Optimal Distributed Data Warehouse System Architecture

Abstract

Many organizations look for a proper way to make better and faster decisions about their businesses. Data warehouse has unique features such as data mining and ad hoc querying on data collected and integrated from many of the computerized systems used in organization. Data warehouse can be built using a number of architectures. Each one of the architecture has its own advantages and disadvantages. This paper investigates five different data warehouse architectures: centralized data warehouse, independent data mart, dependent data mart, homogeneous distributed data warehouse and heterogeneous distributed data warehouse and subsequently a comparison plan will be described and then all data warehouse architectures mentioned above will be parametrically measured by this plan. Then an optimal architecture is proposed in which we tried to resolve drawbacks in previous distributed architectures to reach to a desired design. It must not be forgotten, however, that a desired design will not suit every situation in every project, but should be customized for particular environments.