Skip to content

Latest commit

 

History

History
108 lines (75 loc) · 3.48 KB

04-topic.adoc

File metadata and controls

108 lines (75 loc) · 3.48 KB
Raw

Kafka Topic

These labs pretend to introduce about the topic concept in Kafka.

Prerequisites:

  • AMQ Streams operator installed

Goals:

  • Kafka topic concepts

  • How to create a topic

Kafka cluster works as a group of brokers.

The data is stored in Topics. Each topic has one or many partitions where the data is distributed.

A partition has a data subset of the topic. Thanks to this, the data processing can be parallelized.

Another important point is the replication factor. When a topic is divided in partitions, we earn the ability of parallelization, but we need high availability in most cases. With the replication factor, we indicate to Kafka how many times this partition should be replicated along the brokers. These replications are not productive, they are followers of the main that only work in case of fail in the partition leader.

How to Kafka operate

To deploy a topic we have to apply a KafkaTopic descriptor. The descriptor shows like:

link:https://raw.githubusercontent.com/dbgjerez/workshop-amq-streams/master/content/components/kafka-topic/kafka-simple-topic.yaml[role=include]

We can see a lot of important configuration points:

  • config.retention.ms: represents the time before the data is deleted

  • config.segment.bytes: controls the segment file size for the log

  • partitions: number of partitions of the topic

  • replicas: number of replicas in each partition

  • topicName: name of the topic when the client connects inside the cluster

To deploy the topic is needed a cluster. We’re going to reuse the cluster used to explain the Kafka clusters.

We recreate it:

oc apply -f content/components/kafka-cluster/kafka-cluster-demo.yaml -n $OCP_NS

We can wait while the operator is starting up:

oc get kafka -w
NAME                 DESIRED KAFKA REPLICAS   DESIRED ZK REPLICAS   READY   WARNINGS
kafka-cluster-demo   3                        3
kafka-cluster-demo   3                        3                     True

Now, once the cluster is up, we can deploy the topic:

oc apply -f content/components/kafka-topic/kafka-simple-topic.yaml -n $OCP_NS

We can ask for it with the following command:

oc get kafkatopic -n $OCP_NS simple-topic-demo
NAME                CLUSTER              PARTITIONS   REPLICATION FACTOR   READY
simple-topic-demo   kafka-cluster-demo   1            2                    True

Now, the topic is ready to receive and send the data.

If we enter inside a Kafka broker container, we can navigate to the filesystem when the topic is created and his data is stored.

To list the brokers:

oc get po -n $OCP_NS | grep demo-kafka | awk '{print $1}'
kafka-cluster-demo-kafka-0
kafka-cluster-demo-kafka-1
kafka-cluster-demo-kafka-2

To enter inside and list the filesystem:

oc exec -it -n $OCP_NS kafka-cluster-demo-kafka-0 ls /var/lib/kafka/data/kafka-log0/

As we indicated the replica value=2 and the partition=1, we can see the topic only in 2 brokers.

At this point, we have examined the topic and we don’t need it, so we can delete it:

oc delete kafkatopic simple-topic-demo -n $OCP_NS
kafkatopic.kafka.strimzi.io "simple-topic-demo" deleted

And also the cluster:

oc delete kafka kafka-cluster-demo -n $OCP_NS
kafka.kafka.strimzi.io "kafka-cluster-demo" deleted