Cloud Native Go

1. Cloud Native Go with Azure Kubernetes Service (AKS), KEDA + KEDA HTTP Addon

Walkthrough (vimeo.com)

In this lab you will deploy an Azure Kubernetes Service (AKS) cluster and other Azure services (Container Registry, Managed Identity, Storage Account, Service Bus, Key Vault), the open source KEDA (Kubernetes Event-driven Autoscaling) project, the KEDA HTTP Addon (kedacore/http-add-on) with Azure CLI and Bicep. You will then deploy a sample Go application, go-hello, to your cluster using kubectl and kustomization.

Requirements

• An Azure Subscription (e.g. Free or Student account)
• The Azure CLI
• Bash shell (e.g. macOS, Linux, Windows Subsystem for Linux (WSL), Multipass, Azure Cloud Shell, GitHub Codespaces, etc)
• A GitHub Account
• Recommended: hey for load testing, and k9s to monitor your cluster during testing

Instructions

1. Use the Azure CLI and Bicep templates to deploy the infrastructure for your application.

   # login to the azure cli
   az login
   
   # if you need kubectl, install it
   az aks install-cli
   
   # 01-aks/
   cd 01-aks
   bash deploy-main.sh

   You can optionally deploy an Azure Key Vault. This template is deployed in incremental mode.

   # 01-keyvault/
   cd 01-keyvault
   bash deploy-group.sh

2. Use another Bicep template to run a Deployment Script that runs inside a container and installs the following on your AKS cluster.

   # 01-deploy-script/
   cd 01-deploy-script
   bash deploy-group.sh

3. (Optional) Step 2 uses the bash scripts in 02-keda-http to install the dependencies. If you prefer, you can run these manually at any time via up.sh or down.sh.

   # 02-keda-http/
   cd keda-http
   
   # install
   bash up.sh
   
   # uninstall
   bash down.sh
   
   # authenticate
   bash auth.sh

4. Use the kubectl CLI tool and kubectl apply -k (via kustomize) to deploy a pre-built container hello container image of our go-hello sample application.

   # 03-app/
   cd 03-app
   
   # (optional) run "az aks get-credentials" using the script mentioned in step 3.
   # bash ../02-keda-http/auth.sh
   
   # base
   kubectl apply -k ./base
   
   # keda-http
   # edit: my-image/kustomization.yaml
   # replace "host: 52.151.224.146" with the ip of your ingress controller
   kubectl apply -k ./keda-http
   
   # my-image
   # edit: my-image/kustomization.yaml
   # replace "newName: acrjxasx2.azurecr.io/asw101/test" with the name of your image, and "newTag" with the tag of your image, as applicable
   kubectl apply -k ./my-image

   You may now test and/or load test your application using curl or hey, and monitor it using kubectl or k9s. See 03-app/README.md for sample snippets.

5. Build the go-hello sample app, using a distroless container image, and deploy it from your own container registry.

   # go-hello/
   cd go-hello
   
   # build locally
   bash do/docker-build.sh
   
   # build locally using buildx (e.g. Apple Silicon/M1)
   bash do/docker-buildx.sh
   
   # login to github container registry
   bash do/ghcr-login.sh
   
   # push to github container registry, replace 'asw101' with your GH user
   export OWNER='asw101'
   export IMAGE='test'
   export TAG='latest'
   
   bash do/ghcr-push.sh
   
   # build with azure container registry
   bash do/acr-build.sh

What's next?

Try our Cloud Native labs in Azure Open Source Labs (aka.ms/oss-labs) for Serverless Containers, KEDA, Dapr and more.