With constant hardware advancements, it is crucial to design software-based solutions which exploit new hardware trends in order to build efficient and robust networked systems. In the era of data center networks, software-defined networks and software-defined data centers, it is worth reconsidering how to provide old network functions like security, QoS, load balancing, web caching, besides new network functions. My research work focuses on creating and designing practical algorithmic solutions to build functional systems. I improve the performance of various networked systems, making these systems more efficient in multiple ways – speed, power consumption, storage requirements, cost of building and maintaining such systems, and various other dimensions. My work has had broad impact on different networked systems. 1) My work on software router exploits the parallelism in state-of-art multi-core desktop platforms, and achieves a packet classification speed of more than 10Gbps on a single 8-core Intel Xeon processor desktop platform, which is much less expensive than a high-end router. 2) In the light of new hardware features that TCAMs now support multiple parallel searches and partial TCAM can be addressed, my work on power-efficient TCAM design has reduced the power consumption of TCAMs by more than 80% when used for multi-dimensional packet classification. 3) To accommodate the increasing demand of moving enterprise file systems to cloud, my work on cloud file system has greatly reduced both storage cost and bandwidth cost of maintaining real enterprise file systems on cloud. For future work, I pan to apply algorithmic techniques to software-defined networks (SDN), data center networks and enterprise networks, in order to build efficient systems by leveraging the advantages of both software and hardware.
Yadi Ma is a PhD candidate in Computer Science at the University of Wisconsin-Madison. Her research interests are in networking and distributed systems with emphasis on topics like data center networks, software-defined networking, and design of new algorithmic techniques for such systems. While her work have been published in major venues, some also turned into patents and have been licensed to industry, with great potential of being deployed in real systems.