My project adds a feature to a system utilizing Software Transactional Objects (STOs). STOs began with the idea that being able to handle concurrency control within methods of objects allows for more optimizations when dealing with multicore machines. STOs have been used to implement a high-performance database system. However, when a transaction commits, the commit protocol considers all objects used by the transaction. The goal of this project is to support a new kind of transaction, one in which the first phase filters out irrelevant data, and the second phase carries out the requested computation using the remaining data. We expect that the use of two-phase transactions will significantly improve performance because the work that must be done at commitment time is reduced.

I’m participating in SuperUROP to better understand what the research process consists of and how to communicate findings to those who are researchers and those who aren’t. I like the project because it has some useful applications to databases in industry.