الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 108 |  |  |
| | | from | 108 | | |
Abstract
As SDN technology is emerging and becoming popular in the field of computer networks management, the need for evolvable SDN-based network management systems arises. One of SDN challenges is the lack of a standard SDN north bound interface, which provides an abstraction to lower levels through an easy-to-use high-level interface without getting involved in difficult and complex programming tasks. Higher level abstractions in SDN have been proposed, yet there is space for improvement and innovation. We propose NWManager, an SDN concern-based network management framework. It provides high level abstraction in SDN management layer through the separation of network concerns. It translates a set of input keywords into policies based on network management function, such as behavior monitoring, QoS and security and network concerns, namely, application, communication and resources. Our framework is hybrid; proactive as it inserts OpenFlow rules to switches based on defined policies and reactive where it handles events according to the defined policies. NWManager is not limited to certain network services, and it provides various network management functions in one system. Our conducted network simulation scenarios by Mininet show our framework’s functionality and the high-level policies the framework can support converting them to low level flow rules they also demonstrate that the introduced abstraction and separation of concerns in management layer facilitates and enhances the management of the network through separation of concerns thus dividing the network management problem to smaller easily less-complex solved subproblems. As well, it extends the network control parameters provided to the network administrator. Additionally, it helps closing the semantic gap between the intents of network administrators and the languages in which those intents can be encoded. As a result, it helps achieving network services and applications’ requirements. In addition, the performance is measured through calculating the different processing times.