الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 141 |  |  |
| | | from | 141 | | |
Abstract
The escalating complexity and dynamism of modern data center networks have precipitated the evolution of network paradigms to meet their growing demands. Among these, Software Defined Networking (SDN) is the new network model that uses the notion of centralized administration and control to promote network management and streamlines infrastructure maintenance. SDN has many advantages to promote network performance, offers real-time responsiveness and meets demanding high availability criteria. However, this novel paradigm encounters various technological challenges due to its architecture, some intrinsic and others inherited from pre-existing technologies. This study commences with a systematic review of existing literature, illuminating routing traffic concerns within the realm of SDN and provides insights into the forthcoming challenges that confront this transformative network model, encompassing both protocol and architecture perspectives. Additionally, we aim to explore diverse extant solutions and mitigation strategies that tackle issues of SDN scalability, elasticity, dependability, reliability, high availability, resiliency, and performance. Through a combination of analytical modeling, simulation studies, and practical implementations, this thesis rigorously evaluates the proposed routing algorithms’ performance across various dimensions. Metrics such as network throughput, latency, packet loss, and resource utilization are systematically measured to assess the efficacy of the devised strategies. The findings provide valuable insights into the strengths and limitations of different routing approaches, aiding network architects and administrators in making informed decisions to optimize data center network performance. A centralized controller must not constitute a single point of failure for the network to achieve high availability. This reveals the need for a Real-time fault tolerance mechanism. This mechanism aims to address potential failures by providing redundancy and failover capabilities within the network infrastructure. By distributing control and decision-making responsibilities across the cluster control, the system can continue to operate seamlessly by selecting the better controller by routing the traffic from the cluster to it even if one controller experiences a failure. This real-time fault tolerance mechanism plays a vital role in maintaining uninterrupted network operations and achieving the desired level of availability. This thesis presents a Real-time Efficiently Adaptive Traffic Route (RATR) algorithm That ensures fault tolerance and load balancing which achieves high availability consistency based on real-time measurements of cluster members’ performance monitoring and records the results as votes. Then calculate the collected votes for each performance metrics load which are CPU, Memory, Network traffic and Response time. Finally, run the proposed grading mechanism to elect the leader controller and his vices from among all cluster members. Extensive experiments are conducted to prove the effectiveness of RATR. The results demonstrate that RATR achieves superior performance across key network metrics, including throughput, delay, and packet loss, when compared to existing algorithms such as Round Robin, SMCLBRT, and ESCALB. This improvement can be attributed, to the efficient leader and vice controllers election process employed by RATR.