مقاله انگلیسی رایگان در مورد GSA ابری برای زمانبندی بار در محاسبات ابری – الزویر ۲۰۱۸

Scheduling of load and data plays an important role in the efficient utilization of the resources from one cloudlet to another cloudlet in the cloud computing environment. Cloud computing is an incremental paradigm to brighten the world with its great vision of providing the power of distributed computing through virtual approach. Resource allocation plays an important role in the optimal handling of the load scheduling problem using static and meta-heuristic approaches. The Gravitational Search Algorithm (GSA) is a nature-inspired meta-heuristic optimization technique which is used for solving the load scheduling problem in the cloud computing environment and is based on Newton’s gravitational law dealing with gravity. This paper proposes a near optimal load scheduling algorithm named Cloudy-GSA to minimize the transfer time and the total cost incurred in scheduling the cloudlets to the VMs. These are achieved by increased exploitation of VMs using the particles based on fitness values. The Cloudy-GSA algorithm is implemented on the CloudSim and has been compared with the existing popular algorithms. The results of the algorithm are converged and statistically analysed over a set of iterations. As evident from the results, the proposed Cloudy-GSA algorithm minimizes the transfer time and the total cost for scheduling the load than the existing algorithms.

Introduction

The cloud computing is one of the increasing domains in the area of distributed and grid computing using the concept of virtualization. The computing refers to the processing of the tasks on the virtual machines in the system for the efficient functioning of the tasks. The computational power of networks is increasing day by day to make the world follow and know about the massive hidden potential present inside it. The cloud acts as a repository of the resources enabling the users with large capabilities and computing facilities like storage, processing, extraction and retrieval of the information dealing with a large number of heterogeneous tasks to produce better resource access to the users. The cloud is based on the pay-as-you-go or pay-as-you-use model for accessing the resources. Buyya et al. [1] state that the cloud provides resource and task scalability, on time resource execution, dynamic provisioning, fault tolerance and interoperability of the resources. It provides dynamic allocation of the cloudlets/ tasks to the virtual machines. The functionality of dynamic allocation is achieved by load scheduling of the cloudlets in a near optimal manner in the cloud. This is performed to achieve greater throughput, lesser execution and waiting time, less transfer time, and less cost of computation [2]. The load scheduling is defined as the process of providing, allocating and balancing of the load (tasks/ cloudlets to the virtual machines) in the cloud system efficiently. The main purpose is to reduce the transfer time and the total cost incurred for scheduling the load in the system [3]. The scheduling of the load is performed using various scheduling algorithms. The scheduling algorithms are specified on the basis of nature as static and dynamic algorithms. These are also classified as heuristic and non-heuristic algorithms. The meta-heuristic algorithms play an important role in scheduling the load by using a search mechanism. Vecchiola et al. [4] elaborate the problem solving approach having multiple objectives in the cloud. This method provides solution using an optimization strategy. The load scheduling based on the swarm intelligence methods provides a larger significance in the environment as they involve the real world behaviour of the swarms. A group of objects, particles, ants etc. following the mechanism for locating the food is followed to find the best solution in particle swarm optimization and ant colony optimization mechanisms. The physical laws of gravity are used for finding the near optimal solution among the group of objects. The scheduling of load in the cloud is an important problem that needs to be resolved using the more efficient algorithms than the existing algorithms like Segmented Min-Min, Tabu Search, Simulated Annealing, Genetic Algorithm, FCFS, PSO etc. [5]. In this paper, Newton’s law of gravity is used for performing the scheduling of load.