With a monolith we have one huge api project for which we used plural nouns to represent each entity. For example http://monolith.domain.com/Payments, http://monolith.domain.com/Customers, http://monolith.domain.com/Orders, etc.
With the microservice architecture, we now have the plural noun as the microservice itself. So what is the best practice for naming the endpoints?
It's fairly common to have an API gateway to route traffic from the main URL to a specific microservice. The way you set up the URLs within your microservice really depends on the idioms you choose. In essence, your microservice is the bounded context for the nouns you've identified (Payments, Customers, Orders).
So your API Gateway does the core routing:
The temptation then is to simply have the REST endpoints at the very root of your microservice. Exampled:
However, you may be hampered by your framework. For example in my team's project we found that the Spring Boot infrastructure wouldn't show the Swagger-UI if everything was at the root like that. For that reason we had to add the verbs to the URL. We still use the HTTP Methods, but the URL mapping looks more like this:
The main take-aways from this would be the following:
Consistency is more important than any notion of "purity" or "best", because if one service implements the URL pattern one way and another service does it differently for the same type of functionality--then you are making the consumers of your API pay for it.
Regarding what I am working with microservices architecture, the structure of our endpoints should be as follows:
http(s)://domain.com/{service-name}/{api-version}/
Thus, the action of endpoints is determined by HTTP action: GET, POST, PUT, DELETE
