I’ve also been reading up on AWS and have been wanting to give it a try, but I’m not happy with the billing situation: even for the the free tier they require a billing address and credit card number. The free tier only lasts a year, and [from what I can find] you are only able to setup a “warning” if you are exceeding the free tier usage..
I would be more comfortable giving them billing info if I could setup hard cutoffs: eg “stop service if it exceeds the free tier” or even just “issue warning at x$, but cutoff service at y$”
The reason being: I’ve read horror stories over the years of config errors or (even worse) an API key leaking out resulting in 100s or even 1000s of dollars of AWS service being billed. While I’ve also read the AWS team is very helpful to work through problems like this, I still find it unacceptable, especially in my hobbyist situation.
That being said: my typical usage would likely be pennies a month, and it would be nice to have a more resilient mirror for my personal sites…
“Don’t worry! For this tutorial we guide you through setup of a free tier account for you to try it out!” -> “please enter your credit card info” … nope. Luckily the tutorial and documentation is still good enough I can learn about it without an account, but damn Amazon…
Apparently, this has been an open request since 2011? https://forums.aws.amazon.com/thread.jspa?threadID=58127&start=50&tstart=75
However, I did find this: http://docs.aws.amazon.com/awsaccountbilling/latest/aboutv2/detailed-billing-reports.html
So, ideally, I could setup an monitor instance that will do this for me.
The more and more I read about this, the more I see the dichotomy evolving: hobbyists or personal or non-revenue generating tasks where the person doesn’t care if their service gets cutoff, and the infrastructure/revenue/business cases where costs can be absorbed, but uptime and availability is absolutely vital. The former would benefit from a cutoff, the later needs to adapt best practices, and a cutoff switch could absolutely cripple them.
AWS seems to favor the latter, for lack of a nice solution.