I expect to complete my Ph.D. in Spring 2008. I am looking for Fall 2008
academic/research positions.
• Curriculum Vitae (pdf)
• Research Statement (pdf)
• Teaching Statement (pdf)
• References (pdf)
• Papers (html)
My dissertation work, supervised by Professor Barton Miller, focuses on reliability and scalability for large scale parallel and distributed systems. I am studying ways to use the structure and properties of tree-based overlay networks (TBONs) to develop reliability mechanisms that do not require any computational, storage, or network overhead in the absence of failures.
Dorian C. Arnold and Barton P. Miller. "A Scalable Recovery Model for Tree-based Overlay Networks," UW CS Technical Report, CS-TR-2007-. December, 2007.
Dorian C. Arnold, Dong Ahn, Bronis de Supinski, Gregory Lee, Bart Miller, Martin Schulz. "Stack Trace Analysis for Large Scale Debugging," in Proceedings of the 21st International Parallel and Distributed Processing Symposium (IPDPS '07), Long Beach, California, March 2007.
Philip C. Roth, Dorian C. Arnold and Barton P. Miller. "MRNet: A Software-based Multicast/Reduction Network for Scalable Tools," in Proceedings of SC 2003, Phoenix, AZ, November, 2003.
MRNet The Multicast/Reduction Network (MRNet) is our prototype tree-based overlay network, which uses a hierarchical process organization to provide tools and applications with scalable multicast and gather communication and fully customizable data aggregation capabilities. I developed MRNet with Drs. Philip Roth and Barton Miller.
STAT I developed the stack trace analysis tool (STAT) to sample application stack traces and scalably organize the samples into a call graph prefix tree that distinguishes process equivalence classes, groups of processes with similar behavior. LLNL/UW collaborators: Dong Ahn, Bronis de Supinski, Greg Lee, Bart Miller, and Martin Schulz
Paradyn The Paradyn project researches and develops software technologies for binary instrumentation, binary analysis, and scalability. MRNet is a part of the Paradyn project.
NetSolve At the University of Tennessee, I worked with Professor Jack Dongarra on the Netsolve project researching distributed system performance primarily through dataflow optimizations. NetSolve (originally developed by Dr. Henri Casanova) is a fault-tolerant RPC system for solving complex scientific problems remotely using efficient hardware and software solutions. In 2000, NetSolve won an R & D Top 100 award.
CLUBS At the University of Tennessee, I worked with Professor James Plank to develop CLUBS (Checkpoint Library for Unix Based Systems), a user-transparent checkpoint library based on its predecessor, libckpt. I studied checkpointing optimizations including file formats for low-overhead roll-backs and copy-on-write process forking to decrease checkpointing overhead.