Hacker News new | past | comments | ask | show | jobs | submit login

My take is that there's only a very specific kind of company where on-prem vs cloud is primarily a cost-driven decision.

"Cloud" providers are so much more than just the hardware. There are probably a lot of AWS customers who could absolutely save lots of money if they took all their cloud servers and magically made them physical on-prem servers. But then they might hit a situation like:

"We need to create a whole new prod-sized cluster for load testing but only for a day". Do you rack all those servers, do the test, then take them off and bin them? Or what?

I'm tempted to go on and on with similar examples but imo the biggest deal with cloud providers is that they can operate at such an insane infra scale that it lets you treat extremely large quantities of servers like an abstraction instead of a physical metal box that needs to be bottle-fed and rocked to sleep at night lest it get cranky.

I think at fairly small scales it's easy to build in-house systems that let you treat servers like abstract units as long as you're not trying to 2+x your infra dynamically. But cloud solutions let you do that because your scale <<< the cloud providers' scale even at pretty large values of scale for any one company.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

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

Search: