International Journal of Engineering and Management Research

Acknowledgement

I want to express my sincere gratitude to my academic supervisor, Madhuvan Dixit, for his constant guidance and support throughout the completion of this dissertation.

1. Introduction

With the rapid growth of cloud computing, remote work, and bandwidth-intensive applications, traditional Wide Area Networks (WANs) are no longer sufficient to meet modern enterprise requirements. Legacy WAN architectures primarily depend on Multiprotocol Label Switching (MPLS), which, although reliable, is expensive and lacks flexibility in handling dynamic traffic patterns.

Software-Defined Wide Area Networking (SD-WAN) addresses these limitations by introducing centralized control, application-aware routing, and real-time performance monitoring. It enables organizations to utilize multiple internet links efficiently, improving both performance and cost optimization.

In this research, an SD-WAN environment is implemented using a FortiGate firewall with dual WAN links. The system is designed to monitor network conditions and dynamically route traffic based on predefined Service Level Agreements (SLAs).

Despite these advancements, traditional WAN environments still face several critical challenges, including inefficient utilization of backup links, lack of real-time traffic optimization, and delayed failover mechanisms. These limitations result in increased latency, packet loss, and reduced application performance, particularly for real-time services. Therefore, there is a need for an intelligent network solution capable of dynamically adapting to changing network conditions while ensuring optimal performance and reliability.

2. Literature Review

The increasing demand for scalable, flexible, and cost-effective networking solutions has driven the adoption of Software-Defined Wide Area Networking (SD-WAN) over traditional WAN architectures. Conventional WAN technologies, particularly MPLS-based networks, are often constrained by high operational costs and limited adaptability to dynamic traffic patterns.

Emmanuel presented an SD-WAN architecture that integrates real-time traffic monitoring and dynamic load balancing, identifying key performance metrics such as latency, packet loss, and bandwidth utilization as critical evaluation parameters [6]. These findings highlight the need for a holistic approach that combines algorithmic efficiency with architectural improvements.

Practical implementations further validate the effectiveness of SD-WAN in real-world scenarios. Rahim and Akbi demonstrated that SLA-based bandwidth optimization using FortiGate NGFW enables dynamic traffic distribution across multiple links, resulting in improved bandwidth utilization, reduced jitter, and enhanced network stability [7]. Their study also emphasizes the importance of integrating security mechanisms such as firewall policies within SD-WAN deployments to ensure both performance and data protection. Such implementations reinforce the practical viability of SLA-driven traffic steering mechanisms.

Traffic engineering (TE) has also been widely explored as a means of optimizing network performance. Zheng et al. proposed an online optimization framework that jointly considers traffic engineering and network update costs to improve long-term efficiency [8]. Unlike traditional TE approaches that focus solely on short-term optimization, this method reduces routing instability and operational overhead. In addition, network emulation techniques have been used to evaluate performance under controlled conditions. Elbouanani et al. demonstrated that distributed network emulation can effectively identify delay-related issues and infrastructure limitations in virtualized environments [9]. These methodologies provide valuable insights for validating SD-WAN solutions before deployment.

Overall, the literature indicates that SD-WAN represents a significant advancement in network design by integrating centralized control, intelligent traffic management, and real-time performance monitoring. While existing studies have addressed individual aspects such as load balancing, SLA optimization, congestion control, and traffic engineering [2]–[5], [7]–[9], there remains a gap in developing integrated frameworks that combine these features into a unified solution.

The default gateway for outbound traffic is configured dynamically through the SD-WAN zone, allowing intelligent path selection between the two WAN links. This addressing scheme ensures proper segregation of traffic and enables efficient routing, monitoring, and failover operations within the network.

Step 2: Interface Configuration

Figure 2 shows the configuration of interfaces including LAN and WAN connections on the FortiGate device.

In this step, all physical interfaces of the firewall are configured with appropriate IP addresses and subnet masks. The WAN interfaces are mapped to external networks, while the LAN interface is configured to serve internal users. Administrative access such as HTTPS, SSH, and ping is enabled where required to facilitate monitoring and management. Proper interface configuration ensures stable communication between network components and forms the foundation for SD-WAN implementation.

Step 3: SD-WAN Zone Creation

The configured WAN interfaces are grouped into an SD-WAN zone, creating a logical abstraction that allows centralized traffic management.

Step 5: Performance SLA

Service Level Agreements (SLAs) are configured to monitor critical performance metrics, including latency, jitter, and packet loss. Specific threshold values are defined to determine acceptable performance levels. The firewall continuously probes external servers to measure these metrics in real time. If a link fails to meet SLA requirements, traffic is automatically redirected to a better-performing path. This step is essential for maintaining Quality of Service (QoS) for different applications.

Step 6: Firewall Policy Configuration

Firewall policies are implemented to allow traffic flow between the LAN and WAN networks. Policies are configured with appropriate source and destination parameters, along with services and schedules.

Network Address Translation (NAT) is enabled to ensure that internal users can access external networks. Security profiles may also be applied to enhance protection against threats. This step ensures secure and controlled communication across the network.

4. Results & Analysis

The performance evaluation of the implemented SD-WAN architecture demonstrates significant improvements in network efficiency, reliability, and adaptability when compared to traditional WAN configurations. The system continuously monitored network conditions using Service Level Agreement (SLA) parameters such as latency, jitter, and packet loss, enabling real-time decision-making for traffic routing.

