Building a Free Cloud Infrastructure - From Free High-Spec Servers to Deploying Your Own App (feat. Oracle, k8s)
I want to launch a side project as a real service and study Kubernetes, but managed services from AWS/GCP (EKS, GKE, etc.) are too expensive. On the other hand, running it only on a local computer makes it difficult to experience network configurations or actual deployment environments. We will utilize the "Always Free" tier of Oracle Cloud (OCI), which offers the most powerful free benefits among existing public clouds. By installing the lightweight yet powerful k3s, we will build a Linux server and Kubernetes environment capable of hosting actual services from scratch—all at zero cost. After taking this course, you will gain a general understanding of cloud networking and possess your own Kubernetes cluster infrastructure where you can deploy and operate your applications for free.
148 learners
Level Basic
Course period Unlimited
News
2 articles
Hello.
Lecture Material PDF Update
As of June 2026, the limits for Oracle's free resources have changed, so I have updated the lecture PDF to reflect these changes. I apologize for the delay. For those who need it, please feel free to download the updated version.
Regarding Part 2
Part 2 will proceed with the following content, based on the feedback from those who shared their opinions on the next lecture topic.
1. Upload sample service source + Dockerfile to Dockerhub using GitHub Actions
2. Running a sample service on the built cluster
3. Setting up and connecting a real domain for free (planned to use free DNS, Cloudflare, etc.)
(CI/CD automation using ArgoCD will be decided after checking the total course volume)
Setting up all of Part 1's hands-on practice with AI!
I had been putting off Part 2 due to personal matters and only recently began preparations, but since parts of the cluster had broken in the meantime, I ended up setting everything up again from scratch. I usually use Claude Code, so I tried letting it handle the setup, and it seems that most of the practical results from Part 1 can be configured automatically as long as you have the information below.
1. Class material PDF
2. Claude Code (Opus 4.8)
3. Connection information for the two instances (SSH information such as IP, key, etc.)
4. A few settings such as opening cloud security group ports (can be easily handled with AI)
So, even if you have been putting off the hands-on practice because it seemed cumbersome, you can set it up without much difficulty by utilizing AI. I highly recommend you try the wonderful experience of owning your own personal cluster and deploying various things on it.
(For your information, I am using Oracle Cloud because of its free resources, but since you can expand in the same way using AI even with instances from other clouds, feel free to give it a try.)
Thank you.
Hello.
It is said that there are changes to the Oracle free resource limits.
Existing 4core cpu + 24gb ram
Changed to 2core cpu + 12gb ram
For those who have already created them, master 1core cpu, 6g ram / worker 1core cpu, 6g ram
It seems you need to reduce your usage to avoid being charged. (It is expected that if you continue to use the full capacity, a charge of about $26 will be incurred.)
You can check the details here as well
https://news.hada.io/topic?id=30458It appears that the content of the official Oracle documentation has also been updated.
https://docs.oracle.com/en-us/iaas/Content/FreeTier/freetier_topic-Always_Free_Resources.htm
I will update and upload the lecture PDF files and lecture notes.
