Mentor: Dr. Tuba Kahveci
College of Engineering
"I look forward to being an innovator and contributing to a project that will help improve efficiency in cloud management software."
- Cloud Computing
- Big Data
- Computer Graphics
Hobbies and Interests
Identifying Source-code Level Metrics for MapReduce Programs to Guide Performance Tuning
An open research problem in cloud computing area is how to configure a MapReduce job such that a certain level of efficiency is achieved. The most popular cloud management software that uses the MapReduce paradigm is called Hadoop. Hadoop provides a set of configuration parameters with some default values. This research project will focus on 1) identifying source-code level metrics for a typical MapReduce program, 2) identifying methods on how to automatically compute those metrics, 3) performing experiments to find out if those metrics are correlated to any of the configuration parameters, and 4) providing suggestions on how to tune the relevant configuration parameters based on specific source-code level metrics.