Article Preview
Top1. Introduction
Rapid development in information technology (IT) replaced traditional computing techniques with cloud computing. The Cloud allows consumers to connect many configurable computation assets (computers, memory, networks, apps) to provide 24x7 facilities to customers at pay-per-use rates (Brown, 2011, Buyya, 2010). It also enables resource allocation across the globe to perform various information centers, which allow cost-effective services to both cloud service providers (CSPs) and users (Chiregi, 2016).
As the Cloud offers various services to its users, cloud load balancing is one of the main concerns to the CSPs to avoid overloading scenarios during task estimation. Load balancing provides the capability to divide the burden evenly with the available resources. Thus, load balance aims to reduce the response time for tasks and optimize resources, which increases system performance at a lower cost. Further, load balance's objectives are to reduce energy consumption and carbon emissions, avoid bottlenecks, supply resources and meet the QoS requirements for load balance. Global research groups are interested in the design and development of the best possible methods for resource allocation. Hence the present study and assessments focus on them. Cloud load balance is a method of distributing the load on unbundled virtual machines to improve device flow. Numerous difficulties alongside load managing include asset programming, tracking efficiency, QoS Management, power usage, and internet storage accessibility (Kaur, 2012; Malladi, 2015). This paper provides an extensive examination of the multiple kinds of cloud planning, load balancing, and task sharing methods.
Ghomi et al. performed a comprehensive evaluation of charge equilibrium in memory, dividing the study into six classifications: Hadoop MapReduce charge balance methods, Natural charge equilibrium phenomena methods, agent-based charge balancing methods, General charge balancing methods, Network-aware job planning, and count processing methods. They primarily focused on Hadoop MapReduce and Efficiency of Energy, which affect cloud count balancing. Nevertheless, their job requires task and load balancing dependent on the cluster (Ghomi, 2017).
Milani and Jafari examined and categorized multiple current load balance systems into vibrant and hybrid subdomains. The behavior, disadvantages, and difficulties of these methods were defined based on the various parameters. They have identified challenges in developing more efficient algorithms to reduce resources and energy consumption and increase the efficiency of load balancing technologies. However, they have not discussed task-based load balance, cluster-based load balance, and energy consumption problems (Milani, 2016). Singh et al. provided the comprehensive evaluation of algorithms for metaheuristic workflow planning, recognized numerous problems relating to cloud job planning, and reported a comparative assessment based on the meta-heuristic strategy for both dependency and independence (Singh, P., 2017). Singh and Chana reported six distinct views: planning workload, tracking, QoS necessity, implementation layout, auto-management of workload, and assessed automated cloud resource management (Singh and Chana, 2015). Ivanisenko and Radivilova presented a comparative assessment of the load-balancing techniques centered on metrics (response time, relocation moment, scalability, and asset usage). However, they missed out on the problems in current technology and difficulties and potential developments (Ivanisenko, 2015). Katyal and Mishra have assessed the load-balancing algorithms centered on SLA user requirements for different cloud settings. They address benefits, disadvantages, and difficulties in the primary classifications of current methods (Katyal, 2014). However, they did not assess the basis of varying load balance parameters. Hence, there is a need to define the taxonomy of load balancing techniques in cloud scenarios. The paper reports a detailed taxonomy of load balancing techniques in a cloud scenario.