MOBY: An Architecture for Distributed Expert Database Systems
Jonathan Bein, Roger King · 1987
In thii paper, we consider MOBY. a distributed architecture to support the development of expert database systems in a rule based language. It combines standard indexing and horizontal data partitioning techniques with a rule based interpreter to achieve the reasonable performance. The major difficulty in developing this architecture is to maintain a high effective parallelism as the number of processors increases. Analytic results suggest that when data is masonably well balanced across a local area network, MOBY has a high effective parallelism. Simulation results support this claim by showing that the effective parallelism is proportional to 40 % of the number of processors. A discussion of some crucial issues in our current network based implementation is also given.