Not so sure about your load balancing example

[sachabarber]

Your load balancing example doesnt actually do any load balancing at all, unless I am mistaken.

From what I can see you have a publisher (client) that pushes out on Rabbit channel, and an endpoint (SchoolService which is really just Queue listener/handler) that simply responds to the incoming Rabbit channel message and sends response back to publisher.

This is it, there is no load balancing out to any other services (unless you call your DataStore a service which it really isnt, as its just a repo really) ok you say you are using Rabbit broker as load balancer, but feels a little confused to me

Dont get me wrong this is excellent repo, but this is not load balancing at all, as there is just a pub/sub and competing consumer, its just a rabbit demo really.

I guess what I was expecting to find, was more in line with say how Kubernetes does things, that you have your API endpoints, and you have a rabbit consumer over that, that knows which API services are available and would round robin between them. There would have to be some form of registration associated with that though where each endpoint would register itself with the actual load balancer (this could be app for me I guess) via a rabbit message

Ok I understand this arguably achieved the same thing, but its not really load balancing, its competing consumers really

Thoughts?

[dukecheng]

Same understanding for load balancing part, other part are very good for understanding the service discovery concept.

@sachabarber maybe https://github.com/ThreeMammals/Ocelot is a good project for you for the load balancing, I know its have the load balancing feature, actually concul is a k/v database, its for configuration, not for load balance.