Matt Rasband

Thanks for the shout out John!

I know that we still have a few things outstanding, some of which I will probably use Spring Cloud Cluster for:

1. HA Publisher: Right now we have 1 instance of the scheduler, which is unlikely to fail. But there is always a chance the host dies or something else - so having it distributed would be good

2. Abstraction around that huge annotation - that requires people to know too much about the implementation details of Rabbit.

Fair enough, all I know is after working in both - I prefer something closer to what the article talks of - after all, the article is actually a direct outflow from my work (John is my manager, I can only assume most of this is a direct outflow of my teams work).

if you are doing RESTful HTTP, you don't need any of these technologies

Unfortunately if you are in a system where you are doing modern "services" you really do (and it is hard). You need some way to dynamically determine service addresses - whether that is Eureka or something else is a different question. There is no feasible way to manage those manually unless you can guarantee the same `host:port` config for the apps, which is getting less and less common when you start thinking about "orchestrating" independent services.

Secondarily - a gateway abstracts away the potentially infinite number of unique C-Names for your system.

I actually gave a talk on this at our local meetup (slides), unfortunately I didn't get it recorded :/.