Recently on one of my side projects the time came to start MVP to prove hypothesis. I needed a solution to fast start without my own infrastructure and decided to deploy the project to the cloud. Using kubernetes on a daily basis made me too lazy to use ordinary virtual machines so I glanced at managed kubernetes services.
There are several competitors:
- GKE with 300$ for 90 days (0.1$ per hour ~ 75$ per month)
- Azure with 200$ for 30 days (0.1$ per hour ~75$ per month)
- DigitalOcean for 200$ for 60 days (40$ for 2nodes per month)
- IBM for 200$ for 60days (0.29$ per hour ~ 208$ per month 😞)
- Oracle for 300$ for 30 days (~77$ per month)
- Alibaba ? for free (64.8$ per month)
I started with GKE and after credit comes to end I’ll switch to DO and then to Azure, so I will have at least half a year of free usage before I’ll prove or fail with an idea. Ideal conditions.
GKE bootstrap
- Creating a new google account and attaching newly issued card
- Activate GKE account
- Go to the Kubernetes Engine section in menu
- Creating cluster and choosing autopilot option
- Downloading gcloud console utility
- Following steps on github manual to use in actions
- Creating new kubectl context with
kubectl config use-context humanway - Fetching GKE credentials with
gcloud container clusters get-credentials humanway-1 --project <project> --region <region>, where project and region could be found on the dashboard page of kubernetes cluster in GKE management page. - That’s all, now you can deploy to k8s with helm
Summary
Bright future is some, you can with no effort get free computing resources for proof of concept.