How to configure a good max number or connections in a node app on dedicated-cpu-1x machines ?

I am doing a node app that is going to be used in real world events, this means my traffic consist of several spikes during the day

An example: 1 hour with 15k requests , 5 hours with 100 request, 1 hour with with 13k request , 10 hours of 600 requests

How can I configure my instances so they only scale when they really need to ?
What could be a number of max connections to optimize for cost using dedicated-cpu-1x machines ?

fly supports several autoscaling options, you can read the documentation to learn more. In my practice, this seems to work fine.

For load control of traffic, the load documentation should be worth referring to. (hard_limit & soft_limit)