Novel Service Broker and Load Balancing Policies for CloudSim-Based Visual Modeller

Abstract

Cloud computing enables numerous possibilities to deploy and host applications for application developers without any concern for service demand, installation, and maintenance. The developers can obtain the desired services from the cloud service providers (CSPs) and can pay on a usage basis. However, each CSP offers different pricing schemes, locates in different regions, and provides a different set of services. Therefore, it is challenging for the developers to select the CSPs. As a solution, the Cloud Analyst tool was developed to support large-scale applications for developers. It provides three service broker policies, namely closest data center (CDC), optimize response time (ORT) and reconfigure dynamically with load (RDL), and three load balancing policies, namely round robin (RR), equally spread current execution load (ESCEL) and throttled (TH). These policies map the user bases (UBs) to the data centers (DCs) and further map to the virtual machines (VMs) of the DCs, respectively. The performance of these policies is assessed in terms of four measures, namely load assignment, overall response time, DC request servicing time, and total cost. In this paper, we propose one service broker policy, one VM provisioning policy, and one load balancing policy and compare their performance with existing service broker policies and existing load balancing policies, respectively. Moreover, the proposed policies are incorporated with the existing policies to improve their performance in terms of four performance measures and validated using Cloud Analyst.