/gcp-golang-performance-test

Performance testing on GKE using Go echo application

Primary LanguageShell

Performance testing on GKE using labstack echo application

Build Quality Gate Status Lines of Code

Performance testing on GKE using the https://echo.labstack.com application.

Table of Contents

Prerequisites

Installation

Set environment variables

COMPUTE_ZONE="us-central1"
# replace with your project
PROJECT_ID="sample-project"

Set GCP project

gcloud config set project ${PROJECT_ID}
gcloud config set compute/zone ${COMPUTE_ZONE}

1. Create a GKE cluster

Create an Autopilot GKE cluster. It may take around 9 minutes.

gcloud container clusters create-auto sample-cluster --region=${COMPUTE_ZONE}
gcloud container clusters get-credentials sample-cluster

2. Deploy two applications for checking the performance per Pod and scaling

Build and push to GCR:

cd app
docker build -t go-echo-api . --platform linux/amd64
docker tag go-echo-api:latest gcr.io/${PROJECT_ID}/go-echo-api:latest

gcloud auth configure-docker
docker push gcr.io/${PROJECT_ID}/go-echo-api:latest
kubectl get namespaces

kubectl create namespace echo-test

Two deployments may take around 5 minutes to create a load balancer, including health checking.

2.1. Deploy for performance of one Pod

To check request per seconds(RPS) WITHOUT scaling, create and deploy K8s Deployment, Service, HorizontalPodAutoscaler, Ingress, and GKE BackendConfig using the go-echo-api-onepod-template.yaml template file:

sed -e "s|<project-id>|${PROJECT_ID}|g" go-echo-api-onepod-template.yaml > go-echo-api-onepod.yaml
cat go-echo-api-onepod.yaml

kubectl get namespaces
kubectl apply -f go-echo-api-onepod.yaml -n echo-test --dry-run=client

Confirm Pod logs and configuration after deployment:

kubectl logs -l app=go-echo-api-onepod -n echo-test

kubectl describe pods -n echo-test

kubectl get ingress go-echo-api-onepod-ingress -n echo-test

2.2. Deploy for Scaling Test

To check request per seconds(RPS) with scaling, create and deploy K8s Deployment, Service, HorizontalPodAutoscaler, Ingress, and GKE BackendConfig using the go-echo-api-template.yaml template file:

sed -e "s|<project-id>|${PROJECT_ID}|g" go-echo-api-template.yaml > go-echo-api.yaml
cat go-echo-api.yaml

kubectl apply -f go-echo-api.yaml -n echo-test --dry-run=client
kubectl apply -f go-echo-api.yaml -n echo-test

Confirm Pod logs and configuration after deployment:

kubectl logs -l app=go-echo-api -n echo-test

kubectl describe pods -n echo-test

kubectl get ingress go-echo-api-ingress -n echo-test

Confirm that response of / API.

LB_IP_ADDRESS=$(gcloud compute forwarding-rules list | grep go-echo-api | awk '{ print $2 }')
echo ${LB_IP_ADDRESS}
curl http://${LB_IP_ADDRESS}/

3. Performance Testing

3.1. Install the Taurus

https://gettaurus.org/install/Installation/

sudo apt-get update -y
sudo apt-get install python3 default-jre-headless python3-tk python3-pip python3-dev libxml2-dev libxslt-dev zlib1g-dev net-tools  -y
sudo python3 -m pip install bzt
sudo apt-get install htop -y

3.2. Test for performance of one Pod

cd test
# test with 300 threads and connection:close option
bzt echo-bzt-onepod.yaml

test/echo-bzt-onepod.yaml

kubectl describe hpa go-echo-api-onepod-hpa -n echo-test

kubectl get hpa go-echo-api-onepod-hpa -n echo-test -w

3.3. Test with auto scaling

cd test
# test with 2000 threads and connection:close option
bzt echo-bzt.yaml

test/echo-bzt.yaml

kubectl describe hpa go-echo-api-hpa -n echo-test

kubectl get hpa go-echo-api-hpa -n echo-test -w

Cleanup

kubectl scale deployment go-echo-api-onepod -n echo-test --replicas=0
kubectl scale deployment go-echo-api -n echo-test --replicas=0

kubectl delete -f app/go-echo-api-onepod.yaml -n echo-test
kubectl delete -f app/go-echo-api.yaml -n echo-test

References