Logical distribution of applications and data

A major goal of distributed system design is to bring data processing functions and data closer to the users than is possible with a centralized design. The expected advantageosf this arrangement, such as local autonomy, responsiveness, etc., have been expressed at length in the literature on distributed data processing. However, we have noted that a nodeo f a distributed system does not function in a totally isolated, stand-alone mode. Other nodes depend upon it or it depends upon other nodes. If these interdependencies are too strong, the sought-after advantages of the distributed data processing system are lost. It is essential, therefore, that such a system be designed so that the nodal interdependencies are at an acceptably low level. We have illustrated one approach to this task.