I can't wait for the followup analysis where they discover they saved a ton of money by running it on prem on bare metal hardware without "all the overhead" of a hypervisor.
@vwbusguy my whole career has still been setting up local on prem stuff for people, I just don’t get how people fall for the Amazon suite with crazy high bills… if I need to spin up a new node immediately to handle something brief and on demand, then I don’t mind cloud but long term it’s a cash sink for lower quality support
@CommodoreSpence It is definitely very useful for startups and provisioning infra you don't otherwise have on hand, but the costs definitely scale with it, which is one of the reasons we're now investing in more on prem baremetal stuff in my dayjob. Granted, we already have ops and data centers for it all and that's not a small thing if you don't already have that.
@vwbusguy agreed, it’s easier if you have the DC’s already. But once you get big in cloud it’s a monster, hybrid is the better way, with long term plans to repatriate data from temporary remote cloud nodes as budgets allow
@CommodoreSpence Generally, yes. It also depends on what you're trying to do. For example, if you're setting up a CDN that needs to scale globally with low latency, that's very difficult to do with self-provisioned equipment. But many companies architect for this use case when their company's website is just some basic react app with a database of 20G or less.
@CommodoreSpence But yeah, public cloud is excellent for prototyping. I'm not going to just buy a whole Ampere rack setup without first testing it out on AWS Graviton first to make sure the application stack can run on ARM, etc.
Add comment