data loss. Each broker can handle terabytes of messages without performance impact.

Distributed by Design

Kafka has a modern cluster-centric design that offers strong durability and fault-tolerance guarantees.


Apache Kafka™ is a distributed streaming platform.

Which means:

  1. It lets you publish and subscribe to streams of records. In this respect it is similar to a message queue or enterprise messaging system.
  2. It lets you store streams of records in a fault-tolerant way.
  3. It lets you process streams of records as they occur.

It gets used for two broad classes of application:

  1. Building real-time streaming data pipelines that reliably get data between systems or applications
  2. Building real-time streaming applications that transform or react to the streams of data

Kafka console scripts are different for Unix-based and Windows platforms. In the examples, you might need to add the extension according to your platform. Linux: scripts located in bin/ with .sh extension. Windows: scripts located in bin\windows\ and with .bat extension.


Step 1: Download the code and untar it:

tar -xzf kafka_2.11-
cd kafka_2.11-

Step 2: start the server.

to be able to delete topics later, open and set delete.topic.enable to true.

Kafka relies heavily on zookeeper, so you need to start it first. If you don't have it installed, you can use the convenience script packaged with kafka to get a quick-and-dirty single-node ZooKeeper instance.

zookeeper-server-start config/
kafka-server-start config/

Step 3: ensure everything is running

You should now have zookeeper listening to localhost:2181 and a single kafka broker on localhost:6667.

Quick Start (Single mode)

Create a topic

We only have one broker, so we create a topic with no replication factor and just one partition:

kafka-topics --zookeeper localhost:2181 \
    --create \
    --replication-factor 1 \
    --partitions 1 \
    --topic test-topic

Check your topic:

kafka-topics --zookeeper localhost:2181 --list 

kafka-topics --zookeeper localhost:2181 --describe --topic test-topic
Topic:test-topic    PartitionCount:1    ReplicationFactor:1 Configs:
Topic: test-topic   Partition: 0    Leader: 0   Replicas: 0 Isr: 0

send and receive messages

Launch a consumer:

kafka-console-consumer --bootstrap-server localhost:9092 --topic test-topic

On another terminal, launch a producer and send some messages. By default, the tool send each line as a separate message to the broker, without special encoding. Write some lines and exit with CTRL+D or CTRL+C:

kafka-console-producer --broker-list localhost:9092 --topic test-topic   
a message
another message

The messages should appear in the consumer therminal.

Stop kafka


Cluster mode

Start a multi-broker cluster

The above examples use only one broker. To setup a real cluster, we just need to start more than one kafka server. They will automatically coordinate themselves.

Step 1: to avoid collision, we create a file for each broker and change the id, port and logfile configuration properties.


cp config/ config/
cp config/ config/

Edit properties for each file, for example:

vim config/

vim config/

Step 2: start the three brokers:

kafka-server-start config/ &
kafka-server-start config/ &
kafka-server-start config/ &

Create a replicated topic

kafka-topics --zookeeper localhost:2181 --create --replication-factor 3 --partitions 1 --topic replicated-topic

kafka-topics --zookeeper localhost:2181 --describe --topic replicated-topic
Topic:replicated-topic  PartitionCount:1    ReplicationFactor:3 Configs:
Topic: replicated-topic Partition: 0    Leader: 1   Replicas: 1,2,0 Isr: 1,2,0

This time, there are more information:

  • "leader" is the node responsible for all reads and writes for the given partition. Each node will be the leader for a randomly selected portion of the partitions.
  • "replicas" is the list of nodes that replicate the log for this partition regardless of whether they are the leader or even if they are currently alive.
  • "isr" is the set of "in-sync" replicas. This is the subset of the replicas list that is currently alive and caught-up to the leader.

Note that the previously created topic is left unchanged.

Test fault tolerance

Publish some message to the new topic:

kafka-console-producer --broker-list localhost:9092 --topic replicated-topic
hello 1
hello 2

Kill the leader (1 in our example).

On Linux:

ps aux | grep
kill -9 <PID>

On Windows:

wmic process get processid,caption,commandline | find "java.exe" | find "" 
taskkill /pid <PID> /f

See what happened:

kafka-topics --zookeeper localhost:2181  --describe --topic replicated-topic
Topic:replicated-topic  PartitionCount:1    ReplicationFactor:3 Configs:
Topic: replicated-topic Partition: 0    Leader: 2   Replicas: 1,2,0 Isr: 2,0

The leadership has switched to broker 2 and "1" in not in-sync anymore. But the messages are still there (use the consumer to check out by yourself).


Delete the two topics using:

kafka-topics --zookeeper localhost:2181 --delete --topic test-topic
kafka-topics --zookeeper localhost:2181 --delete --topic replicated-topic