Congratulations to the KNIT10 Award Winners Kriti Goel, Saurya Saha, Vaneshi Ramdhony, and Hyunsuk Bang! Awardees were selected for their innovative uses of FABRIC testbed resources, experiment design, and execution. Learn more about their work below.
Best Experiment
Expedited Internet Bypass Protocol
Kriti Goel
This project addresses the growing complexity in Internet routing by simplifying the Internet Protocol and its routing protocols. We start with simplifying routing in an intra-AS communications System and comparing with OSPF the defacto intra-AS routing protocol. Currently routing protocols such as OSPF designed in 1989 for technologies and networks of those years is the preferred routing protocol for intra-AS communications. OSPF has scalability issues, which is addressed by the area concept. It has convergence issue, which is addressed by Link-Layer Failure detection or BFD. It requires network or area-wide update dissemination. Blast Radius is high – resulting in unstable systems.
The researchers’ solution protocol addresses all the above challenges. It is scalable, has inherent fast failure detection and is more stable. The project demonstrates the superior performance of novel approaches to Internet routing and also simplify routing techniques. The team adopted a novel technique that uses physical/virtual structure in the AS to set up routes without using routing protocols and IP addresses. The solution can scale without adding operational complexity.
The team has a scalable routing technique that can replace OSPF and BGP. This would be a significant reduction on router operations, instability and cost, while improving performance. They have designed and tested a novel routing technique to address several challenges faced in the Internet. The broader impacts: This would be starting point for network researchers to investigate simple, green solutions to address the escalating energy concerns and complexity in network operations. The simple approaches would provide significant performance and security benefits.
Best Classroom Use
Optimizing CDN Architectures: Multi-Metric Algorithmic Breakthroughs for Edge and Distributed Performance
Saurya Saha
The paper titled “Optimizing CDN Architectures: Multi-Metric Algorithmic Breakthroughs for Edge and Distributed Performance” focuses on enhancing Content Delivery Networks (CDNs) by deploying dynamic algorithms and multi-metric analyses. These improvements aim to address challenges in scalability, load balancing, and latency optimization across both edge and distributed computing environments. We use two experimental setups— a local system simulating an edge environment and a scalable FABRIC testbed—to evaluate the effectiveness of their algorithms. We measure performance metrics like round-trip time (RTT) and CPU usage, emphasizing the trade-offs between scalability and resource consumption. The research highlights how multi-metric analyses provide a more comprehensive understanding of CDN dynamics and can lead to more effective deployment strategies in various network environments.
Best Paper
Running GraphBLAS on FABRIC
Vaneshi Ramdhony & Hyunsuk Bang
This paper describes the first FABRIC deployment of GraphBLAS—a powerful linear algebra-based framework for network traffic analysis. FABRIC is an international network testbed that is used for teaching and research in networking. This work has two goals to provide a foundation for further research on network security: (1) deploying an advanced network monitoring technique on FABRIC, and (2) making this deployment shareable with other FABRIC users. These goals are achieved by creating FABRIC experiments that build and deploy GraphBLAS to process network traffic on FABRIC. These experiments are described as executable Jupyter notebooks that are shareable with other researchers. These experiments also include the replaying of network workloads (from pcap files), which we use to test and evaluate the deployment.