I seem to recall Netflix asking me to consult and I told them how to fix this using a SINGLE REQUEST/RESPONSE with a FORWARD

GraphQL uses a separate request/response for EVERY endpoint call and a redirect, thus dropping the thread, going outside the DMZ and coming back in and reissuing the request/response with each endpoint call.

This creates HIGH IO and UNECCESSARY excessive PROCESSING.

This could have all been fixed had they not used the old API pattern built in the 70’s and meant for centralized architectures and instead used a more modern pattern meant for distributed architectures.

When I last showed this to the API manager at Netflix, he stated (and I quote) ‘this fixes everything we are having issues with’

Unfortunately, He also stated that Netflix’s ‘bean counters’ were most likely not going to spend the money to do a rewrite.