Resumen
The article focuses on failure detection. Identifying the fundamental abstractions in distributed computing and classifying these abstractions according to some inheritance hierarchy lies at the heart of Object Oriented distributed programming. The aim is not to describe a specific failure detection protocol but rather to sketch a modular architecture to compose and customize failure detection protocols according to the topology of the system and the communication pattern of the application. The reduction of computer network traffic, especially when a lot of monitorable objects and clients are involved, is the main reason for the good scalability of this hierarchical approach. The modularity of these services would be increased if failure detection were encapsulated in a separate component. It is also important to encapsulate the complexity of failure detection inside first-class (application level) components with well-defined interfaces, namely first-class objects