with cloud being the talk of the industry these days. how pervasive are SRE roles. I would wager a good system admin | SRE can save a company thousands if they provisioned their own hardware i.e colo instead of cloud instances etc
There is an incredible number of metrics around every AWS service, and the decoupling of things via queues, topics, APIs etc. mean there are a lot of moving parts.
It’s unlikely that a “good” sysadmin / SRE, or even a team of them, could efficiently replicate the versatility and reliability offered by cloud services.
You need a half dozen engineers working full time to replicate what the cloud gives you, and still buy the hardware ahead of time. That's 6 months of straight project development with 6 people and capital for a rack full of gear.. that's like $1.5M USD outlay for one year. For close to feature parity for core services I mean, nobody would normally build all that just to put a crappy web app online.
OR hire one SRE to set up a cloud stack in two months and only pay for what you use, $200K for one year or $100K for a short contact.
The decision for where the hardware will exist may be advised by SREs, but is generally not exactly in their wheelhouse. SREs tend more towards lowering the cost and maximizing the utilization of whatever methods the company as a whole decides to use.
Running our primary DB cluster in AWS for a month costs more than the purchase price of the hardware the cluster runs on.