I am working on a scala + web application and am considering how to configure everything. I will be launching this on Amazon EC2 and I will have a pretty flexible architecture.
My current approach is to use an ELB configured with sticky sessions (elevator requirement), a nginx interface, with glass chips (or possibly a jetty) in the backend. nginx would be good for rewriting url and pushing static content (although looking at my static content in an Amazon S3 bucket is under consideration). I'm not sure if I should throw away the ELB and use HAProxy instead. I will need to add and remove instances of glass fish to dynamically scale depending on the load. Can nginx dynamically balance the load across multiple instances of genuine leather? Do I have to maintain a 1: 1 ratio from nginx to schooling flocks? Will ELB + nginx handle this accordingly? How would you configure ELB / nginx / glassfish to make it work? Does it make sense to use nginx? Should I just load the balance directly through a cluster of glass fish?
Thanks for any help you can provide!
dhoyt source share