During normal operating conditions, both WAN links actively participated in traffic forwarding, ensuring balanced utilization of available bandwidth. The distribution of traffic across multiple links resulted in reduced congestion and improved throughput. Unlike traditional WAN environments, where backup links often remain idle, the SD-WAN configuration ensured that all available resources were effectively utilized. This behavior highlights the efficiency of dynamic load balancing in optimizing network performance.

As network conditions varied, the system demonstrated intelligent traffic steering by dynamically selecting the most optimal path based on SLA metrics. When one of the WAN links experienced increased latency or fluctuations in jitter, the system automatically redirected traffic to the alternate link without requiring manual intervention. This adaptive behavior ensured that application performance remained consistent, particularly for latency-sensitive services.

The failover capability of the system was evaluated by intentionally disabling one of the WAN links during operation. The SD-WAN architecture successfully detected the link failure in real time and seamlessly shifted traffic to the remaining active link. This transition occurred with minimal disruption, maintaining continuous connectivity and preventing significant packet loss. The rapid response of the system highlights its ability to handle network failures effectively and maintain service availability.

Routing behavior was also analyzed to verify the correctness of the configuration. The presence of multiple default routes through the SD-WAN zone confirmed that the system was capable of supporting redundancy and dynamic path selection.

Another significant aspect of the implementation is the system’s ability to handle network failures. The seamless failover observed during testing highlights the robustness of the SD-WAN architecture. Unlike traditional failover mechanisms that rely on routing protocol convergence, SD-WAN enables near-instantaneous switching based on SLA violations. This ensures minimal disruption to ongoing communication and enhances overall network reliability.

From a practical perspective, the study also highlights the importance of centralized control and simplified network management. By abstracting multiple WAN links into a single SD-WAN zone, the complexity of managing individual interfaces is significantly reduced. This not only simplifies configuration but also improves scalability, making the solution suitable for enterprise environments.

While the results are promising, it is important to acknowledge certain limitations. The implementation was carried out in a simulated environment, which may not fully replicate real-world network conditions such as ISP variability and large-scale traffic patterns. Additionally, the study focuses primarily on performance aspects and does not extensively evaluate security features, which are also critical in SD-WAN deployments.

Overall, the discussion establishes a clear connection between theoretical concepts and practical implementation. It demonstrates that SD-WAN is not only a conceptual advancement but also a viable solution capable of improving network performance, reliability, and operational efficiency in modern enterprise environments.

6. Conclusion

This research study presented the design, implementation, and evaluation of an SD-WAN architecture using a FortiGate firewall in a simulated environment. The primary objective was to analyze the effectiveness of intelligent traffic steering and load balancing based on SLA parameters such as latency, jitter, and packet loss.

The results demonstrate that SD-WAN significantly enhances network performance by enabling dynamic path selection and efficient utilization of multiple WAN links. Unlike traditional WAN architectures, which rely on static routing and underutilized backup links,

[4] Houssam ElBouanani, Chadi Barakat, Walid Dabbous, & Thierry Turletti. (2023). Troubleshooting distributed network emulation. ICIN 2023 - 26th Conference on Innovation in Clouds, Internet and Networks. DOI: 10.1109/ICIN56760.2023.10073480.

[5] Divya Natolana Ganapathy & Dr. Karuna Pande Joshi. (2022). Semantically rich framework to automate cloud service level agreements. IEEE Transactions on Services Computing, pp. 1-12. DOI: 10.1109/TSC.2022.3140585.

[6] Maha Zedan, Gamal Attiya & Nawal El-Fishawy. (2021). Load balancing based active monitoring load balancer in cloud computing. 2^nd IEEE International Conference on Electronic Engineering, Menoufia University, Egypt.

[7] Jiaqi Zheng, Yimeng Xu., Li Wang, Haipeng Dai., & Guihai Chen. (2021). Online joint optimization on traffic engineering and network update in software-defined WANs. IEEE INFOCOM 2021 - IEEE Conference on Computer Communications.

[8] Kasun Jayasinghe. (2020). AI-based SLA monitoring and optimization. International Journal of Trend in Research and Development, 7(6).

[9] Dr.S. Rajagopalan. (2020). An overview of SD-WAN load balancing for WAN connections. Fourth International Conference on Electronics, Communication and Aerospace Technology (ICECA-2020).

[10] Ning Cai., Yanni Han., Yongming Ben., Wei An., & Zhen Xu. (2019). An Effective Load Balanced Controller Placement Approach in Software-Defined WANs. MILCOM 2019 - 2019 IEEE Military Communications Conference (MILCOM).

[11] Van Rossem, W. Tavernier, D. Colle, M. Pickavet, & P. Demeester. (2018). Introducing development features for virtualized network services. IEEE Commun. Mag., 56(8), pp. 184–192.

[12] Pablo L., Jack F. Bravo-Torres Víctor M., Larios-Rosillo Paúl E., Vintimilla-Tapia Iván F., Yuquilima-Albarado Santiago J., & Arévalo-Cordero. (2017). Living lab concept for cloud analysis in networks of metropolitan sensors applying the concept of SD-WAN and hybrid networks. CHILEAN Conference on Electrical, Electronics Engineering, Information and Communication Technologies.