Putting OO Distributed Programming to Work

stractions underlying distributed computing. We attempted to keep our preaims at an abstract and general level. In this column, we make those claims more concrete. More precisely, we describe OO distributed programming through an exercise consisting of abstracting and factoring out a fundamental component of a distributed system: failure detection. 1 Adopting a distributed architecture for a given application might be driven by various motivations. One might adopt a distributed solution for an application made of inherently distributed components or decide to distribute a centralized application to take some advantage of distribution. Among the distribution advantages are resource sharing, load balancing, and fault-tolerance. However, these advantages have a dark side, and, to paraphrase L. Lamport, a distributed system is also one that stops you from completing any work because of the crash of a machine you have never heard about. The notion of partial failure is a fundamental charact...

• Type: Journal
• Year: 1999
• Where: CACM
• Authors: Pascal Felber, Rachid Guerraoui, Mohamed Fayad
• Keywords: CACM 1999, Distributed Computing, Distributed Systems, Failure Detection