This project leverages techniques from the real-time systems domain to construct a scalable Network Function Virtualization (NFV) platform that can provide latency and throughput guarantees in a cloud computing setting. Real-time systems have been successfully providing performance guarantees on a wide range of devices, including critical ones such as airbags and pacemakers, where even small delays must be carefully avoided; hence, this technology can provide a solid foundation for an NFV platform with predictable performance.
The intellectual merit of the proposed research is 1) the development of novel, scalable real-time scheduling techniques suited for NFV platforms, 2) the integration of elasticity and run-time adaptation with these scheduling techniques, 3) the application of declarative networking and query planning techniques to analyze and efficiently schedule virtual network functions, and 4) the development of suitable diagnostic primitives.
The broader impact of the proposed research lies in the development of next-generation NFV platforms that will both simplify network management in data centers and support emerging performance-critical applications.
