We're live-coding on Twitch! Join us!
Google Cloud Platform I: Deploy a Docker App To Google Container Engine with Kubernetes

Google Cloud Platform I: Deploy a Docker App To Google Container Engine with Kubernetes

Code Project


Google Cloud Platform (GCP) is a cloud computing service by Google that offers a set of enterprise cloud services that leverages the same techologies that are used at Google.

Notable aspects of GCP are:

In this article, we will be deploying a containerized NodeJS application to GCP with Kubernetes, thanks to the platform's Google Container Engine.

What Is Kubernetes?

Kubernetes is an open-source platform for automating deployment, scaling, and operations of application containers across clusters of hosts, providing container-centric infrastructure.


Developed and powered by Google, Kubernetes provides an efficient way to manage Docker containers. Some it's provisions are:

  • Deploying Docker containers quickly and predictably.
  • Scaling your applications depending on the traffic needs on the fly.
  • Easily rolling out new features.

Some of the few things that make Kubernetes stand out are:

  • Automated rollouts and rollbacks - Easily roll new features to an existing application and rollback in case something goes wrong.
  • Horizontal Scaling - Scale up and down depending on your CPU usage.
  • Service discovery and load balancing - Use Kubernetes services to externally access containers and distribute the traffic to Kubernetes pods.
  • Secret and configuration management - Easy and secure management of sensitive data such as passwords, API keys and Auth tokens without risk of compromise.
  • Healthchecks perfomance. - Constantly perform healthchecks on your applications via HTTP to easily detect and remedy failures.
  • Resource Usage Monitoring

Enough chatter about what Kubernetes can do! The list is endless. Let's get our hands dirty with it.


To build and deploy the container, a background in the following is a plus but not necessary.

  • Docker - A lightweight containerization platform.
  • NodeJS - Javascript runtime built on Chrome's V8 JavaScript engine.
  • Kubernetes - An open-source Google managed platform for automating deployment, scaling, and management of containerized applications.

You will also need a Google Cloud Platform Subscription or Trial Account. Currently, Google offers a 60 day free subscription to all GCP services worth 300 dollars.

Sweet! Let's get started!

Installation and Setup


Go to the Docker download page and download the file that suites your Operating System and install.

If you are on a mac, Simply run

$ brew install docker


gcloud is a command-line tool for managing resources on Google Cloud Platform and is provided as part of Google Cloud SDK. To install, download the latest version here and extract it to your local system. Ensure that you have Python 2.7 installed.

Once downloaded, extract the folder and run the following command to install gcloud and add it to your path.

OS X/Linux users


Windows users


You can now authenticate gcloud with your Google account on your terminal by running the initialization command.

gcloud init

This will authenticate your gcloud environment, set a default project if any, set the default compute engine region (such as us-east1) and zone (such as us-central1-a).

To check your current configurations, run the configuration list command.

gcloud config list

gcloud config list


Kubernetes provides a number of ways to install the kubectl command used to interact with it on your local system. The easiest approach to setup is installing it as part of Google Cloud SDK.

Simply run

gcloud components install kubectl

To verify your installation, check the kubectl version on your command-line. You should get back a response with the version details.

If you need to install any other provisions of Google Cloud SDK, simply run the command gcloud components list

gcloud components list

Creating The GCP Project

Every Google project begins with creating a project on it's dashboard. Once you have your subscription or trial account ready, head over to https://console.cloud.google.com to create your first project.

Click on create new project and give it a memorable name. Take note of the project ID as we will be using it across the entire process.

Creating a new GCP project

This may take a while to finish. Once it's done, click on the Products & Services icon and select Container Engine which is where our Docker images will live.

Container Engine view

The Container clusters manages our Google Compute Engine instances also reffered to as nodes.

The Container Registry is a repository for all our Docker images from which containers will be created.

Preparing our Deployment Environment

Set default project

Once we have the gcloud and kubectl command-line tools installed, we can now set the project we created earlier as our default. We can easily do this with the config option.

gcloud config set project scotch-155622

Like we mentioned earlier, you can see a list of your configurations by running the gcloud config list command.


