My point is, that you don't want to do this inside of a load balancer there. I don't recall any traffic filtering abilities that would suffice, but I'm not fully up-to-date with the configurability either. If the load balancer supports it, a short search in the net or docs should surface an easily-applicable guide, and if not, I'd probably put that blocking closer to my app server.
And the reason against AWS for this would be both the general cost (AWS is not cost-efficient in many cases, unless you have complicated infrastructure that takes a lot of management, where the Infrastructure-as-code approach can give you a sizable benefit), the bandwidth cost in particular, and the lack of configurability of their services to e.g. apply suitable tar-pitting against such crawlers.
