Vashistha & Verma International Journal on Emerging Technologies 10(3): 10-15(2019) 10 International Journal on Emerging Technologies 10(3): 10-15(2019) ISSN No. (Print) : 0975-8364 ISSN No. (Online) : 2249-3255 Economic-Driven Strategies for Virtual Machine Allocation in Cloud Data Center Avneesh Vashistha 1,2 and Pushpneel Verma 1 1 Department of Computer Science & Engineering, Bhagwant University, Ajmer, (Rajasthan), India. 2 Department of Information Technology, IMS Ghaziabad, (Uttar Pradesh), India. (Corresponding author: Avneesh Vashistha) (Received 25 May 2019, Revised 02 August 2019 Accepted 18 August 2019) (Published by Research Trend, Website: www.researchtrend.net) ABSTRACT: In the cloud environment, applications have different requirements and priorities. These applications require the dynamic provision of resources into different types of virtual machines based on the priority requirements. In general, fixed price model is used for allocating the virtual machine to an end-user that may not support an optimal or economic-driven allocation. In this paper, we describe economic-driven techniques for VM allocation and classified these techniques based on specific characteristics required by VM in the cloud. We introduce technical debt as a novel approach for VM allocation and maps the concept of technical debt into the context of VM allocation. Furthermore, we discuss some critical situations that incurred technical debt while operating VMs in the cloud data center. Keywords: Virtual Machine, Dynamic Allocation, Cloud Computing, Economic Strategies, SLA, Computing Resources, Technical Debt I. INTRODUCTION Cloud computing is an on-demand and dynamically scalable computing platform that provides network- enabled resources such as storage, servers, networks, databases, and software services, etc. However, modern computing platform fosters the economically driven service model while delivering these resources in the cloud market. In the cloud environment, several Virtual Machines (VMs) with varying capacity could be instantiated on-demand by a Physical Machine (PM). For optimizing a VM essential network characteristic such as bandwidth and latency required for communication between VMs may lead to communication cost and significant delays [1-2]. Besides Service Level Agreement (SLA) and Quality of Service(QoS), but some other parameters for VM allocation have already been considered in earlier studies including energy consumption, performance, minimize execution time, cost reduction, load balancing, network delays, congestion, and service downtime [3,4, 20]. In cloud environment, each application has different requirements or resource priorities for performing time- dependent tasks. Such application requires dynamic provisioning of resources while instantiating a VM. Moreover, applications are developed and deployed on the multi-tenant architecture which facilitates Shared Resouces as a Service (SRaaS). In this architecture pattern, several users can share the same resource instance on different levels such as application, databases, and VM, etc. For example, several users are participating in a globally accessedSaaS survey application. Since SaaS applications have multi-tenancy in nature and hosted on different VMs located on different locations that require communication among them. In this case, network bandwidth scarcity is a major factor for VM allocation. Also, VMs communication cost influences overall performance. Further, we may consider a situation where a request has been made for a large capacity VM but service provider fails to provide it immediately just because of currently available individual VMs can not fulfill the required capacity. In this situation, we may approach for joint VM provisioning or server consolidation. Moreover, a VM is allocated based on job characteristics. For each job, a different kind of strategy may be implemented for VM allocation. For example, profit and response time are key parameters for utility-based applications; QoS, throughput, and response time parameters are required for SLA based applications.In literature, several research studies have shown that most of the VMs in data cloud center severely under-utilized because of over-provisioning under peak demand that affects the revenue and make the operating environment sub- optimal from VM execution point of view [7-9]. In general,both cases, under-utilization or over-utilization leads the problem of sub-optimal utilization of a VM capacity.A Virtual Machine leads the debt whenever operated sub-optimally in the cloud environment and reasons could be the strategic, managerial, or even unintentional.For addressing these problems, we propose a technical debt approach for VM allocation in the Cloud. Technical debt could be the results of non-strategic or inappropriate engineering decisions that affect the utility of underlying computing resources and leads sub- optimal utilization of VM in the cloud data center. Besides, VM could be inevitably operated under dynamic changes in requests workload generated by several users in the cloud environment, and consequently, encounter the problem of under/over- utilization of VM; for example, VM under-utilization could be linked with a situation where VM provides more computing resources than the demands of users and going into debt as the cost of unused resources over the underlying VM. In this case, technical debt denotes the cost of engineering efforts required for maintaining optimal utilization of VM plus accrued interest over the technical debt. On the other hand, VM over-utilization could be the consequences of a high volume of requests received on the VM and in response; VM e t