AWS

Hello everyone! 👋

Last weekend, I had the amazing opportunity to be a speaker at AWS Community Day Indonesia 2025. It was one of the most rewarding and eye-opening experiences in my tech journey and in this post, I’d love to share what the day was like, what I learned, and why it meant so much to me. BTW, I shared about “Building a Multi-Tenant Machine Learning Platform on AWS EKS with Ray and JupyterHub”.

Before the Event: Preparation

A few days before the event, to be honest, I felt a mix of excitement and nervousness. Even though I’ve presented in smaller settings before, speaking at a community event like this felt special and a bit more serious!

Here’s what I did to prepare:

- Finalized my presentation topic, making sure the content was practical and relevant for developers and cloud enthusiasts.

- Practiced my talk several times out loud to get comfortable with the flow.

- Set my mindset: I’m not just here to present. I’m here to connect, share, and learn from others.

That mindset shift really helped me calm down before stepping on stage.

During the Event: On Stage and Beyond

When the day finally came, the atmosphere was incredible buzzing with conversations, excitement, and passion for AWS and cloud technologies.

Some of my favorite moments were:

- Meeting so many people from the AWS community, many of whom I had only interacted with online before.

- Taking the stage to deliver my session “Building a Multi-Tenant Machine Learning Platform on AWS EKS with Ray and JupyterHub”.

- Even though I was nervous at first, once I started talking, everything clicked. It didn’t feel like a “presentation” anymore it felt like a conversation with friends who shared the same curiosity and enthusiasm.

Key Takeaways & Reflections

Looking back, here are a few lessons that really stuck with me:

- Community is everything. Behind every technology, it’s always about people learning, sharing, and growing together.

- Preparation builds confidence. The more I practiced, the more natural it felt on stage.

- Sharing is a two-way process. I came to give a talk, but I left with more insights from the audience than I expected.

- Stay adaptable. Every crowd is different. Reading the room and adjusting how you explain something is an important skill.

- Keep growing. This experience reminded me that growth doesn’t just happen behind the keyboard it happens when you step out, speak, and share what you’ve learned.

What’s Next?

To be honest, there are still many aspects I haven’t fully covered in my presentation. After this event, I plan to expand this talk into a deeper technical series exploring the end-to-end ML lifecycle using Ray on AWS.

While this session mainly focused on the infrastructure layer, there’s so much more to uncover about how Ray can power the entire machine learning pipeline from data preprocessing and distributed training with Ray Train, to hyperparameter tuning with Ray Tune, and serving models in production with Ray Serve.

You can see my deck here: Deck

Let’s start again. Now we will do some practices.

In this part of the Kubernetes series, we will explore how to create a Kubernetes cluster in different environments. Whether you’re running Kubernetes locally or in the cloud, understanding how to set up a cluster is fundamental to deploying and managing containerized applications efficiently.

We will cover three different ways to create a Kubernetes cluster:

- Kind (Kubernetes in Docker) - A lightweight way to run Kubernetes clusters locally for testing and development.

- K3D (K3S in Docker) - A more lightweight Kubernetes distribution, optimized for local development and CI/CD workflows.

- EKS (Amazon Elastic Kubernetes Service) - A managed Kubernetes service provided by AWS for running Kubernetes workloads in the cloud.

Each approach has its own use cases, advantages, and trade-offs. Let’s dive into each one and see how to set up a cluster.

Setting Up a Kubernetes Cluster with Kind

Kind (Kubernetes in Docker) is one of the simplest ways to spin up a Kubernetes cluster for local development and testing. It runs Kubernetes clusters inside Docker containers and is widely used for CI/CD and development workflows.

Prerequisites

- Docker installed on your machine. (installation guide)

- KIND CLI installed. (installation guide)

- Kubectl CLI installed. (installation guide)

Create a Cluster with Kind

- Create a new Kind cluster:

$  kind create cluster --name kind-cluster

Creating cluster "kind-cluster" ...
 ✓ Ensuring node image (kindest/node:v1.31.0) 🖼
 ✓ Preparing nodes 📦
 ✓ Writing configuration 📜
 ✓ Starting control-plane 🕹
 ✓ Installing CNI 🔌
 ✓ Installing StorageClass 💾
Set kubectl context to "kind-kind-cluster"
You can now use your cluster with:

kubectl cluster-info --context kind-kind-cluster

Thanks for using kind! 😊

• Check the cluster:

