Info Hub Distinguished Lecture Series 信息枢纽卓越系列讲座 - Enhancing Relational Database Management Systems with Increased Power and Efficiency

Relational Database Management Systems (RDBMS) have been extensively studied over decades to manage large amounts of data, and there has been a long history of attempts to support complex structures in RDBMS since early 1970. In this talk, we will review various approaches to manage complex structures in RDBMS.  As the need for large-scale graph processing and the development of graph algorithms and systems has grown, we revisit the issue of supporting graph processing in RDBMS. This is motivated by the fact that many real-world applications involve relations closely related to large graphs stored in RDBMS, requiring integrated graph and data analytics.  To support graph analytics in RDBMS, we will discuss new relational algebra operations that can be defined using basic relational algebra operations combined with group-by and aggregation. We will also discuss new SQL recursive queries that are guaranteed to have a fixpoint with these new operations. Furthermore, we will explore how to support these new operations and SQL recursive queries using SparkSQL and GraphX on the Spark platform, as well as how to optimize SQL queries by separating communication from computation in a distributed system to achieve high efficiency.