Create an internal API fleet
To perform many operations most of our servers either use a local API (connecting to unicorn on localhost) or connect to the public API endpoint.
I would like to create an internal API fleet to better split client facing traffic from our own. The benefits for this would be a more relaxed rate limiting and a cleaner experience for our API users out there.
Decoupling the API would mean stopping unicorn on many nodes, like sidekiq, which would then only run their main service, pushing ourselves one step closer to cloud native. This will scale independently from the public API, adapting to our needs.
In order to do this we need an internal load balancing tier, which I have already created in staging. We will need to pair at least with the Build team in order to test pointing our apps to the internal endpoint.