The NSI operations accommodate two distinct models for defining servers: the service model and the resource model. These models express different views of how clients use servers and how servers can present themselves in the directory service database. The models are not mutually exclusive, and an application may need to implement both models to meet diverse goals. By evaluating these models before designing an RPC application, you can make informed decisions about whether and how to use object UUIDs, how many server entries to use per server, how to distinguish among instances of a server on a system, whether and how to use groups or profiles or both, and so forth. The two models are the service model and resource model.
The service model views a server exclusively as a distributed service composed of one or more application-defined interfaces that meet a common goal independently of specific resources. The service model is used by applications whose servers offer an identical service and whose clients do not request an RPC resource when importing an interface. Often, with the service model, all the server instances of an application are equivalent and are viewed as interchangeable. However, the service model can accommodate applications that view each server instance as unique. The implications of whether server instances are viewed as interchangeable or unique are significant, so the following topics address these alternatives separately.
More:
Interchangeable Server Instances
Distinct Service Instances on a Single Host