A cluster is a group of Google Compute Engine instances running Kubernetes. Pods and Services which we will be looking at later on, all run on top of a cluster.

Kubernetes coordinates a highly available cluster of computers that are connected to work as a single unit.

  • Creating a cluster

You can create a cluster on the Container Clusters section on the project dashboard by going to Container Engine > Container Clusters > New container cluster or via the command-line using the kubectl command. Simply run

gcloud container clusters create scotch-cluster

The cluster may take some time to complete. Once it's done, again, set it as the default cluster.

gcloud config set container/cluster scotch-cluster

You will notice that the project dashboard console will reflect our newly created cluster.

GCP Kubernetes cluster

  • Getting the cluster credentials.

We now have a Kubernetes cluster on Our GCP project that will contain our containerized application. Let's notify our local setup that we are authorized to use the cluster.

On clicking the connect button shown in the cluster above, you will be presented with the command to configure Kubernetes with the newly created scotch-cluster.

Configuring Kubernetes cluster

Run the command on your terminal and we are now ready to push our Docker container.

A Look At Our NodeJS App.

Our app is going to be very simple and straight forward. We are going to have a single index.js file that serves 3 routes using express.

We will look at Dockerfile, deployment.yml and service.yml files in later sections.

├── .gitignore
├── package.json
├── index.js
├── Dockerfile
├── deployment.yml
├── service.yml

Here is a snapshot of the index.js file.

const app = require('express')();

const port = process.env.PORT || 8000;

app.get('/', (req, res) => {
  res.status(200).json('welcome to root');

app.get('/foo', (req, res) => {
  res.status(200).json('welcome to foo');

app.get('/bar', (req, res) => {
  res.status(200).json('welcome to bar');

app.listen(port, () => console.log('Magic happens on port', port));

module.exports.app = app;

Pretty straightforward, right? Sweet!

Dockerizing Our Node App

Make sure you have Docker running on your local setup before proceeding. While we will not go to the details of how Docker works, take a look at this article on getting started with Docker.

Creating a Docker Image

To create an image for our application, we will use the docker build command which looks for a Dockerfile in the set directory.

Here is a look at our Dockerfile.

#Create our image from Node 6.9-alpine
FROM node:6.9-alpine

MAINTAINER John Kariuki <johnkariukin@gmail.com>

#Create a new directory to run our app.
RUN mkdir -p /usr/src/app

#Set the new directory as our working directory
WORKDIR /usr/src/app

#Copy all the content to the working directory
COPY . /usr/src/app

#install node packages to node_modules
RUN npm install

#Our app runs on port 8000. Expose it!

#Run the application.
CMD npm start

To create our Google Cloud platform-ready Docker image, we have to abide by some rules before it is pushed to the Google Container Registry. The image has to be created in the following format. gcr.io/{$project_id}/{image}:{tag}

$ docker build -t gcr.io/scotch-155622/node-app:0.0.1 .

We should be able to see the newly created image once it's done by running docker images.

Pushing The Docker Image to Container Registry

With our Docker image ready, pushing it to the Container Registry is simply done with the gcloud command.

gcloud docker -- push gcr.io/scotch-155622/node-app:0.0.1

This may take a while but once it's done, we should have the image listed on Compute > Container Engine > Container Registry. which can be managed/orchestrated by Kubernetes!

GCP Container Registry

Deployments: Instantiating A Container From the Docker Image

All we have now is a Docker image on the Container Registry. To actually run the application, we need to create an instance from the image - A Docker container.

To achieve this, we will create a Kubernetes Deployment which creates a pod for our container.

A Kubernetes pod is a group of containers, tied together for the purposes of administration and networking that are always co-located and co-scheduled, and run in a shared context.

Containers within a pod share an IP address and port space, and can find each other via localhost

There are two ways to create a deployment using the kubectl command. You can either specify the parameters in a yml file or the command-line.

