Summary

Chapter one we have evolved a view of Open Distributed Systems that is more subtle than simple distributed services or distributed operating systems. We have introduced the ideas Properties and Design Freedoms of Open Distributed Systems that enable the systems designer and implementor to identify the objects required in a system from its specification, and to make engineering and other choices in the design and implementation. Few computing services (from the computational viewpoint) inherently require distribution from the computational point of view alone. Indeed, only those highly fault tolerant algorithms usually employed for actually managing a network itself (e.g. routing algorithms, fault isolation algorithms) are o themselves distributed. However, it is clear from the other viewpoints (enterprise and informational) give us justifications for distribution of services. Computing facilities, personnel and information may already exist at more than one site. Engineering reasons may constrain some facilities to exist closer to one facility than another (e.g. oil well monitoring closer to the well than to the magnate's office). We have argued that Open Distributed Systems with publicly available interfaces are the correct approach to building solutions to these types of problems.