Abstract
Server virtualization and cloud based data centre solutions are becoming popular for effective resource utilization, better manageability and cost reduction. However, advantage of maximizing resource utilization may elevate performance issues due to poor resource allocation mechanism such as network bandwidth. The inefficient provisioning of network bandwidth to application results in performance degradation, despite having sufficient processing and memory resources. Traditional approaches prioritize network traffic on the basis of application type and allocate the network bandwidth accordingly. In this paper we propose an OpenFlow based adaptive QoS framework for efficient network bandwidth allocation, which monitors the application's load in a real time and allocate network bandwidth dynamically according to application popularity. As a result, sufficient network bandwidth is allocated to popular applications. The performance evaluation on a real test bed shows that by using our proposed framework, maximum number of user sessions associated with popular application can meet their QoS targets, which can eventually improve the end user satisfaction level.