The Federated View Selection Problem

Database federation is becoming an increased reality for business and clinical organizations due to rapid data accumulation, making a single, monolithic data warehouse difficult to achieve. A commonly used technique to decrease query execution time in data warehousing environments is to precompute and store beneficial views. Known as the view selection problem, this process seeks to identify these views subject to resource constraints. A federated model, however, has yet to be developed. In this work, we submit such a model, the federated view selection problem, and its constituent elements: an optimization function, search lattice, and suite of heuristics adapted from state-of-the-art single and distributed configurations. The notion of restricted materialization is also defined to deal with read-only situations where policy, security, and/or legal concerns disallow data storage across the federated network. This condition significantly diminishes the number of valid solutions in the search space, necessitating a novel twophased heuristic approach. As this is the initial work in this area, the article focuses on definitions, various search methodologies, and the application of heuristics to provide a practical aspect to the theoretical model. Keywords—Data warehouse, federation, heuristics, optimization, view selection problem.