MTLBP: A Novel Framework to Assess Multi-Tenant Load Balance in Cloud Computing for Cost-Effective Resource Allocation

The recent advancement in cloud computing enabled technologies to offer scalable and elastic computing components at a reduced cost to manage the IT infrastructure organizations effectively. It addresses the cost optimization problem, which arises in the context of traditional web-enabled computing models where bottle-neck conditions arise owing to managing high-scale dynamic clients along with the flexible infrastructure to map heterogeneous/complex data traffic with storage elements associated with appropriate server units. The trend of research evolution has given birth to the conceptualization of cloud computing, which offers storage, infrastructure, and platform as services with a virtualized mechanism to deal with uncertain complex and high-scaled data traffic irrespective of application domains. It has got a wide range of applications starting from health-care to service industries. Hence, it has become de-facto for any digital and information technology (IT) oriented application design and deployment aspects. Managing complex data flow in the context of the cloud is a highly challenging task, and the principle of multi-tenancy has come up as a solution approach to enhance resource sharing among cloud clients with isolated virtualization and application instances with services. The prime agenda here is to reduce the cost of computation and cloudlet handling time. The study introduces a unique load-balancing approach, namely, multi-tenant load balance (MTLBP), to cope with the service-oriented architecture of multi-tenancy in the cloud environment. It also ensures that the end client user can perform their task execution efficiently on the multi-tenant environment. The experimental approach adopts a behavioral traffic analysis, where the formulated approach's performance is compared with the traditional baseline approaches. The outcome obtained for the proposed approach found quite promising in terms of resource utilization and task execution in comparison with the other models. It shows that the formulated MTLBP approach outperforms the baseline approaches with approximately 26.83% when small value cloudlet processing time is concerned and 33.625% when medium volume cloudlet processing time is concerned and approximately 62% when large volume cloudlet handling time is concerned. It is also observed that for small volume cloudlets, MTLBP attains approximately 12.55% performance improvement on average when resource utilization cost is concerned. For medium and large volume cloudlets, it attains approximately the performance improvement of 9.85% and 3.14%, respectively.