Stop Using AWS for Your MVP

How many goddamn times are we going to watch someone spin up an entire AWS Rube Goldberg machine for a project that three people ever use? You’ve seen it. Hell, maybe you’ve done it. The MVP no one asked for, built like it’s prepping for the goddamn Super Bowl of uptime: Lambda functions everywhere API Gateway spaghetti Cognito because, sure, why not DynamoDB, S3, CloudFront, IAM, CloudWatch... It looked like an AWS certification vomited on your whiteboard. And then what? Nobody gave a shit. Nobody used it. Here’s the truth: You don’t need AWS. You need a fucking user. Stop Architecting for Imaginary Scale Unless you're Netflix (spoiler: you're not), your “product” doesn’t need 14 microservices and zero users. You read one Hacker News post and decided your todo app needs auto-scaling and distributed tracing? No. What you need is to ship. Something. Anything. Before you burn out configuring Terraform modules for a product no one asked for. Most products don’t fail because they couldn’t scale. They fail because they suck, or no one needed them, or the login button didn’t fucking work. What You Actually Need You’re building a side project, not launching SpaceX. Here’s the whole goddamn checklist: A €5/month VPS. Hetzner, DigitalOcean, pick your poison. Docker Compose. It works. Use it. Maybe a managed platform if you’re too lazy to ssh (no shame). That's it. No Kubernetes. No VPC wizardry. No 3 AM panic attacks over IAM roles. One server. One port. One fucking docker-compose up. When AWS Is Okay Alright, let’s not be assholes for no reason. AWS has its place: You’re learning it for a job. Fine. You’re legally required to store data on AWS. Cool. You actually need global scale. Doubtful, but okay. You’re already knee-deep in AWS and it’s too late to go back. Godspeed. But unless one of these is true, you’re just playing pretend DevOps while your actual product gathers dust. How to Actually Launch Something Want to go live in one day instead of six weeks? Write code docker-compose it Deploy to a cheap VPS with SSH Point your domain Done Need auth? Use open source. Need jobs? Use cron or a queue. You don’t need a goddamn AWS certificate to show a webpage. TL;DR You don’t need AWS. You need focus. You need to ship. AWS is not a substitute for a useful product. Simplicity is not your enemy—complexity is. Build. Launch. Learn. Scale if you need to. Until then, just fucking use a VPS.

May 12, 2025 - 20:05
 0
Stop Using AWS for Your MVP

How many goddamn times are we going to watch someone spin up an entire AWS Rube Goldberg machine for a project that three people ever use?

You’ve seen it. Hell, maybe you’ve done it. The MVP no one asked for, built like it’s prepping for the goddamn Super Bowl of uptime:

  • Lambda functions everywhere
  • API Gateway spaghetti
  • Cognito because, sure, why not
  • DynamoDB, S3, CloudFront, IAM, CloudWatch...

It looked like an AWS certification vomited on your whiteboard. And then what? Nobody gave a shit. Nobody used it.

Here’s the truth: You don’t need AWS. You need a fucking user.

Stop Architecting for Imaginary Scale

Unless you're Netflix (spoiler: you're not), your “product” doesn’t need 14 microservices and zero users. You read one Hacker News post and decided your todo app needs auto-scaling and distributed tracing?

No. What you need is to ship. Something. Anything. Before you burn out configuring Terraform modules for a product no one asked for.

Most products don’t fail because they couldn’t scale. They fail because they suck, or no one needed them, or the login button didn’t fucking work.

What You Actually Need

You’re building a side project, not launching SpaceX. Here’s the whole goddamn checklist:

  • A €5/month VPS. Hetzner, DigitalOcean, pick your poison.
  • Docker Compose. It works. Use it.
  • Maybe a managed platform if you’re too lazy to ssh (no shame).

That's it. No Kubernetes. No VPC wizardry. No 3 AM panic attacks over IAM roles. One server. One port. One fucking docker-compose up.

When AWS Is Okay

Alright, let’s not be assholes for no reason. AWS has its place:

  • You’re learning it for a job. Fine.
  • You’re legally required to store data on AWS. Cool.
  • You actually need global scale. Doubtful, but okay.
  • You’re already knee-deep in AWS and it’s too late to go back. Godspeed.

But unless one of these is true, you’re just playing pretend DevOps while your actual product gathers dust.

How to Actually Launch Something

Want to go live in one day instead of six weeks?

  • Write code
  • docker-compose it
  • Deploy to a cheap VPS with SSH
  • Point your domain
  • Done

Need auth? Use open source. Need jobs? Use cron or a queue. You don’t need a goddamn AWS certificate to show a webpage.

TL;DR

You don’t need AWS. You need focus. You need to ship. AWS is not a substitute for a useful product. Simplicity is not your enemy—complexity is.

Build. Launch. Learn. Scale if you need to.

Until then, just fucking use a VPS.