Vertical Fragmentation Design of Distributed Databases Considering the Nonlinear Nature of Roundtrip Response Time

One of the challenges of applications of distributed database (DDB) systems is the possibility of expanding through the use of the Internet, so widespread nowadays. One of the most difficult problems in DDB systems deployment is distribution design. Additionally, existing models for optimizing the data distribution design have only aimed at optimizing query transmission and processing costs overlooking the delays incurred by query transmission and processing times, which is a major concern for Internet-based systems. In this paper a mathematical programming model is presented, which describes the behavior of a DDB with vertical fragmentation and permits to optimize its design taking into account the nonlinear nature of roundtrip response time (query transmission delay, query processing delay, and response transmission delay). This model was solved using two metaheuristics: the threshold accepting algorithm (a variant of simulated annealing) and tabu search, and comparative experiments were conducted with these algorithms in order to assess their effectiveness for solving this problem.