mateo-maza

I see, someone commented it on this post. His argument is that most hardware issues are detected in power-on-self-test, and live kernel patching stops this. To be completely honest, I'm too unexperienced to know if it's pointless to consider this or not. Your point does make sense to me, if you only update the OS after several months, then it can't be a crucial thing for the comparison.

Really interesting that you have found on-prem 2x10x faster within a region, thanks for that info. I'll let you know how things go with the experiment, cheers :D

:D I'll make a post and hit you up when I have something. As someone adviced me here, it would be important for the comparison to consider if the on-premises example does live kernel patching. Could you share me this info please?

Wow this is extremely good info, I'm going to investigate on this for the comparison. Thanks for the tip, your comment was just too helpful!

Awesome! Thanks, really appreciate it :)

The idea for this comparison came when I read in article that 40% of internet downtime is due to hardware issues (not sure if this is true). The experiment aims to compare the uptime between a website using a reliable cloud provider and one managed in-house due to compliance or other reasons. I understand your point about the variety of configurations that exist beyond simple cloud vs. on-premises. However, my focus is specifically on comparing traditional on-premises data center management against using established cloud services. Here's the repo of the project if that helps you believe in what I say https://github.com/mateomaza/cloudXground. I don't even have the knowledge to do something wrong with the info tbf, I mean just look at my GitHub.