An Effective Low-Cost Cloud Service Brokering Approach for Cloud Platforms

Abstract

Cloud computing is characterized by efficient provision of services and access to shared resources through its pay-per-use model. The service brokers in this architecture mediate among cloud users and cloud service providers to redirect requests to appropriate data centers while aiming at minimization of response time and monetary cost for cloud users. For selection of data centers, the optimization techniques incur large overhead during execution of applications. Consequently, the recent brokering approaches resort to heuristics which do not guarantee optimal solutions in terms of response time and cost that are pivotal for executing compute intensive applications in a cloud environment. In this paper, we propose multi-objective service brokering with availability-based load balancing (MOSB_ALB) approach that minimizes response time and monetary cost for efficient low-cost provision of services in a cloud environment. The MOSB_ALB approach performs static and dynamic selection of data centers while using availability-based load balancing for distributing load among virtual machines. The static computation of data center index incorporates MOEA/D algorithm and uses z-score values corresponding to indexes in optimal solutions. The dynamic computation of data center indexes uses criteria based on weights and allocation counts. The experimentation performed through a large number of configurations shows that the MOSB_ALB approach outperforms existing well-known cloud service brokering approaches by improving cumulative response time with a speedup factor of 2.19, along with a significant reduction of monetary cost.