Quickstart: Create Apache Kafka cluster in Azure HDInsight using Bicep

In this quickstart, you use a Bicep to create an Apache Kafka cluster in Azure HDInsight. Kafka is an open-source, distributed streaming platform. It's often used as a message broker, as it provides functionality similar to a publish-subscribe message queue.

Bicep is a domain-specific language (DSL) that uses declarative syntax to deploy Azure resources. It provides concise syntax, reliable type safety, and support for code reuse. Bicep offers the best authoring experience for your infrastructure-as-code solutions in Azure.

The Kafka API can only be accessed by resources inside the same virtual network. In this quickstart, you access the cluster directly using SSH. To connect other services, networks, or virtual machines to Kafka, you must first create a virtual network and then create the resources within the network. For more information, see the Connect to Apache Kafka using a virtual network document.

Prerequisites

If you don't have an Azure subscription, create a free account before you begin.

Review the Bicep file

The Bicep file used in this quickstart is from Azure Quickstart Templates.

@description('The name of the Kafka cluster to create. This must be a unique name.')
param clusterName string

@description('These credentials can be used to submit jobs to the cluster and to log into cluster dashboards.')
param clusterLoginUserName string

@description('The password must be at least 10 characters in length and must contain at least one digit, one upper case letter, one lower case letter, and one non-alphanumeric character except (single-quote, double-quote, backslash, right-bracket, full-stop). Also, the password must not contain 3 consecutive characters from the cluster username or SSH username.')
@minLength(10)
@secure()
param clusterLoginPassword string

@description('These credentials can be used to remotely access the cluster.')
param sshUserName string

@description('SSH password must be 6-72 characters long and must contain at least one digit, one upper case letter, and one lower case letter.  It must not contain any 3 consecutive characters from the cluster login name')
@minLength(6)
@maxLength(72)
@secure()
param sshPassword string

@description('Location for all resources.')
param location string = resourceGroup().location

@description('This is the headnode Azure Virtual Machine size, and will affect the cost. If you don\'t know, just leave the default value.')
@allowed([
  'Standard_A4_v2'
  'Standard_A8_v2'
  'Standard_E2_v3'
  'Standard_E4_v3'
  'Standard_E8_v3'
  'Standard_E16_v3'
  'Standard_E20_v3'
  'Standard_E32_v3'
  'Standard_E48_v3'
])
param HeadNodeVirtualMachineSize string = 'Standard_E4_v3'

@description('This is the worerdnode Azure Virtual Machine size, and will affect the cost. If you don\'t know, just leave the default value.')
@allowed([
  'Standard_A4_v2'
  'Standard_A8_v2'
  'Standard_E2_v3'
  'Standard_E4_v3'
  'Standard_E8_v3'
  'Standard_E16_v3'
  'Standard_E20_v3'
  'Standard_E32_v3'
  'Standard_E48_v3'
])
param WorkerNodeVirtualMachineSize string = 'Standard_E4_v3'

@description('This is the Zookeepernode Azure Virtual Machine size, and will affect the cost. If you don\'t know, just leave the default value.')
@allowed([
  'Standard_A4_v2'
  'Standard_A8_v2'
  'Standard_E2_v3'
  'Standard_E4_v3'
  'Standard_E8_v3'
  'Standard_E16_v3'
  'Standard_E20_v3'
  'Standard_E32_v3'
  'Standard_E48_v3'
])
param ZookeeperNodeVirtualMachineSize string = 'Standard_E4_v3'

var defaultStorageAccount = {
  name: uniqueString(resourceGroup().id)
  type: 'Standard_LRS'
}

resource storageAccount 'Microsoft.Storage/storageAccounts@2021-08-01' = {
  name: defaultStorageAccount.name
  location: location
  sku: {
    name: defaultStorageAccount.type
  }
  kind: 'StorageV2'
  properties: {}
}