$   kubectl cluster-info --context kind-kind-cluster

Kubernetes control plane is running at https://127.0.0.1:43417
CoreDNS is running at https://127.0.0.1:43417/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

- List available nodes:

$ kubectl get nodes

NAME                         STATUS   ROLES           AGE   VERSION
kind-cluster-control-plane   Ready    control-plane   75s   v1.31.0

Create Simple Deployment

- Use the kubectl create deployment command to define and start a deployment:

$  kubectl create deployment nginx --image=nginx

deployment.apps/nginx created

- Check deployment status using kubectl get deployment command:

$ kubectl get deployment

NAME    READY   UP-TO-DATE   AVAILABLE   AGE
nginx   0/1     1            0           29s

- Expose the deployment:

$ kubectl expose deployment nginx --port=80 --type=LoadBalancer

service/nginx exposed

- Verify the Pod status and then try to access Nginx using your browser:

$ kubectl get pods

NAME                     READY   STATUS    RESTARTS   AGE
nginx-676b6c5bbc-wd87x   1/1     Running   0          12m

- Delete the cluster when no longer needed

$  kind delete cluster --name kind-cluster

Deleting cluster "kind-cluster" ...
Deleted nodes: ["kind-cluster-control-plane"]

Setting Up a Kubernetes Cluster with K3D

K3D is a tool that allows you to run lightweight Kubernetes clusters using K3S inside Docker. It is a great choice for fast, local Kubernetes development.

Prerequisites

- Docker installed on your machine. (installation guide)

- K3D CLI installed. (installation guide)

- Kubectl CLI installed. (installation guide)

Create a Cluster with K3D

- Create a new K3D cluster:

$ k3d cluster create my-k3d-cluster

INFO[0000] Prep: Network
INFO[0000] Created network 'k3d-my-k3d-cluster'
INFO[0000] Created image volume k3d-my-k3d-cluster-images
INFO[0000] Starting new tools node...
INFO[0000] Starting node 'k3d-my-k3d-cluster-tools'
INFO[0001] Creating node 'k3d-my-k3d-cluster-server-0'
INFO[0001] Creating LoadBalancer 'k3d-my-k3d-cluster-serverlb'
INFO[0001] Using the k3d-tools node to gather environment information
INFO[0001] HostIP: using network gateway 172.20.0.1 address
INFO[0001] Starting cluster 'my-k3d-cluster'
INFO[0001] Starting servers...
INFO[0001] Starting node 'k3d-my-k3d-cluster-server-0'
INFO[0008] All agents already running.
INFO[0008] Starting helpers...
INFO[0008] Starting node 'k3d-my-k3d-cluster-serverlb'
INFO[0016] Injecting records for hostAliases (incl. host.k3d.internal) and for 2 network members into CoreDNS configma
p...
INFO[0018] Cluster 'my-k3d-cluster' created successfully!
INFO[0018] You can now use it like this:
kubectl cluster-info

- Check the cluster status:

$ kubectl cluster-info

Kubernetes control plane is running at https://0.0.0.0:46503
CoreDNS is running at https://0.0.0.0:46503/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy
Metrics-server is running at https://0.0.0.0:46503/api/v1/namespaces/kube-system/services/https:metrics-server:https/p
roxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

- List available nodes:

$ kubectl get nodes

NAME                          STATUS   ROLES                  AGE     VERSION
k3d-my-k3d-cluster-server-0   Ready    control-plane,master   2m33s   v1.30.4+k3s1

Create Simple Deployment

- Use the kubectl create deployment command to define and start a deployment:

$ kubectl create deployment httpd --image=httpd

- Check deployment status using kubectl get deployment command:

$ kubectl get deployment

NAME    READY   UP-TO-DATE   AVAILABLE   AGE
httpd   0/1     1            0           45s

- Verify the Pod status:

$  kubectl get pods

NAME                     READY   STATUS    RESTARTS   AGE
httpd-56f946b8c8-84ww8   1/1     Running   0          9m11s

- Expose the deployment:

$ kubectl expose deployment httpd --port=80 --type=LoadBalancer

- Try to access using browser:

- Delete the cluster when no longer needed:

$ k3d cluster delete my-k3d-cluster

INFO[0000] Deleting cluster 'my-k3d-cluster'
INFO[0001] Deleting cluster network 'k3d-my-k3d-cluster'
INFO[0001] Deleting 1 attached volumes...
INFO[0001] Removing cluster details from default kubeconfig...
INFO[0001] Removing standalone kubeconfig file (if there is one)...
INFO[0001] Successfully deleted cluster my-k3d-cluster!

