A Predictive Priority-Based Dynamic Resource Provisioning Scheme With Load Balancing in Heterogeneous Cloud Computing

Abstract

Listen

Translate article

In cloud computing, resource provisioning is a key challenging task due to dynamic resource provisioning for the applications. As per the workload requirements of the application’s resources should be dynamically allocated for the application. Disparities in resource provisioning produce energy, cost wastages, and additionally, it affects Quality of Service (QoS) and increases Service Level Agreement (SLA) violations. So, applications allocated resources quantity should match with the applications required resources quantity. Load balancing in cloud computing can be addressed through optimal scheduling techniques, whereas this solution belongs to the NP-Complete optimization problem category. However, the cloud providers always face resource management issues for variable cloud workloads in the heterogeneous system environment. This issue has been solved by the proposed Predictive Priority-based Modified Heterogeneous Earliest Finish Time (PMHEFT) algorithm, which can estimate the application’s upcoming resource demands. This research contributes towards developing the prediction-based model for efficient and dynamic resource provisioning in a heterogamous system environment to fulfill the end user’s requirements. Existing algorithms fail to meet the user’s Quality of Service (QoS) requirements such as makespan minimization and budget constraints satisfaction, or to incorporate cloud computing principles, i.e., elasticity and heterogeneity of computing resources. In this paper, we proposed a PMHEFT algorithm to minimize the makespan of a given workflow application by improving the load balancing across all the virtual machines. Experimental results show that our proposed algorithm’s makespan, efficiency, and power consumption are better than other algorithms.