Autonomic management of software defined networks : DAIM can provide the environment for building autonomy in distributed electronic environments - using OpenFlow networks as the case study

Publication Type:
Thesis
Issue Date:
2016
Full metadata record
Files in This Item:
Filename Description Size
01front.pdf101.63 kB
Adobe PDF
02whole.pdf10.82 MB
Adobe PDF
Next generation networks need to support a broad range of services and functionalities with capabilities such as autonomy, scalability, and adaptability for managing networks complexity. In present days, network infrastructures are becoming increasingly complex and challenging to administer due to scale and heterogeneous nature of the infrastructures. Furthermore, among various vendors, services, and platforms, managing networks require expert operators who have expertise in all different fields. This research relied on distributed active information model (DAIM) to establish a foundation which will meet future network management requirements. The DAIM is an information model for network solutions which considers challenges of autonomic functionalities, where the network devices can make local and overall network decisions by collected information. The DAIM model can facilitate networks management by introducing autonomic behaviours. The autonomic behaviours for communication networks lead networks to be self-managed and emerge as promising solutions to manage networks complexity. Autonomic networks management aims at reducing the workload on network operators from low-level tasks. Over the years, researchers have proposed a number of models for developing self-managed network solutions. One such example is the common information model (CIM), which is described as the managed environment that attempts to merge and extend the existing conventional management and also uses object-oriented constructs for overall network representation. However, the CIM has limitations coping in complex distributed electronic environments with multiple disciplines. The goal of this research is defined as development of a network architecture or a solution based on the DAIM model, which is effectively distribute and automate network’s functions to various network devices. The research first looks into the possibilities of local decision-making and programmability of network elements for distributed electronic environments with an intention to simplify network management by providing abstracted network infrastructures. After investigating and implementing different elements of the DAIM model in network forwarding devices by utilising virtual network switches, it discovers that a common high-level interface and framework for network devices are essential for the development of network solutions which will meet future network requirements. The outcome of this research is the development of (DAIM OS) specification. The DAIM OS is a network forwarding device operating system which is compliant with the DAIM model when it comes to network infrastructure management and provides a high-level abstracted application programming interface (DAIM OS API) for creating network service applications. Through the DAIM OS, network elements will be able to adapt to ever changing environments to meet the goals of service providers, vendors, and end users. Furthermore, the DAIM OS API aims to reduce complexity and time of network service applications development. If the developed DAIM OS specification is implemented and if it functions as predicted in the design analyses; that will result in a significant milestone in the development of distributed network management. This dissertation has an introduction in chapter 1 followed by five parts in order to draw a blueprint for information model as a distributed independent computing environment for autonomic network management. The five parts include lending weight to the proposition, gaining confidence in the proposition, drawing conclusions, supporting work and lastly is appendices. The introduction in chapter 1 includes motivations for the research, main challenges of the research, overall objectives, and review of research contributions. After that, to lend weight to the proposition as the first part of the dissertation, there is chapter 2 which presents the background and literature review, and chapter 3 which has a theoretical foundation for the proposed model. The foundation consists of a generic architecture for complex network management and agents to aggregate distributed network information. Moreover, chapter 3 is probably more about a state of the art in software engineering than about real implementation to engineer autonomic network management. The second part of the dissertation is to gain confidence in the proposition which includes attempting to implement the DAIM model in chapter 4 with some tests to report good performance regarding convergence and robustness for the service configuration process of network management. Also, the second part has a specification of true abstraction layers in chapter 5. The specification of true abstraction layers proposes a high-level abstraction for forwarding networking devices and provides an application program interface for network service applications developed by network operators and service providers. The implementation in chapter 4 is supported by the fourth part of the dissertation in chapter 10 which supports the theoretical foundation, designing, modelling, and developing the distributed active information model via simulation, emulation and real environments. The third part of this dissertation provides the way to draw conclusions as shown in chapter 7 which has the overall research summary, validation of the propositions, contributions and discussion, limitations and finally recommendations for future works. Finally are the appendices in Appendix A, Appendix B, Appendix C and Appendix D which provide a developing code of the core DAIM model and show different setting up for testbed environments.
Please use this identifier to cite or link to this item: