ABSTRACT
This research is aimed at the development of a modified East-West Interface for distributed control plane in Software Defined Network (SDN) for Wide Area Network (WAN). The East-West Interface is important in achieving communication across a Software Defined WAN, to enable scalability and distribution of the control plane. The Communication Interface for Distributed Control Plane (CIDC) was developed to ensure communication in WANs. However, the interface does not address the synchronization of different modules in the controller. This synchronization enables a consistent high availability, policy updating and efficient communication among controllers, needed for SDN to scale in a WAN environment. The modified-CIDC (mCIDC) is therefore developed using the ISyncService that ensures synchronization of state across the different modules in the controller, and among controllers in the WAN. The performance of the mCIDC and CIDC was compared using captured TCP (Transmission Control Protocol) packets, TCP errors and inter-controller communication overload (ICO). The results indicated that for Claranet_2, mCIDC showed a better performance in minimizing the number of Captured TCP Packets, TCP Errors and ICO by 26.55%, 17.89%, and 19.35% respectively when compared with CIDC. While for Claranet_3; 15.82%, 21.60% and 29.25% for Captured TCP Packets, TCP Errors and ICO respectively. For network policies with Claranet_2, mCIDC indicated a better performance in minimize Captured TCP Packets, TCP Errors and ICO by 16.34%, 17.77% and 44.99% respectively. While network policies with Claranet_3; 15.51%, 29.85% and 22.98% for Captured TCP Packets, TCP Errors and ICO respectively. This shows that the mCIDC ensures communication by transmitting the necessary required packets (information) among controllers with reduced TCP errors and fewer overloads.
TABLE OF CONTENTS
DECLARATION ii
CERTIFICATION iii
DEDICATION iv
ACKNOWLEDGMENT v
ABSTRACT vi
TABLE OF CONTENTS vii
LIST OF FIGURES x
LIST OF TABLES xii
LIST OF APPENDICES xiii
LIST OF ABBREVIATIONS xiv
CHAPTER ONE: INTRODUCTION
1.1 Background of Research 1
1.2 Significance of Research 3
1.3 Statement of Problem 3
1.4 Aims and Objectives 3
1.5 Scope of the Study 4
CHAPTER TWO: LITERATURE REVIEW
2.1 Introduction 5
2.2 Review of Fundamental Concepts 5
2.2.1 Traditional networks (Legacy Networks) 5
2.2.2 Software Defined Networking (SDN) 6
2.2.3 SDN Architecture 8
2.2.4 SDN Communication Interface 9
2.2.4.1 Southbound interfaces 9
2.2.4.2 Northbound interfaces 10
2.2.4.3 East/Westbound interfaces 10
2.2.5 Controller 12
2.2.6 Control Architecture 13
2.2.6.1 Single Control Plane 13
2.2.6.2 Multiple Control Plane 13
2.2.7 SDN Controllers 15
2.2.7.1 Floodlight 16
viii
2.2.8 Open Flow 17
2.2.8.1 Open Flow Protocol 18
2.2.9 SDN and WAN 18
2.2.9.1 Multi-domain 19
2.2.10 CIDC 20
2.2.11 Simulation and Emulation Tools 21
2.2.11.1 Mininet Network Emulator 21
2.2.11.2 Graphical Network Simulator-3 (GNS3) 22
2.2.11.3 Iperf 22
2.2.13 Wireshark 23
2.2.14 Performance Metrics 23
2.2.13.1 Captured TCP packets 23
2.2.13.2 TCP errors 23
2.2.13.3 Inter-controller Communication Overload (ICO) 23
2.3 Review of Similar Works 24
CHAPTER THREE: MATERIALS AND METHODS
3.1 Introduction 32
3.2 Materials 32
3.2.1 Hardware 32
3.2.2 VM Workstation Pro v12.0.1.3 32
3.2.3 Floodlight Controller 32
3.2.4 Mininet 33
3.3 Methods 33
3.3.1 Claranet Network Topology 34
3.3.2 GNS3 35
3.3.3 CIDC Operation 36
3.3.4 Modification of CIDC 37
3.3.5 Information Exchange by ISyncService 38
3.3.6 Development of Network Policies 39
3.3.7 Experimental Description 39
3.3.8 Performance Evaluation 41
ix
CHAPTER FOUR: RESULTS AND DISCUSSION
4.1 Introduction 42
4.2 Experimental Connection 42
4.3 Performance Analysis without Network Policies 42
4.3.1 Result Analysis of CIDC without Network Policies 43
4.3.2 Result Analysis for mCIDC without Network Policies 46
4.4 Performance Analysis with Network Policies 49
4.4.1 Result Analysis of CIDC with Network Policies 50
4.4.2 Result Analysis of mCIDC with Network Policies 53
4.5 Performance Percentage Improvement of mCIDC over CIDC 56
4.5.1 Comparison of CIDC and mCIDC without Network Policies 57
4.5.2 Comparison of CIDC and mCIDC with Network Policies 58
CHAPTER FIVE: CONCLUSION AND RECOMMENDATIONS
5.1 Summary 59
5.2 Conclusion 60
5.3 Limitation 60
5.4 Significant Contributions 60
5.5 Recommendations for Further Work 61
REFERENCES 62
APPENDICES 66
CHAPTER ONE
INTRODUCTION
1.1 Background of Research
Computer networks need network devices such as routers and switches to transmit information over the network. These network devices require protocols and policies for effective communication (Nunes et al., 2014). Traditional networks are faced with many challenging concerns (Chen et al., 2015) such as; implementing high-level policies in the network, and manually configuring each network device using low-level configuration commands while considering the network conditions. There is also the need for network environments to be dynamic (Fault-tolerant, Load changes, Automatic Reconfiguration and Response Mechanism), making it challenging to enforce required network policies (Kreutz et al., 2015). In addition to the increasing complexity of Network devices, Traditional networks, also known as legacy networks, are vertically integrated (Kreutz et al., 2015). The control plane and data plane are bundled inside the networking devices, preventing innovation and network flexibility. Another challenge posed by traditional networks is “Internet ossification”; the difficulty in evolving the Internet in terms of physical infrastructures, protocols and performance (Nunes et al., 2014). Emerging Internet applications and services have become increasingly complex and demanding, as such the Internet needs to evolve to address these challenges (Nunes et al., 2014).
