Optimal provisioning and scheduling of analytics as a service in cloud computing

We propose an analytic request-centric multiobjective provisioning and scheduling method based on hybrid particle swarm optimization, called analytics-as-a-service (AaaS) provisioning and resource scheduling (APARS) system, to optimally provision analytics service and schedule cloud resources by taking into account of user specified constraints. Analytics-as-a-service provisioning and resource scheduling is employed to maximize accuracy of the analytic service and minimize the execution time and cost of the analytic request by combining multiobjectives with the constraint of minimal deadline and budget. Provisioning AaaS and scheduling cloud resources for analytic request is challenging as the analytic request requires scaling up and down of cloud resources dynamically. Thus, this domain invites a rigid system that can provide optimal analytics using cloud services for specified quality of service parameters. Our proposed system can handle diverse set of analytic requests with the goal of minimizing execution time, cost, critical performance, load balancing level and maximizing resource utilization, AaaS utilization, and success rate. The proposed APARS system is tested and evaluated under different use cases and compared with other metaheuristic algorithms, and we found that the results to be convincing our claims. The experimentation is carried out for various scenarios and the proposed APARS is evaluated with other metaheuristic algorithm. Simulation results validate the APARS system and show that the proposed method performs better than other existing methods with various use cases.