Collaborative Research: NeTS: Small: Resource-efficient Consolidated Data Plane for Network Functions

Computer networks, from wireless to the Internet, are crucial enablers of the United States economy and industry, and of many applications important to society, such as education, healthcare, and workforce development. As such, scientific research on this topic explores new innovations that have the potential to improve the performance and cost of such networks.

This project is focused on network functions (NFs), which are crucial for managing networks and enhancing user experience from different perspectives such as providing low-latency applications and stronger security guarantees. Supporting multiple NFs simultaneously on a single network device or multiple heterogeneous devices, while minimizing resource usage and maximizing throughput, has gained significant attention.

However, manually managing these co-existing NFs is labor-intensive and unsustainable. This project aims to design efficient tools that automatically manage the algorithm design, deployment and resource management of multiple co-existing NFs through a new concept called consolidated data planes. These management tools will allow network operators to efficiently handle co-existing NFs, optimizing resource usage and throughput, and thus reducing economic costs.

This project proposes consolidated data planes based on the finding that there is a huge space for optimization if efficient and space-compact data structures and algorithms considering multiple NFs can be designed. It achieves its goals through the following: 1) An automatic tool to construct consolidated data plane programs for multiple NFs; 2) Deployment of consolidated data plane programs to meet the NF objectives and the resource limitations of the network devices; 3) Network-wide management of consolidated data plane resources.

The proposed design will be fully implemented and evaluated on a Tofino switch platform at the University of Connecticut and the University of California Santa Cruz, as well as a heterogeneous devices platform on CloudLab. This project will introduce a new research direction about programmable data plane algorithms that potentially impact network algorithm research.

This project will also result in the development of open-source tools to manage co-existing NFs effectively.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.