Vertical Class Fragmentation in a Distributed Object Based System

Processing costs in distributed processing environments is most often dominated by the network communications required for interprocess communication. It is well-known from the distributed relational database design research that careful placement of data \near" the users or processors where it is used is mandatory or system performance will suuer greatly. Data placement in relational database systems is comparatively simple because the data is at, structured, and passive. Objects are characterized by an inheritance hierarchy (other hierarchies could also be considered including, class composition and execution), unstructured (possibly dynamic data), and contain a behavioral component that deenes how the \data" is accessed by encapsulating it within the object per se. Algorithms currently exist for fragmenting relations, but the fragmentation and allocation of objects is still a relatively untouched eld of study. Similar to relations, objects can be fragmented both horizontally and vertically. This paper describes the formalisms necessary to reason about vertical fragmentation in a model consisting of simple attributes (no composition hierarchy is considered) and simple method invocations (nested transaction calls are not permitted). The paper then contributes by outlining an algorithm that fragments the classes into a set of \smaller" equivalent classes (actually fragments of the class' extent) that can be placed precisely where they are used. The algorithms will place data so network traac is minimized thereby achieving optimal overall throughput while producing excellent response times for transactions executed locally on a distributed network.