Abstract: Modern databases are moving to the cloud for elastic resource allocation and high availability. Cloud-native databases adopt a unique storage-disaggregation architecture, where the computation and storage are decoupled as two separate services connected through the data center network. The new storage-disaggregation architecture demands a revisit of database system design. In this talk, I first discuss our recent research on optimizing the two-phase commit protocol, a fundamental building block in distributed transactional databases. We leverage the disaggregated storage service to reduce the protocol latency and eliminate blocking. Then, I discuss our work on leveraging storage-layer computation (i.e., a pushdown layer) to accelerate data analytics processing. 

Bio: Xiangyao Yu is an Assistant Professor at the University of Wisconsin-Madison. His research interests include (1) cloud-native databases, (2) new hardware for databases, and (3) transactions and HTAP. Before joining UW-Madison, he finished postdoc and PhD at MIT and bachelor at Tsinghua University.