Evaluation of virtualized osmotic cloud network using discrete event Branch-and-Bound heuristics

Abstract

During heavy traffic transactions on a network, there is an extreme impact on network resources such as bare metals, routers, I/Os, enterprise applications and services. The arrival rates usually exceed service rates despite Ethernet TCP windowing and layer-3 fair queuing. This is unacceptable in Osmotic computing paradigm which houses machine to machine cloud elasticity especially for IoT plug and play edge devices. This research proposes Fog and server consolidation (SC) in distributed high performance data centers (D-HPDCs) as a middleware solution to solve the problem. A candidate scheme called green virtualization is used to share the capabilities of physical servers (bare metals) by splitting resources among operating systems that host an on-demand smart grid application. AType-I virtual machine hypervisor is used to mitigate the various pitfalls in legacy datacenters (DCs) such as resource availability, query response time, scalability and network quality of service (QoS). MATLAB SimEvent simulation is used to realize a case-based Branch-and-Bound heuristics as an optimization scheme in the green virtualized DC. Within the elasticity zone, performance evaluation between Type-I virtualized and non-virtualized DC is carried out. The benefits of virtualized DCs using cloud elasticity design framework is highlighted. The results showed that the Type-1 virtualized DC scenario had better QoS performance metrics compared with the non-virtualized DC. Consequently, SC using virtualization is shown to be efficient for supporting multi-constrained DC environments.