An Evolutive Scoring Method for Cloud Computing Provider Selection Based on Performance Indicators

Application of deterministic, stochastic, and hybrid methods for cloud provider selection

“…The individual is coded as a vector using a binary encoding, where the number of coded variables is equal to the amount of CPs registered in the CP database. Each variable uses a position in the binary vector, and it is represented by a single bit, indicating whether that CP belongs to the solution set (bit 1) or not (bit 0) [17]. All CPs with bit 1 will be called "employed CPs" for that individual, i.e., CPs that will be effectively used in the solution set.…”

“…The penalty is calculated proportionally as a function of how much the solution Ind infringes on the problem constraint. For the CPS problem, there are two constraints [17]: the individual must not be all encoded with bit 0, and the employed CPs must maximize attendance of all user requested PIs. The first constraint prevents functions that generate the initial random population and the genetic operators (modify individual's codification vector) from generating an encoding vector filled with bit 0 for any individual.…”

“…Unlike our previous works concerning the CPS problem[15][16][17], this paper brings up other selection methods, an entirely new set of experiments, new analysis, and exciting findings. Hence, the proposal of building a CPS approach offering several other cloud provider selection methods beyond previously developed ones makes this paper a significant new contribution to the community.2 criteria/attributes/performance indicators 3.…”