resource cluster 'Microsoft.HDInsight/clusters@2021-06-01' = {
  name: clusterName
  location: location
  properties: {
    clusterVersion: '4.0'
    osType: 'Linux'
    clusterDefinition: {
      kind: 'kafka'
      configurations: {
        gateway: {
          'restAuthCredential.isEnabled': true
          'restAuthCredential.username': clusterLoginUserName
          'restAuthCredential.password': clusterLoginPassword
        }
      }
    }
    storageProfile: {
      storageaccounts: [
        {
          name: replace(replace(concat(reference(storageAccount.id, '2021-08-01').primaryEndpoints.blob), 'https:', ''), '/', '')
          isDefault: true
          container: clusterName
          key: listKeys(storageAccount.id, '2021-08-01').keys[0].value
        }
      ]
    }
    computeProfile: {
      roles: [
        {
          name: 'headnode'
          targetInstanceCount: 2
          hardwareProfile: {
            vmSize: HeadNodeVirtualMachineSize
          }
          osProfile: {
            linuxOperatingSystemProfile: {
              username: sshUserName
              password: sshPassword
            }
          }
        }
        {
          name: 'workernode'
          targetInstanceCount: 4
          hardwareProfile: {
            vmSize: WorkerNodeVirtualMachineSize
          }
          dataDisksGroups: [
            {
              disksPerNode: 2
            }
          ]
          osProfile: {
            linuxOperatingSystemProfile: {
              username: sshUserName
              password: sshPassword
            }
          }
        }
        {
          name: 'zookeepernode'
          targetInstanceCount: 3
          hardwareProfile: {
            vmSize: ZookeeperNodeVirtualMachineSize
          }
          osProfile: {
            linuxOperatingSystemProfile: {
              username: sshUserName
              password: sshPassword
            }
          }
        }
      ]
    }
  }
}

output cluster object = cluster.properties

Two Azure resources are defined in the Bicep file:

Deploy the Bicep file

  1. Save the Bicep file as main.bicep to your local computer.

  2. Deploy the Bicep file using either Azure CLI or Azure PowerShell.

    az group create --name exampleRG --location eastus
    az deployment group create --resource-group exampleRG --template-file main.bicep --parameters clusterName=<cluster-name> clusterLoginUserName=<cluster-username> sshUserName=<ssh-username>
    

    You need to provide values for the parameters:

    • Replace <cluster-name> with the name of the HDInsight cluster to create. The cluster name needs to start with a letter and can contain only lowercase letters, numbers, and dashes.
    • Replace <cluster-username> with the credentials used to submit jobs to the cluster and to log in to cluster dashboards. Uppercase letters aren't allowed in the cluster username.
    • Replace <ssh-username> with the credentials used to remotely access the cluster.

    You'll be prompted to enter the following:

    • clusterLoginPassword, which must be at least 10 characters long and contain at least one digit, one uppercase letter, one lowercase letter, and one non-alphanumeric character except single-quote, double-quote, backslash, right-bracket, full-stop. It also must not contain three consecutive characters from the cluster username of SSH username.
    • sshPassword, which must be 6-72 characters long and must contain at least one digit, one uppercase letter, and one lowercase letter. It must not contain any three consecutive characters from the cluster login name.

    Note

    When the deployment finishes, you should see a message indicating the deployment succeeded.

Review deployed resources

Use the Azure portal, Azure CLI, or Azure PowerShell to list the deployed resources in the resource group.

az resource list --resource-group exampleRG

Get the Apache Zookeeper and Broker host information

When working with Kafka, you must know the Apache Zookeeper and Broker hosts. These hosts are used with the Kafka API and many of the utilities that ship with Kafka.

In this section, you get the host information from the Ambari REST API on the cluster.

  1. Use ssh command to connect to your cluster. Edit the command below by replacing CLUSTERNAME with the name of your cluster, and then enter the command:

    ssh sshuser@CLUSTERNAME-ssh.azurehdinsight.net
    
  2. From the SSH connection, use the following command to install the jq utility. This utility is used to parse JSON documents, and is useful in retrieving the host information:

    sudo apt -y install jq
    
  3. To set an environment variable to the cluster name, use the following command:

    read -p "Enter the Kafka on HDInsight cluster name: " CLUSTERNAME
    

    When prompted, enter the name of the Kafka cluster.

  4. To set an environment variable with Zookeeper host information, use the command below. The command retrieves all Zookeeper hosts, then returns only the first two entries. This is because you want some redundancy in case one host is unreachable.

    export KAFKAZKHOSTS=`curl -sS -u admin -G https://$CLUSTERNAME.azurehdinsight.net/api/v1/clusters/$CLUSTERNAME/services/ZOOKEEPER/components/ZOOKEEPER_SERVER | jq -r '["\(.host_components[].HostRoles.host_name):2181"] | join(",")' | cut -d',' -f1,2`
    

    When prompted, enter the password for the cluster login account (not the SSH account).

  5. To verify that the environment variable is set correctly, use the following command:

     echo '$KAFKAZKHOSTS='$KAFKAZKHOSTS
    

    This command returns information similar to the following text:

    <zookeepername1>.eahjefxxp1netdbyklgqj5y1ud.ex.internal.cloudapp.net:2181,<zookeepername2>.eahjefxxp1netdbyklgqj5y1ud.ex.internal.cloudapp.net:2181

  6. To set an environment variable with Kafka broker host information, use the following command:

    export KAFKABROKERS=`curl -sS -u admin -G https://$CLUSTERNAME.azurehdinsight.net/api/v1/clusters/$CLUSTERNAME/services/KAFKA/components/KAFKA_BROKER | jq -r '["\(.host_components[].HostRoles.host_name):9092"] | join(",")' | cut -d',' -f1,2`
    

    When prompted, enter the password for the cluster login account (not the SSH account).

