Shengquan Ni: Supporting time-travel debugging in Texera

Abstract: Dataflow systems, traditionally used for relational analysis, now support a variety of tasks including complex user-defined functions. As dataflow jobs become more diverse and complex, there is an increasing need for better debugging support to understand their runtime behaviors and identify issues either in data or the analysis. To achieve the goal in the Texera system, we develop techniques to support “time-travel debugging.” In particular, the system allows users to interact with an execution during runtime to retrieve an execution state, which is a consistent snapshot of the engine. The user has the ability to “travel back to the past” to access the execution state of a previous interaction, thus retrospectively explore and analyze a previous execution state.  We will show a demo of this powerful feature, and give an overview of the underlying techniques.

Bio: Shengquan Ni is a Ph.D. student in the Department of Computer Science advised by Professor Chen Li. His current research interests include big data processing, distributed systems, data analytics and data science. He was a summer intern at Google.