Role-based intelligent application state computing for OpenFlow distributed controllers in software-defined networking

摘要

Software-defined networking, in which a network is programmable and controlled with soft computing techniques, is widely used in various network testbeds. To satisfy the demands of large networks, control planes are constructed with distributed controllers, which are a routine aspect of OpenFlow research. Distributed controllers are scalable and fault tolerant; thus, they can replace centralized controllers in large-scale networks. In the foreseeable future, there will emerge more applications based on soft computing techniques. Because there is scant research on application management, a distributed controller with a role-based mechanism for properly managing applications and their states based on their properties has yet to be developed. Thus, in this study, we propose a systematic approach for classifying applications according to their roles to dynamically deploy applications and their states. Both applications and their states are managed based on their properties, including CPU, memory and network bandwidth. This intelligent mechanism, which computes the overhead of applications, provides a compromise between storage and bandwidth usage in OpenFlow distributed controllers. We propose a hierarchical system to differentiate applications and design a controller module for dynamically determining the status of an application.