Yunhao Zhang

Cornell University


I am a fourth year PhD candidate in Computer Science at Cornell University, advised by Lorenzo Alvisi. I am interested in the theory and practice of fault-tolerant distributed systems. One fundamental building block in such systems is called consensus which requires machines to agree on the same sequence of commands for execution. For example, Facebook’s consensus-based system Delos provides fault-tolerant storage to Facebook applications. My current research focuses on consensus-based systems deployed across multiple parties, such as those supporting permissioned blockchains. Consider a blockchain for exchanges: Different *order* of commands usually lead to different financial benefits to the parties involved, resulting in vulnerabilities to order manipulation which has been observed in real-world blockchains. Such relation between order and financial benefits didn’t exist in classic consensus-based systems like Delos. Therefore, the problem is that classic theory and practice didn’t consider requirements on how a consensus-based system should order the commands. Our first paper addressing this problem won the Jay Lepreau Best Paper Award at OSDI’20. We introduced new abstractions and natural theories for specifying and reasoning about how consensus-based systems should order the commands. Before starting my PhD and working on this problem, I was a researcher at Shanghai Jiao Tong University, advised by Rong Chen and Haibo Chen. I worked on distributed data processing systems leveraging the latest hardware features available in data centers