Agent Based Simulation as an Efficient Way for HPC I/O System Tuning

Generally, evaluating the performance offered by an HPC I/O system with different configurations and the same application allows selecting the best settings. This paper proposes to use agent-based modeling and simulation (ABMS) to evaluate the performance of the I/O software stack to allow researchers to choose the best possible configuration without testing the real system. Testing configurations in a simulated environment minimizes the risk of disrupting real systems. In particular, this paper analyzes the communication layer of an HPC I/O system, more specifically the communication layer of the parallel file system (PVFS2). ABMS has been selected because it enables a rapid change in the level of analysis for modeling and implementing a simulator. It can focus on both macro- and micro-levels (how the aggregate behavior of system agents is born and analyzing their individual behavior).