Instability in parallel I/O systems

The performance of a shared-disk parallel I/O system with several concurrent processes is studied. It is observed that under some naturally occurring conditions such a system can exhibit instability, with a subset of the processes racing ahead of the others and monopolizing I/O resources. This race can lead to serialization of the processes and poor disk utilization. The study has implications for both data placement in multiple disk systems and task partitioning for parallel processing.