Tsinghua University has won the 2023 ACM SIGMOD Research Highlight Award. This award is presented by the Special Interest Group on Management of Data (SIGMOD) of the Association for Computing Machinery (ACM) to recognize projects that exemplify core database research.
The awarded paper, titled “Building Write-Optimized Tree Indexes on Disaggregated Memory”, was authored by Qing Wang, a 2018 doctoral student from the Department of Computer Science and Technology, and his advisors Professor Jiwu Shu and Associate Professor Youyou Lu from the same department. The paper was originally published at the SIGMOD 2022 conference and focuses on data index design under the Disaggregated Memory architecture. The paper proposes a high-performance B+ tree index named Sherman, incorporating three aspects of design: concurrency control, network communication, and data structure layout. It uses the on-chip memory of network card to design an efficient distributed lock primitive, reduces the number of network round trips for index operations, and introduces a two-level versioning mechanism to reduce write amplification. The paper shows that under typical write-intensive workloads, Sherman improves throughput by an order of magnitude and reduces tail latency by an order of magnitude.
The paper is a result of the research conducted by the Storage Research Group of the Institute of High-Performance Computing (IHPC), one of the earliest research teams engaged in building storage systems in China. The laboratory has made a series of achievements in various aspects of storage systems, such as network storage systems, cloud storage systems, flash storage systems, persistent memory storage systems, and edge intelligent storage systems. The laboratory has also collaborated with leading IT companies to apply related technologies and achievements into practice. Based on the research results, several start-up companies have been successfully incubated. The research group has published over 100 academic papers in top conferences and authoritative journals including FAST, SOSP, OSDI, ASPLOS, SIGMOD, USENIX ATC, EuroSys, MICRO, ISCA, SC, ACM/IEEE Trans, among others, and has won many national and provincial awards.
The ACM SIGMOD Research Highlight Award was established in 2016 to honor projects that address an important problem, represent a definitive milestone in solving the problem, and have the potential of significant impact. The award covers all conference and journal papers published by the data management community in 2022. A total of 12 papers won the award this year.