Software defined networking (SDN) is a new networking paradigm that addresses the limitations of traditional networks by simplifying network management and enable innovation and evolution (Kreutz et al., 2015). SDN separates the control plane from the data plane. While the control plane decides how to handle the traffic, the data plane forwards traffic according to decisions that the control plane makes (Feamster et al., 2013). SDN has gained immense interest from the industry and academia (Jarraya et al., 2014).
2
It transfers the intelligence from the traditional network devices to a centralized control plane and enables network programming. SDN is now envisioned for multi-datacenter environments e.g. B4 (Jain et al., 2013) and WANs (Wide Area Networks). This can only be achieved by distributing the SDN control plane. A distributed control plane is necessary to tackle single point of failure (SPOF); which makes SDN architecture highly vulnerable to attacks (Kreutz et al., 2013). It also addresses the challenges of scalability and performances in large networks.
The distributed control plane involves the use of multiple controllers, and this is divided into: logically centralized and logically distributed control planes (Blial et al., 2016). The logically centralized control plane balances charges between controllers and uses a shared database to unify decisions. However, it requires extensive synchronization between controllers and it is not suitable for large and highly distributed networks (e.g. Multi-domain networks). The logically distributed control plane (logically distributed controllers) is suitable for large distributed networks, where each controller manages its domain and distributes the necessary data to other controllers. The primary use of this category is in large data centers and WAN networks that suffer from the high cost and latency, due to the complexity of the infrastructure and protocol e.g. border gateway protocol (BGP), and multi-protocol label switching (MPLS) that handle the traffic (Benamrane et al., 2017).
In a logically distributed SDN architecture (e.g. WAN), the communication between multiple controllers is of primary importance (Benamrane et al., 2017). The communication between the controllers at the control plane is handled by the East-West Interface (or API). Currently, there is no standard for the East-West Interface (Jarraya et al., 2014). Existing research on East-West interface does not consider several characteristics of a real WAN such as different network policies, with high availability.
3
1.2 Significance of Research
Networks and network technologies have evolve over the years, providing a need for effective and efficient network management across all kinds of networks (small, medium and large networks). This network management must be reliable, scalable and provide fine-grained performance. This necessitate the development of SDN for WAN; providing forwarding, distribution and specification abstractions. This development will enable simple, scalable, cost-effective, efficient, secure, and enhance connectivity across the network(s). The significance of this research work is to ensure communication in WAN with different policies through policy updating and high availability; encouraging the development of SD-WAN. Previous researchers have not considered consistent high availability across controllers for East-West Communication.
1.3 Statement of Problem
The logically distributed architecture control plane address the issue of scalability in large networks such as WAN by providing an East-West Interface for communication among this networks. The East-West Interface needs to provide scalability through efficient and effective communication among controllers in a WAN network; considering real-time characteristics of WAN. Therefore, there is a need for an Interface that considers high availability for policy updating and decision making, and ensures communication among WANs with different policies. This research is aimed at developing a modified East-West Interface for WANs by incorporating high availability across the controllers.
1.4 Aim and Objectives
The aim of this research is to develop a modified East-West interface for distributed control plane in Software Defined Network (SDN) for Wide Area Networks (WANs).
To achieve this aim, the following objectives were set:
4
1. Emulation of the network environment on a virtual machine (VM) running Ubuntu 16.04 LTS necessary for the implementation of the East-West communication interface for distributed control plane (CIDC) developed by Benamrane et al., 2017.
2. Development of an improved East-West interface based on the network emulated in (1) called the “Modified-CIDC” (mCIDC) to connect the different WANs and network policies.
3. Evaluation of performance of the mCIDC with CIDC using captured TCP packets, TCP errors, and inter-controller communication overload (ICO) as performance metrics.
1.5 Scope of the study
The research study considers the use of the same controller across the different WAN network.
5
IF YOU CAN'T FIND YOUR TOPIC, CLICK HERE TO HIRE A WRITER»
