I'm a full-stack developer moving toward DevOps and want to run a small app with a Node server and SQL database. I'd like to learn containers, CI/CD, cloud services and monitoring without spending more than the project needs.
Is it worth using AWS for the experience, or should I start with a cheaper host? How would you split the budget?
Start with one small server and keep the monthly cost predictable. Run the app and database as containers, put a reverse proxy in front, and use a CI/CD pipeline to test and deploy changes. Add backups and a way to restore them before treating the database as anything you depend on.
That leaves some budget for short cloud experiments instead of keeping a load balancer and managed database running all month. You can rebuild the same setup with infrastructure as code later and compare what the managed services take off your hands.
If you specifically want AWS jobs, there is a good case for using AWS. You can learn its permissions model, command-line tools and service configuration alongside the general deployment work. Keep the architecture small rather than adding services just to collect names for a resume.
Check current free-tier rules and prices before creating anything. Set a billing alert on day one, review charges regularly, and shut down experiments when you're done. An alert warns you about spending, but don't assume it will stop resources or cap the bill.
verify!then~vacate+:
Managed databases, load balancers and storage can have separate charges, so price the whole setup rather than just the server.
A cheaper platform or another major cloud is fine if AWS isn't your target. Container builds, automated tests, deployments, monitoring and rollbacks are useful skills wherever you run them. Provider-specific details still differ, but you don't need an expensive setup to start learning the workflow.
For job applications, be ready to explain what you built and why. A clear account of how a change gets tested, deployed and rolled back says more than a list of services you opened once.
Don't put the whole learning budget into hosting. Leave time for Linux, shell scripting and networking too. Those basics help when a container won't start or the app can't reach its database, regardless of which provider you pick.
I'd also check free tiers and introductory credits for cloud and monitoring tools, but read the limits and expiry dates first. Offers change, and a free trial isn't the same thing as a predictable ongoing budget.
Answers are reviewed before publication.
CLOSET SPACE THEORY:
A single server is useful for learning, but it is also a single point of failure. Keep that tradeoff clear if anyone starts relying on the app.