My current research focuses on intelligent grouping for purposes of caching, disk management, and predication. Specifically, I'm currently working on grouping schemes (both on-line and off-line) for disk management. Additional topics that I am interested in (but am not currently pursuing) include artificial intelligence, specifically machine learning algorithms, intelligent adversarial game play and strategies, and high-dimensional learning problems and applications, as well as algorithmic theory.
Back to top
Storage Research Group (SRG)
Advisor: Ahmed Amer
Webpage: http://www.cs.pitt.edu/~amer/
Intelligent data grouping has many applications, including caching, access prediction, and data layout policies. I work on developing predictive grouping strategies and the investigation of their performance limits under Dr. Ahmed Amer.
Our work in these areas has included formulation of a generalized data grouping problem, and the development and testing of optimal off-line algorithms for more restricted forms of the problem. We are currently investigating the on-line adaptation and implementation of these techniques and algorithms. We are particularly interested in their application to storage optimization and are evaluating the performance impact of integrating such policies into the storage subsystem.
Back to top
Interesting Problems
Favorite storage problems encountered (read “headaches”) to date include:
- The developement of a provably suboptimal algorithm for solving the Predictive Data Grouping Problem.
- Large memory processes, and the surprising challenges of constructing very large hashes.
Back to top
Recent Publications
- David Essary, Ahmed Amer, "Avoiding State-space Explosion of Predictive Metadata with SESH," International Performance Computing and Communications Conference (IPCCC) 2009. Phoenix, Arizona, December 2009.
- David Essary, Ahmed Amer, "Space-Efficient Predictive Block Management," International Workshop on Software Support for Portable Storage (IWSSPS) 2009. Grenoble, France, October 2009.
- David Essary, Ahmed Amer, "Predictive Data Grouping: Defining the bounds of energy and latency reduction through predictive data grouping and replication," ACM Transactions on Storage. May 2008.
Recent Tech Reports
- David Essary, Ahmed Amer, "Space-Efficient Predictive Block Management," TR-09-165, University of Pittsburgh, April 2009.
- David Essary, Ahmed Amer, "DrNO Clustering: Mozart Trace Analysis", TR-112, University of Pittsburgh, August 2004.
- David Essary, Ahmed Amer, "Project DrNO: Proof of Optimality", TR-113, University of Pittsburgh, March 2004.
Recent Presentations
- "Testing the Limits of Power and Latency Reductions by Predictive Data Grouping" University of Pittsburgh's School of Arts & Sciences Grad Expo 2007. Pittsburgh, PA. March, 2007.
- "Predictive Layout Strategies and Disk Management" University of Pittsburgh's Department of Computer Science Graduate Research Competition 2007. Pittsburgh, PA. February, 2007.
- "Limits of Power and Latency Reductions by Intelligent Grouping" 7th USENIX Symposium on Operating Systems Design and Implementation (OSDI '06). Seattle, Washington. November, 2006.
Degrees
- University of Pittsburgh, April 2003
Bachelor of Science, Computer Science Department - University of Pittsburgh, April 2003
Bachelor of Science, Department of Mathematics
Recent Awards and Accomplishments
- University of Pittsburgh's Grad Expo 2007 poster finalist
- University of Pittsburgh's Department of Computer Science 2007 poster finalist
- University of Pittsburgh's Department of Computer Science Grad Research Competition 2007 finalist
- Member of Upsilon Pi Epsilon Honor Society in the Computing and Information Disciplines (April, 2005)
- Graduated Magna Cum Laude with dual BS degrees (April, 2003)
Service
- Served as paper reviewer for NAS 2009.
- Served as paper reviewer for IPCCC 2008.
- Served as paper reviewer and referee for ICAC 2006.
Doctoral Studies
- Proposal reading list coming soon.
- Proposal document coming soon.
- Proposal presentation coming soon.