Article Preview
TopIntroduction
Living in an increasingly advancing dematerialized world with more organization dependency on Cloud services has allowed entering into an advanced technological era. With the adoption of the Cloud as the major contributing technology by leading companies, the Cloud can be defined as an epitome of several technologies that include parallel computing, distributed computing, ubiquitous computing, utility computing, grid and pervasive computing (Mohiuddin & Almogren, 2019; Ijaz et al., 2021). The Cloud serves as the most promising distributed-domain methodology with the pay-per-use provision and on-demand resources for large-scale complex and scientific applications (Dang et al., 2019; Beloglazov et al., 2012). But because of this large-scale deployment, big data and dependency on the cloud, have raised certain security as well as privacy concerns in the cloud environment (Gupta et al., 2018). With consideration of the dependency and deadline constraints of the applications, the Cloud has become the most reliable deployment platform in contrast to the traditional grid and other high-performance computing services. It provides a customized environment where scientific computation is performed by the provision of the desired resources in advance to several users by employing the concept of Virtualization.
Virtualization could be well understood as the multiplexing of resources such that multiple users could be allowed to access the resource as per their requirements. Thus, the Cloud has several data centers for service provision, consisting of multiple physical machines, generally termed host machines, that turn millions of VMs for workflow execution. These workflow applications are complex, burdensome, and challenging, with many tasks requiring a considerable amount of real-time data and consuming considerable energy (Xiao et al., 2019). Cloud data centers alone contribute to 20% of total contributed energy in the digital world (Aziza & Krichen, 2020), as 94% of the workflow execution is carried out at the data center (Alboaneen et al., 2021). The more the computation demand for the workflow execution, the more will be the energy consumption leading to alarming environmental hazards of massive CO2 emission and heat dissipation (Mohanapriya et al, 2018).
The energy consumption also increases the operational and maintenance cost of the server. Cloud data centers adverse impact, i.e. high energy consumption, has attracted researchers to contribute to green computing by proposing approaches to cut back on energy consumption and promote resource utilization (N. Garg et al, (2021)) i.e. how to make maximum utilization with minimum hosts for workflow execution. The workload scheduling at the hosts is discussed using several scheduling algorithms (Data-Aware based (Tao et al., 2015), Greedy approach, Heuristic, Meta-heuristic (Askarizade et al., 2019), which is later checked for any load imbalance. In case the load is not balanced, VM consolidation, i.e. based on the live migration of VMs between the hosts (Mohanapriya et al, 2018; Li et al., 2019; Lin et al., 2019). Research works (Mohiuddin, & Almogren, 2019; Xiao et al., 2019; Li et al., 2019; Lin et al., 2019, Wu et al., 2016, Singh & Sindhu, 2020; Khaleel & Zhu, 2016; Zharikov et al., 2019; Nasim et al., 2018; Shaw et al., 2020; Sharma et al., 2019; Pyati et al., 2020; Khattar et al., 2020) have contributed to define several VM consolidation schemes for load balancing and energy-efficient computation.