Backend instances should indeed accept concurrent requests, and the same '
default' concurrent requests that Automatically scaled instances have should also be seen by Backend instances. The only difference being, Automatic scaling allows for a user to change the 'default' amount up to a pre-defined max limit.