Self Managed Kubernetes

Reez Patel

·

Follow

5 min read

·

May 24, 2020

3

Listen

Share

As the technology is evolving, “cloud computing” is becoming more and more popular. Developing cloud-native application is modern standard. Having your application auto scaled to satisfy incoming traffics is a blessing. With this any company can scale their resources according to usage.

Motivation

Well, any person who is setting his/her foot in kubernetes realm would first look at a managed kubernetes solution, it can be Amazon’s AWS, Google’s GCP, Microsoft’s Azure, DigitalOcean the list goes on and on. You can find official certified kubernetes provider here.

So when I started, I chose to go with DigitalOcean Kubernetes. Creating a kubernetes cluster in DigitalOcean is pretty straight forward, just click on some button.. give your cluster a name, choose number and type of node and you done. SIMPLE AF.

For my DigitalOcean’s k8s cluster I choose to go with 3 10$ nodes, installed some application and I was short of nodes to spin up new pods. Just with this setup I was paying 50$ (10$ Loadbalancer + 30$ Nodes + 10$ Storage). Certainly I could do better.

Search for other solutions

So my main aim was to find a VPS which is cheaper than 50$ which offers more computing power, memory and storage then DigitalOcean. Lucky I found many VPS provider which offers this. I chose to go with Contabo . Specifically there “VPS XL SSD” plan. For just 26.99 Euro they provide 10 vCore CPU, 60 GB RAM and 1.6TB SSD. 😱😱😱

So with this I was all set to build my self managed kubernetes cluster. I purchases 2 VPS (for 2 nodes). This will more than enough for me to get started. Contabo’s process of spinning up a new VPS is very slow. They say it may take up 48 hours to set up VPS, and sure it did take close to 48 hours to send me my SSH credentials.

Creating a self managed cluster

There are many options to run your own k8s cluster, for local cluster “minikube” is a good choice. I decided to with…