  7. To verify that the environment variable is set correctly, use the following command:

    echo '$KAFKABROKERS='$KAFKABROKERS
    

    This command returns information similar to the following text:

    <brokername1>.eahjefxxp1netdbyklgqj5y1ud.cx.internal.cloudapp.net:9092,<brokername2>.eahjefxxp1netdbyklgqj5y1ud.cx.internal.cloudapp.net:9092

Manage Apache Kafka topics

Kafka stores streams of data in topics. You can use the kafka-topics.sh utility to manage topics.

  • To create a topic, use the following command in the SSH connection:

    /usr/hdp/current/kafka-broker/bin/kafka-topics.sh --create --replication-factor 3 --partitions 8 --topic test --zookeeper $KAFKAZKHOSTS
    

    This command connects to Zookeeper using the host information stored in $KAFKAZKHOSTS. It then creates a Kafka topic named test.

    • Data stored in this topic is partitioned across eight partitions.

    • Each partition is replicated across three worker nodes in the cluster.

      If you created the cluster in an Azure region that provides three fault domains, use a replication factor of 3. Otherwise, use a replication factor of 4.

      In regions with three fault domains, a replication factor of 3 allows replicas to be spread across the fault domains. In regions with two fault domains, a replication factor of four spreads the replicas evenly across the domains.

      For information on the number of fault domains in a region, see the Availability of Linux virtual machines document.

      Kafka isn't aware of Azure fault domains. When creating partition replicas for topics, it may not distribute replicas properly for high availability.

      To ensure high availability, use the Apache Kafka partition rebalance tool. This tool must be ran from an SSH connection to the head node of your Kafka cluster.

      For the highest availability of your Kafka data, you should rebalance the partition replicas for your topic when:

      • You create a new topic or partition

      • You scale up a cluster

  • To list topics, use the following command:

    /usr/hdp/current/kafka-broker/bin/kafka-topics.sh --list --zookeeper $KAFKAZKHOSTS
    

    This command lists the topics available on the Kafka cluster.

  • To delete a topic, use the following command:

    /usr/hdp/current/kafka-broker/bin/kafka-topics.sh --delete --topic topicname --zookeeper $KAFKAZKHOSTS
    

    This command deletes the topic named topicname.

    Warning

    If you delete the test topic created earlier, then you must recreate it. It is used by steps later in this document.

For more information on the commands available with the kafka-topics.sh utility, use the following command:

/usr/hdp/current/kafka-broker/bin/kafka-topics.sh

Produce and consume records

Kafka stores records in topics. Records are produced by producers, and consumed by consumers. Producers and consumers communicate with the Kafka broker service. Each worker node in your HDInsight cluster is a Kafka broker host.

To store records into the test topic you created earlier, and then read them using a consumer, use the following steps:

  1. To write records to the topic, use the kafka-console-producer.sh utility from the SSH connection:

    /usr/hdp/current/kafka-broker/bin/kafka-console-producer.sh --broker-list $KAFKABROKERS --topic test
    

    After this command, you arrive at an empty line.

  2. Type a text message on the empty line and hit enter. Enter a few messages this way, and then use Ctrl + C to return to the normal prompt. Each line is sent as a separate record to the Kafka topic.

  3. To read records from the topic, use the kafka-console-consumer.sh utility from the SSH connection:

    /usr/hdp/current/kafka-broker/bin/kafka-console-consumer.sh --bootstrap-server $KAFKABROKERS --topic test --from-beginning
    

    This command retrieves the records from the topic and displays them. Using --from-beginning tells the consumer to start from the beginning of the stream, so all records are retrieved.

    If you're using an older version of Kafka, replace --bootstrap-server $KAFKABROKERS with --zookeeper $KAFKAZKHOSTS.

  4. Use Ctrl + C to stop the consumer.

You can also programmatically create producers and consumers. For an example of using this API, see the Apache Kafka Producer and Consumer API with HDInsight document.

Clean up resources

When no longer needed, use the Azure portal, Azure CLI, or Azure PowerShell to delete the resource group and its resources.

az group delete --name exampleRG

Next steps

In this quickstart, you learned how to create an Apache Kafka cluster in HDInsight using Bicep. In the next article, you learn how to create an application that uses the Apache Kafka Streams API and run it with Kafka on HDInsight.