In the past, our applications lived on a single physical server where we had a 1:1 relation between the application and the backend implementing it. In this case, looking for a service is really simple: you only need to know the server IP address or the associated DNS name.

Later on, applications were distributed, which means that they lived on many physical servers to provide high availability. In this case, we have a 1:N relationship between a service and the backend servers, where N can represent more than one. Incoming requests are managed using a load balancer to route the requests among the available servers.

The same approach is used when the physical servers are replaced by virtual machines. Load balancers need some configuration to register the new servers available and route the requests properly. This task used to be executed by the operations team.

Today, it's common to find applications deployed within containers, which we will discuss further in Chapter 10, Containerizing your Applications. Containers are constantly being provided and destroyed every millisecond, so registering new servers manually is an impossible task and must be automated. For this purpose, Netflix created a project named the Eureka project.