Command-line approach

The deployment name is a unique identifier for your deployment that will be used to reference it later on. Specify the image name to create the deployment from and finally a port which will be mapped to our applications port that is exposed (see Dockerfile).

kubectl run {deployment_name} --image=gcr.io/$PROJECT_ID/{name}:{tag} --port={port}

file approach

I personally prefer this approach as it is easier for me to play around with the configurations of the deployment. Here is a look at our Deployment file, subtly named deployment.yml

apiVersion: extensions/v1beta1
kind: Deployment
  name: scotch-dep
    #Project ID
    app: scotch-155622
  #Run two instances of our application
  replicas: 2
        app: scotch-155622
      #Container details
        - name: node-app
          image: gcr.io/scotch-155622/node-app:0.0.1
          imagePullPolicy: Always
          #Ports to expose
          - containerPort: 8000

One of the interesting aspects is that deployments allows you to set the number of instances for your application. This scaling Kubernetes provision allows you to add or remove instances/replicas/pods depending on the traffic needs of your app.

To create the deployment, run:

kubectl create -f deployment.yml

You can view the deployment and the created pods by running kubectl get deployments and kubectl get pods respectively. Adding a -w enables us to see how the pods are created.

Services: Expose application to External traffic

Up until this point, there has been no way for us to actually run the application externally. Currently, the deployment is only accessible within the Kubernetes cluster (say, between two services in the same cluster).

Pods, like containers, can also be created and destroyed anytime, say, when scaling up or down or when rolling to an updated image. This means we cannot rely on the internal IPs since we cannot keep track of them.

Enter left, Services.

A Kubernetes Service is an abstraction which defines a logical set of Pods and a policy by which to access them - sometimes called a micro-service. It enables external traffic exposure, load balancing and service discovery for those Pods.

Just like a deployment, You can create a service from a them command-line or from a file.

command-line approach

kubectl expose deployment {service_name} --type="LoadBalancer"

file approach

Here is our Service file, again, subtly name service.yml.

kind: Service
apiVersion: v1
  #Service name
  name: node-app-svc
    app: scotch-155622
    - protocol: TCP
      port: 8000
      targetPort: 8000
  type: LoadBalancer

The Kubernetes master creates the load balancer and related Compute Engine forwarding rules, target pools, and firewall rules to make the service fully accessible from outside of Google Cloud Platform.

To create the service, run:

kubectl create -f service.yml

Once the service is created, we can get the externally accessible IP address by listing all the services (kubectl get services). The external IP may take a few seconds to be visible. You should get this.

Kubernetes services

From this, you can now access your application by visiting the IP address on port 8000 ( in our case. It works!.

Try out the other routes as well.

Scaling Up/Down your Application

Like we mentioned, one of the best provisions of Kubernetes is that you can scale your application up and down depending on your traffic requirements. This can be done by running the following command:

kubectl scale deployment {deployment_name} --replicas=n

In our case, the deployment name is scotch-dep and n can be 3 replicas to scale up from 2 instances or 1 to scale down. You can actually see the pods come alive or get terminated by running the kubectl get pods -w command.


You can also enable autoscaling in the cluster and set the minimum and maximum number of Pods based on the CPU utilization from the existing pods

kubectl autoscale deployment nginx-deployment --min=5 --max=10 --cpu-percent=75

Kubernetes Web UI

Kubernetes has a very neat web User Interface for every cluster that makes it easy to interact with the Kubernetes functionality - From deploying to autoscaling to managing and monitoring clusters.

To access the UI locally, run the following command.

$ kubectl proxy

Go to http://localhost:8001/u, you should see a dashboard like this. This particular service section displays all the pods running in the cluster.

Kubernetes Web UI dashboard

Read more on what you can do on the Kubernetes web UI here.


In this article, we have been able to:

  • Look at a few provisions of Google Cloud Platform,
  • Understand Kubernetes and it's role in deploying Docker containers to GCP.
  • Deploy a simple NodeJS application in a Docker container to GCP.
  • Understand Kubernetes Deployments, Services and scaling features.

There is still a lot more to learn on GCP. Next, we will look at Continous Integration/ Continous Deployment of Docker Containers with CircleCI which will give us the opportunity to automatically roll out upgrades to the application.

Like this article? Follow @_kar_is on Twitter