Expanding to dev, and not actually having people available on-call to deal with emergencies seems like an implementation problem.

The basic setup has worked well for me at several companies with many large teams on-call.