Hi, I initially posted a question on slack about more efficient replication: http://stackoverflow.com/q/43555490/2831606. Since then, I've been told about upcoming changes to the replicator in CouchDB that should solve some of my problems.
That being said, I still think there is a place for the solution I developed for a proprietary project. I've modified the design for CouchDB 2 and made it scalable and plan on open sourcing it. I'd appreciate any feedback: https://github.com/redgeoff/spiegel Also, if it is helpful in anyway, please feel free to use any pieces of this design in CouchDB core. I'm not sure what would be best located in the core DB vs an external project, but I personally find that Spiegel saves you from writing a lot of code.
