> Yes, you get the benefit of backups, auto-restore, increased reliability etc, but.. How to set up HA databases with frequent backups and auto-restore is pretty common knowledge at this point.<p>Oh my god I’m gonna make this into a poster, my DBA friends will love it. But seriously these are not common skills — these are skills you pay $200k for and still struggle to find quality people. If you found yourself somewhere where the average SWE can set up a metal to production ready HA DB with turnkey live restores dm me your location so I can start looking on Zillow.<p>Also we’re just gonna gloss over that the AWS setup includes a HA managed load balancer that Hetnzer doesn’t as well as the k8s control plane.<p>Normally I’m on here telling people how not scary and complicated on-prem (ish in this case) deployments are but I think I’ve actually found someone who actually swung the other way.
I think people should just be more open to use other providers.
Not immediately jump to AWS when they want to host anything, especially due to it's pricing and overhead.<p>Linode, DigitalOcean, Hetzner, OVH and many others will probably be enough or sometimes better for your use case while still being cheaper.
The writer concludes that we should deploy and manage our PaaS. This answer is in the same vain as "just buy your own servers". The point of using AWS is that I don't need to manage or even worry about of it. I pay the premium of using EKS & ECS so that I don't personally need to manage my clusters. Going backwards isn't an answer.
In the AWS 80% of the bill is Cloudfront. Why not putting Cloudflare on top of AWS and you have almost the same price. This article is basically comparing paid vs free CDN...