Setting Up a Kubernetes Cluster on AWS EKS

Amazon Elastic Kubernetes Service (EKS) is a fully managed Kubernetes service on AWS, designed for running production workloads.

Prerequisites

- AWS CLI installed and configured. (installation guide)

- EKSCTL CLI installed. (installation guide)

- Kubectl CLI installed. (installation guide)

Create a cluster on EKS

To create a Kubernetes cluster in AWS, you can use the AWS Console (dashboard) or the eksctl CLI. For this guide, we will use eksctl.

We will provisions an EKS cluster with two t4g.small nodes in the us-east-1 region, making it ready for running Kubernetes workloads.

$ eksctl create cluster  \
--name cluster-1  \
--region us-east-1 \
--node-type t4g.small \
--nodes 2 \
--nodegroup-name node-group-1

2025-02-01 19:52:35 [ℹ]  eksctl version 0.202.0
2025-02-01 19:52:35 [ℹ]  using region us-east-1
2025-02-01 19:52:37 [ℹ]  setting availability zones to [us-east-1c us-east-1f]

...

2025-02-01 20:02:04 [ℹ]  creating addon
2025-02-01 20:02:04 [ℹ]  successfully created addon
2025-02-01 20:02:05 [ℹ]  creating addon
2025-02-01 20:02:06 [ℹ]  successfully created addon
2025-02-01 20:02:07 [ℹ]  creating addon
2025-02-01 20:02:07 [ℹ]  successfully created addon
"us-east-1" region is ready

- Access AWS console, navigate to the EKS service and you can see the cluster is successfully created.

- List available nodes:

kubectl get nodes
NAME                             STATUS   ROLES    AGE   VERSION
ip-192-168-xx-yy.ec2.internal   Ready    <none>   17m   v1.30.8-eks-aeac579
ip-192-168-xx-yy.ec2.internal   Ready    <none>   17m   v1.30.8-eks-aeac57

Create Simple Deployment

- Use the kubectl create deployment command to define and start a deployment:

$ kubectl create deployment nginx --image=nginx

deployment.apps/nginx create

- Check deployment status using kubectl get deployment command:

$  kubectl get deployment

NAME     READY   UP-TO-DATE   AVAILABLE   AGE
nginx    1/1     1            1           23s

- Verify the Pod status:

$ kubectl get pods

NAME                     READY   STATUS         RESTARTS   AGE
nginx-bf5d5cf98-9dld5    1/1     Running        0          43s

- Expose the service:

$ kubectl expose deployment nginx --type=LoadBalancer --port=80 --name=nginx-service

- Try to access using the browser:

- Delete the cluster when no longer needed:

$  eksctl delete cluster --name cluster-1 --region us-east-1

2025-02-01 20:51:59 [ℹ]  deleting EKS cluster "cluster-1"
2025-02-01 20:52:02 [ℹ]  will drain 0 unmanaged nodegroup(s) in cluster "cluster-1"
2025-02-01 20:52:02 [ℹ]  starting parallel draining, max in-flight of 1
2025-02-01 20:52:04 [ℹ]  deleted 0 Fargate profile(s)
2025-02-01 20:52:13 [✔]  kubeconfig has been updated
2025-02-01 20:52:13 [ℹ]  cleaning up AWS load balancers created by Kubernetes objects of Kind Service or Ingress
2025-02-01 20:52:56 [ℹ]
...

2025-02-01 21:02:00 [ℹ]  waiting for CloudFormation stack "eksctl-cluster-1-nodegroup-node-group-1"
2025-02-01 21:02:01 [ℹ]  will delete stack "eksctl-cluster-1-cluster"
2025-02-01 21:02:04 [✔]  all cluster resources were deleted

Conclusion

Setting up a Kubernetes cluster is the first step in running containerized applications at scale. In this guide, we’ve explored three different ways to create a Kubernetes cluster and do a simple deployment: using Kind and K3D for local development and using EKS for cloud-based deployments. Each method has its own advantages depending on your use case.

Thanks for reading this post. Stay tuned!

References:

- KUBERNETES UNTUK PEMULA. https://github.com/ngoprek-kubernetes/buku-kubernetes-pemula.

- How do I install AWS EKS CLI (eksctl)?. https://learn.arm.com/install-guides/eksctl