Title: Understanding and Improving Database-backed Applications
Advisor: Alvin Cheung
Supervisory Committee: Alvin Cheung (Chair), Nicholas Weber (GSR, iSchool), Magdalena Balazinska, and Dan Suciu
Abstract:
From online shopping to social media websites, many applications are built using imperative language like Java, Python or Ruby while store its data in a relational database. Such architecture enables fast development of application logic using object-oriented data model while leveraging the convenient persistent-data management of relational databases. However, the separation between the application and the database often leads to performance issues and buggy behaviors.
Our work aims to understand the performance and functionality issues of these applications, and build tools to automatically or interactively improve the performance, make performance-functionality tradeoff and fix application bugs. In this talk, I will introduce our prior work focusing on the performance: (1) Quro, a compiler to reorder transaction queries, (2) Chestnut, an in-memory data layout designer, and (3) Panorama, an interactive tool for view-driven optimizations. I will also propose our work on understanding and fixing data consistency problems, and leveraging data consistency for query optimization.