Análise de decisão multicritério de cenários alternativos para o sequenciamento de tarefas em computadores em datacenter

Abstract

In current data centers, processing routines are scheduled to perform in a given period of time and, in case of delay, make it impossible to conduct the activities of everyday life of the various sectors of the company. Even with modern servers, due to the large volume of data to be processed, it is possible that certain computers are overloaded causing a drop in performance, a longer time to implement and consequently delays in the delivery of results. In this case, alternative scenarios can be created to sequence the tasks again. This work aimed to apply the multi-criteria decision analysis method on the results of the simulation of alternative scenarios for sequencing tasks on computers in a datacenter, in order to overload the resolution of problems. To achieve this purpose have been developed simulation models of task sequencing scenarios and replacement of computer overloaded. The analysis of simulation results was made based on three criteria, namely the Total Processing Time (makespan), the Total delay time, and the number of Delayed Tasks. Subsequently, the results were submitted to analysis method of AHP multicriteria decision (Analitycal Hierarchy Process), allowing you to choose the best setting according to established criteria. It was concluded that the modeling and computer simulation, combined with multi-criteria decision analysis method, can aid in decision making, as it enables sequencing to identify the best solution possible alternative scenarios in front of a computer overload situations.