The analysis don't seems to take in account the labor to maintain that infrastructure up and running to AWS's SLA level. You can't do the same operation with the same team on prem vs cloud without shifting some priorities.
Don't get me wrong, I do believe that cloud is way too overprice compare to having hardware on-prem, especially if you don't leverage autoscaling to spin up and down, but the analysis seems a bit too simplistic.
reply
You're right, the analysis is oversimplified.
I wanted to find out more about it, because a client of mine happens to be in a similar situation and is reconsidering their cloud strategy. Also, I'm quite familiar with 37signal's services and expected them to have done their homework on this.
Turns out they did. Found a lot more detail on their reasoning in a podcast of theirs in late 2022. I only read the transcript, which unfortunately may have been auto-generated since it seems to contains some errors. Still quite informative, though.
You can find the podcast here, if you're interested:
reply
A second part of this podcast was published yesterday: https://37signals.com/podcast/leaving-the-cloud-part-2/
reply
Thanks a lot, I'll listen to it ! :)
reply
Quite a contradictionary move from another big tech company in a other (more consumer) market: https://newsletter.pragmaticengineer.com/p/uber-move-to-cloud
reply
This was a good read, thanks.
reply
I don't think these people had to upgrade server firmwares or replace failed SSDs. They might be in for multiple surprises in the next five years.
reply
wen aws on umbrel
reply
Let’s gooo!
reply