But the idea that AWS provides some sort of white glove 24/7 support is laughable for anyone that's ever run into issues with one of their products...
You will definitely get support reasonably fast if something breaks because of them but that’s not where breakage happens most of the time. The issue will nearly always be with how you use the services. To fix that, you need someone who understands both the tech you use and how it’s offered by your cloud provider. At which point, you have an expert on board anyway so what’s the point of the huge bill?
A hoster will cost you less for most of the benefits. They already offer most of the bricks required for easy scalability.
1. Dumping a backup every so often?
2. Exporting its performance via Prometheus, and displaying in a dashboard?
3. Machine disk usage via Prometheus?
4. An Ansible playbook for recovery? Maybe kicking that into effect with an alert triggered from bullet 2 and 3.
5. Restoring the database that you backed up into your staging env, so you get a recurring, frequent check of its integrity.
This would be around 100 to 500 lines of code of which an LLM can do for you.
What am I missing?
Besides this we also use - ECS to autoscale app layer - S3 + Athena to store and query logs - Systems Manager to avoid managing SSH keys. - IAM and SSO to control access to the cloud - IoT to control our fleet of devices
I’ve never seen how people operate complex infrastructures outside of a cloud. I imagine that using VPS I would have a dedicated dev. ops acting as a gatekeeper to the infrastructure or I’ll get a poorly integrated and insecure mess. With cloud I have teams rapidly iterating on the infrastructure without waiting on any approvals and reviews. Real life scenario 1. Let use DMS + PG with sectioned tables + Athena 2. Few months later: let just use Aurora read replicas 3. Few months later: Let use DMS + RedShift 4. Few months later: Zero-ETL + RedShift.
I imagine a dev. ops would be quite annoyed by such back and forth. Plus he is busy keeping all the software up to date.
That’s your issue. If all you have is a hammer, everything looks like a nail.
I have the same issue with the junior we hire nowadays. They have been so brain washed that the idea that the cloud is the solution and they can’t manage without them that they have no idea of what to do instead of reaching for them.
> I imagine that using VPS I would have a dedicated dev. ops acting as a gatekeeper to the infrastructure or I’ll get a poorly integrated and insecure mess.
You just describe having a real mess after this.
> I imagine a dev. ops would be quite annoyed by such back and forth.
I would be quite annoyed by such back and forth even on the cloud. I don’t even want to think about the costs of changing so often.
While I admit lack of experience at scale I had my share of Linux admin experience to understand how it could be done. My point is that building a comparable environment without cloud would be much more than just 500 LoC. If you have relevant experience please share.
>I would be quite annoyed by such back and forth even on the cloud. I don’t even want to think about the costs of changing so often.
In cloud it took 1-2 weeks per iteration with several months in between when we have been using the solution. One person did it all, nobody in the team even noticed. Being able to iterate like this is valuable.
This is not the case. The reason for iteration is the search for solution in the space we don’t know well enough. In this particular case cloud made iteration cheap enough to be practical.
I asked you to think about what it would take to build well integrated suite of tools (PG + backups + snapshots + prom + logs + autoscaling for DB and API + ssh key management + SSO into everything). It is a good exercise, if you ever built and maintained such a suite with uptime and ease of use comparable to AWS I genuinely would like to hear about it.
It's their first "leadership principle" (their sort of corporate religion, straight out of the lips of Jeff himself)
Also, the idea that using VPS or non-hyperscaler clouds means “poorly integrated and insecure mess” feels like AWS marketing talking. Good ops doesn’t mean gatekeepers — it means understanding your system so you don’t need to swap out components every quarter because the last choice didn’t scale as promised.
I’d rather spend time building something stable that aligns with my compliance and revenue goals, than chasing the latest AWS feature set. And by the way, someone still has to keep all that AWS software up to date — you’ve just outsourced it and locked yourself into their way of doing it.
I don’t understand why people keep propagating this myth which is mostly pushed by the marketing department of Azure, AWS and GCP.
The truth is cloud provider doesn’t actually provide 24/7 support to your app. They only ensure that their infrastructure is mostly running for a very loose definition of 24/7.
You still need an expert on board to ensure you are using them correctly and are not going to be billed a ton of money. You still need people to ensure that your integration with them doesn’t break on you and that’s the part which contains your logic and is more likely to break anyway.
The idea that your cloud bill is your TCO is a complete fabrication and that’s despite said bill often being extremely costly for what it is.