Supertubes

Benchmarking Kafka

How to setup the environment for the Kafka Performance Test using Amazon PKE, GKE, EKS.

We are going to use Banzai Cloud CLI to create the cluster:

brew install banzaicloud/tap/banzai-cli
banzai login

PKE 🔗︎

Create your own VPC and subnets on Amazon Management Console.
1. Use the provided wizard and select VPC with Single Public Subnet. (Please remember the Availability Zone you chose.)
2. Save the used route table id on the generated subnet
3. Create two additional subnet in the VPC (choose different Availability Zones)
  - Modify your newly created subnet Auto Assign IP setting
  - Enable auto-assign public IPV4 address
4. Assign the saved route table id to the two additional subnets
  - On Route Table page click Actions and Edit subnet associations
Create the cluster itself.
```
banzai cluster create
```
The required cluster template file can be found here

Please don’t forget to fill out the template with the created ids.

This will create a cluster with 3 nodes for ZK 3 for Kafka 1 Master node and 2 node for clients.

Create a StorageClass which enables high performance disk requests.

kubectl create -f - <<EOF
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: fast-ssd
provisioner: kubernetes.io/aws-ebs
parameters:
  type: io1
  iopsPerGB: "50"
  fsType: ext4
volumeBindingMode: WaitForFirstConsumer
EOF

GKE 🔗︎

Create the cluster itself:
```
banzai cluster create
```
The required cluster template file can be found here

Please don’t forget to fill out the template with the created ids.

Create a StorageClass which enables high performance disk requests.

kubectl create -f - <<EOF
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: fast-ssd
provisioner: kubernetes.io/gce-pd
parameters:
  type: pd-ssd
volumeBindingMode: WaitForFirstConsumer
EOF

EKS 🔗︎

Create the cluster itself:
```
banzai cluster create
```
The required cluster template file can be found here

Please don’t forget to fill out the template with the created ids.

Once your cluster is up and running we can move on to set up the Kubernetes infrastructure.

Create a StorageClass which enables high performance disk requests.

kubectl create -f - <<EOF
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: fast-ssd
provisioner: kubernetes.io/aws-ebs
parameters:
  type: io1
  iopsPerGB: "50"
  fsType: ext4
volumeBindingMode: WaitForFirstConsumer
EOF

Install other required components 🔗︎

Create a Zookeeper cluster with 3 replicas using Pravega’s Zookeeper Operator.

helm repo add banzaicloud-stable https://kubernetes-charts.banzaicloud.com/
helm install --name zookeeper-operator --namespace zookeeper banzaicloud-stable/zookeeper-operator
kubectl create -f - <<EOF
apiVersion: zookeeper.pravega.io/v1beta1
kind: ZookeeperCluster
metadata:
  name: example-zookeepercluster
  namespace: zookeeper
spec:
  replicas: 1
EOF

Install the latest version of Koperator , the Operator for managing Apache Kafka on Kubernetes.
```
helm install --name=kafka-operator banzaicloud-stable/kafka-operator
```
Create a 3 broker Kafka Cluster using the provided yaml.

This will install 3 brokers partitioned to three different zone with fast ssd.

Create a client container inside the cluster

kubectl create -f - <<EOF
apiVersion: v1
kind: Pod
metadata:
  annotations:
    linkerd.io/inject: enabled
  name: kafka-test
spec:
  containers:
  - name: kafka-test
    image: "wurstmeister/kafka:2.12-2.1.1"
    # Just spin & wait forever
    command: [ "/bin/bash", "-c", "--" ]
    args: [ "while true; do sleep 3000; done;" ]
EOF

Exec into this client and create the perftest, perftest2, perftes3 topics.

kubectl exec -it kafka-test bash
./opt/kafka/bin/kafka-topics.sh --zookeeper zookeeper-client.zookeeper:2181 --topic perftest --create --replication-factor 3 --partitions 3
./opt/kafka/bin/kafka-topics.sh --zookeeper zookeeper-client.zookeeper:2181 --topic perftest2 --create --replication-factor 3 --partitions 3
./opt/kafka/bin/kafka-topics.sh --zookeeper zookeeper-client.zookeeper:2181 --topic perftest3 --create --replication-factor 3 --partitions 3

Monitoring environment is automatically installed. To monitor the infrastructure we used the official Node Exporter dashboard available with id 1860.

Run the tests 🔗︎

Run perf test against the cluster, by building the provided Docker image

docker build -t yourname/perfload:0.1.0 /loadgens
docker push yourname/perfload:0.1.0

Submit the perf test application:

kubectl create -f - <<EOF
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  labels:
    app: loadtest
  annotations:
    linkerd.io/inject: enabled
  name: perf-load
  namespace: default
spec:
  progressDeadlineSeconds: 600
  replicas: 4
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      app: loadtest
  strategy:
    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%
    type: RollingUpdate
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: loadtest
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: nodepool.banzaicloud.io/name
                operator: In
                values:
                - pool3
      containers:
      - args:
        - -brokers=kafka-0:29092,kafka-1:29092,kafka-2:29092
        - -topic=perftest
        - -required-acks=all
        - -message-size=512
        - -workers=20
        image: yourorg/yourimage:yourtag
        imagePullPolicy: Always
        name: sangrenel
        resources: {}
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30
---

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  labels:
    app: loadtest
  name: perf-load
  namespace: default
spec:
  progressDeadlineSeconds: 600
  replicas: 4
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      app: loadtest
  strategy:
    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%
    type: RollingUpdate
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: loadtest
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: nodepool.banzaicloud.io/name
                operator: In
                values:
                - clients
      containers:
      - args:
        - -brokers=kafka-0,kafka-1,kafka-2:29092
        - -topic=perftest2
        - -required-acks=all
        - -message-size=512
        - -workers=20
        image: yourorg/yourimage:yourtag
        imagePullPolicy: Always
        name: sangrenel
        resources: {}
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30
---

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  labels:
    app: loadtest
  name: perf-load
  namespace: default
spec:
  progressDeadlineSeconds: 600
  replicas: 4
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      app: loadtest
  strategy:
    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%
    type: RollingUpdate
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: loadtest
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: nodepool.banzaicloud.io/name
                operator: In
                values:
                - clients
      containers:
      - args:
        - -brokers=kafka-0,kafka-1,kafka-2:29092
        - -topic=perftest
        - -required-acks=all
        - -message-size=512
        - -workers=20
        image: yourorg/yourimage:yourtag
        imagePullPolicy: Always
        name: sangrenel
        resources: {}
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30
EOF

Edit this page on GitHub