Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I wish people were as careful when thinking about adopting S3 as they seemingly are careful when others move away from S3.

Somewhere along the line, people starting defaulting to "at least 3 nodes for backend" and "cloud services for all infrastructure" even if the product they're building haven't even found product market fit.

Sure, if you know for a fact that your traffic will go and down more than 50% during a normal day, go for something that scales up and down quickly. But for most other use cases, the extra cost of cloud doesn't really make much financial sense, unless you're a fat VC-funded startup cat.



That reminds me of a friend who ran a SaaS business off a Heztner VPC.

It was a Node.js app he deployed via SSH and ran under a systemd job.

Used directories of JSON files as a database and the business logic was handled by a single endpoint that took JSON RPC payloads with different action types and metadata.

The app scaled to ~10,000 daily users like this.


I run multiple SaaSes, some popular, some not so much, off a single $20/mo Hetzner VPS with Dokku on it. It works great, and I've never needed to worry about anything.

Meanwhile, I see friends working on MVPs with 1-2 non-paying customers who already have costs in the thousands of dollars a month, but "it's fine because we got free money for a year". Yes, but that means that your company now has an expiration date of a year.


Why use Dokku, and not, say, docker-compose?


I like the automated deployments. These days I'd probably use Harbormaster, but I'm biased because I wrote it.


Hear, hear.

My personal AWS bill is roughly $10/month, all for S3. We're not talking millions here :). Personal compute is a mix of OVH and on-prem.

Work is an entirely different kettle of fish, at an entirely different scale, and primarily runs compute on spot: https://aws.amazon.com/blogs/aws/capacity-optimized-spot-ins...

Being able to scale down, rather than needing to pay for peak capacity, genuinely does save us large amounts of money. But it's a capability that we needed to build out, not something that happened by magic. And it does require that our services are big enough to scale for load, not just for redundancy.


Scale down-to/up-from zero is a helluva drug. You're describing poor engineering within the cloud itself. If you're using cloud, use it for the right things. Container-as-a-service is awesome. SaaS databases are nifty. Build up your startup on the cloud, then save costs by migrating off.




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: