Optimizing services selection in a cloud multiplatform scenario
Everton Cavalcante, Thaı́s Batista, Frederico Lopes, Flávia C. Delicato, Paulo F. Pires, Noemi de La Rocque Rodriguez, Ana Lúcia de Moura, Reginaldo Mendes · 2012
Services selection is an important challenge for applications that use a composition of services provided by different cloud platforms. This paper presents an optimized cloud services selection approach that evaluates each alternative set of services that composed an execution plan. This approach considers cost and quality parameters for each cloud service in the execution plan, and excludes coincident services in the calculations. Coincident services are those services present in all available execution plans and therefore equally contribute in the calculations encompassing costs and quality parameters regarding these plans. By excluding coincident services, the service selection process by itself should be performed more quickly since it involves a smaller number of services, but with the trade-off of running an additional algorithm for identifying the coincident services among the available ones. In order to evaluate such trade-off and illustrate the proposed approach, we present a case study and an experimental evaluation that compares our new approach with a previous one that considers all services that compose all available execution plans.