Article Preview
TopIntroduction
There is a great pace escalation in the introduction of newer technologies in the modern era of computing. One such technology is cloud computing which has been able to attract both, the IT community and the research society in the last decade. The extensive power, high speed processors and the enormous increase in data storage capability have raised the interests of many researchers and motivated them to share the resources on the network. This has led to the emergence of cloud computing. In cloud computing, the resources are provided to multiple users on the on demand and sharing basis. Various services provide by Cloud computing include Infrastructure as a Service (IaaS), Software as a Service (SaaS) and Platform as a Service (PaaS) (Beloglazov et al., 2012). In order to fulfil the demands of the end users, various resources are provided to them in the form of storage, processing and network. This demand has increased exponentially in past years and thus has given rise to the concept of virtualization. Virtualization enables a physical machine host multiple virtual machines (VMs) each having its own operating system so as to optimally utilise the available resources. The concept of Virtualization enables the end users and the service providers to have an efficient utilization of cloud resources with optimum usage and least cost. This virtualization technique is responsible for effectively handling the increasing need of the users in terms of needed resources in Cloud Data Centres (CDC). Various objectives like balancing of load, energy management, consolidation, the sharing of users among multiple users, making the system fault free, can be achieved with the help of virtualization (Noshy et al., 2018) However, the aroused interest of the users in cloud computing has led to a tremendous growth of demand for various cloud resources, making energy consumption a critical issue. The energy demand in hyper scale data centers has increased from 31.11 terawatt hours in 2015 to 76.23 in 2020 and is expected to reach 86.58 by the end of 2021(Energy demand data centers globally by type 2021, 2021). It is predicted that 90% of organizations will have personal data on IT systems they don't own or control and the information technology (IT) sector will consume up to 13% of global electricity by 2030 which is at present 7% (Gartner Inc., 2013). COVID-19 has also resulted in the acceleration of digital businesses, depending upon ruling technologies like cloud computing. According to the recent report by Gartner, the spending on remote working during pandemic will increase by 4.9% in the year 2021 (Costello & Rimol, 2021). This will, in turn, increase the energy consumption in the data centres due to increased workload. One of the main reasons of energy consumption in data centres is the inefficient resource utilization. With lower resource utilization, the energy efficiency of the system will also be low. Also, the number of active hosts will increase, leading to more cooling employments. According to a study by Srikantaiah, optimal resource utilization leads to minimum energy consumption (Srikantaiah et al., 2008). Dynamic VM consolidation has proved to be one of the magical solutions for reducing energy consumption by improving utilization of resources. VM consolidation is best achieved with the help of live VM migration. To minimize the number of the active servers and to save the energy consumption, migration of VMs from overloaded/underloaded servers takes place. Therefore, the process of consolidation comprises of a) finding overloaded/ underloaded servers b) selecting a VM from the server to be migrated c) finding a suitable for the migration of the selected VM, also called the VM Placement. Various researches have been done to carry out the process of consolidation. VM placement has attracted many academic researchers since it is considered an important issue for efficient VM consolidation. Since VM placement is an NP-hard problem(Békési et al., 2000), finding a deterministic solution to this problem is quite difficult. Various heuristic and meta-heuristic VM placement techniques, proposed by different researchers, to solve the problem are discussed in the latter part of the background work. Though it is easy and quick to implement the heuristic techniques, they may fall into local optima. Metaheuristic techniques have been proved to be able to find near optimal solutions to such NP-hard problems (Donyagard Vahed et al., 2019). They have been used in several kinds of researches (Bangyal, Ahmad, et al., 2019) (Pervaiz et al., 2021) to show their importance. In this paper, a Particle Swarm Optimization based VM Placement technique, RAPSO_VMP (Resource Aware Particle Swarm Optimization), has been proposed. The robustness of the PSO algorithm towards the control parameters makes its implementation easy with a high convergence rate. The proposed scheme considers multiple resources like CPU, storage, and memory while trying to optimize many factors like energy efficiency, power consumption, and the overall resource utilization of the system. The contributions of the proposed scheme are as follows: