microservices-demo

How I implemented the sample cloud-native application with 11 microservices showcasing Kubernetes and managed by helm. Original repo; https://github.com/GoogleCloudPlatform/microservices-demo

Must have for this to run; minikube, kubectl, docker, helm, helmfile. Then following these steps;

• Fork the repo
• Clone the repo into you local machine
• cd into the directory
• Run the command; helmfile sync
• Check if all the service were deploy correctly, run the command; kubectl get pod
• Get the url address, first run the command; kubectl get svc
• Copy the frontendservice name, then run the command; minikube service frontendservice
• Copy and paste the url address on the browser. Voila!!!

Table of contents

• A journey into microservice
  • Problem Statements
  • Solution
  • Project requirements
  • Demo
    • Prepare kubernetes environment and get helm install
    • Create a common chart for 10 microservices
    • Set default and actual value for each microservices
    • Create redis helm chart
    • Deploy microservices to kubernetes with helmfile
  • Lesson Learnt

Problem Statements

Problem Statements: Problem Statements How can I deploy and manage existing microservices in a kubernetes cluster?

Solution

The diagram describes my solution in picture as shared in the original repo.

Project requirements

It is assume that this demo is for learning purpose, so best production and security practices are not fully considered:

• I will deploy 11 microservices with helmfile
• Frontend service - entrypoint for the application that is accessible externally
• Redis - 3rd party service that store data
• All the microservices images, env, port are based on the original repo from gcr sample
• I use a single namespace
• All the microservices commmunicate using API call

Demo

If you want to follow along this demo, fork the repo. Clone the repo into your local machine. Change into the directory where it was clone to.

Prepare kubernetes environment and get helm install

• Set up minikube on your machine and get helm, helmfile install.
• Pretty simple process; check out the steps from official doc at https://kubernetes.io/docs/tasks/tools/ and https://helm.sh/docs/intro/install/

Create a common chart for 10 microservices

• I created a common chart for the 10 microservices of the application and later a chart for the 3rd party service, in this case redis. Run the command; helm create common-chart
• At the creation, the chart will contain the following;
  • chart folder - chart dependencies
  • template folder - dir where actual k8s YAML files created, blueprint
  • Chart.yaml - meta info about the chart
  • values.yaml - default values for the template files
  • .helmignore - files need not include in the helm chart

Set default and actual value for each microservices

• Need to validate if all the values define is correct up till this point. Run the command; helm template -f email-service-values.yaml common-chart & helm lint -f email-service-value.yaml common-chart
• Then, deploy one of the service to check if this work. Run the command; helm install common-chart ./helm-charts/common-chart/ & helm list

Create redis helm chart

• Same process as when creating common charts but this time for just redis. Run the command; helm create redis
• Put this redis chart and common chart under same folder - helm-charts

Deploy microservices to kubernetes with helmfile

• Alternative to helmfile is to individually install microservices by running the command; helm install -f file.yaml release_name chart_name for each services. Imagine doing this for hundred of services, not practical right.
• Create helmfile.yaml which define the releases for all the microservices.
• First install helmfile into the machine, I am using macOS, so I ran; brew install helmfile
• Then run the command; helmfile sync the result should be similar to image below
• Check with kubectl if all the pods are created, running with kubectl get pod
• Then check the application on the browser through frontend service which serve as entry point
• By runnning kubectl get svc to get the address expose externally. Remember I use NodePort as serviceType.
• I copied the frontendservice name, then run the command; minikube service frontendservice
• Then again copy and paste the address on the browser. It should display similar image below if all things went well

Lesson Learnt

• Helm makes managing microservices more ease, smooth and time-saving
• One will appreciate helm better if you had previously creating multiple deployment manifest for kubernetes cluster manually
• Alternative to helm is carvel ytt, more of programmatic approach to packaging and managing microservices
• There is no one-fit-all kind of solution to problems. It just use case understanding