Fighting the politics with technology

Abstract

In this position paper we agree with the thesis of the workshop that a significant body of experience in the design and implementation of distributed systems has been developed. This makes us wonder why the concepts that have been developed are so little used in the world of real systems. Our view is that so far the distributed systems community has paid insufficient attention to the nature of the organizations that use distributed systems and the impact. We argue that several of the paradigms for distributed computing will fail when tackling the problem of interworking between heterogeneous systems under the control of autonomous administrative domains. In particular we argue against architectures dependent upon global naming, access control and interaction models based upon a data communication paradigm.When distribution spans organizational and technology boundaries it becomes necessary to consider the overall structure as an evolving federation of cooperating autonomous systems: the federation is not itself a distributed system in the sense of being a superior naming, management and security domain. We are content to be accused of describing distributed unsystems, since we feel this is the basis for successful distributed computing.This paper is based on our experiences in developing the ANSA architecture in the context of the ESPRIT ISA Project.