Abstract

Abstract State Machines (ASMs) encourage highlevel system specifications without forcing the development into the "formal methods straightjacket ". This makes them an ideal formal method for applications in areas, where otherwise only semi-formal methods are used. One such area is the development of data warehouse and on-line analytical processing (OLAP) applications to which this article contributes. Based on an ASM ground model for data warehouses we show which problems have to be solved in the case of distribution. This mainly amounts to making decisions on materialised views. In this article we develop simple refinement rules for this purpose. Then we develop a cost model that combines the costs of query processing with the maintenance costs arising from redundancy in the local data warehouse fragments. This cost model indicates, whether it is advantageous to apply a refinement rule or not. However, as the refinement process is non-deterministic, there is no guarantee that a global cost optimum will be reached. Keywords. Abstract State Machine, Data Warehouse, Distribution, Cost Model, Refinement 1

Citations