Workload characterization using lightweight system call tracing and reexecution

This paper shows how system call traces can be obtained with minimal interference to the system being characterized, and used as realistic, repeatable workloads for experiments to evaluate operating system and file system designs and configuration alternatives. Our system call trace mechanism, called ULTra, captures a complete trace of each UNIX process's calls to the operating system. The performance impact is normally small, and it runs in user mode without special privileges. We show how the resulting traces can be used to drive full, repeatable reexecution of the captured behaviour, and present a case study which shows the usefulness and accuracy of the tool for predicting the impact of file system caching on a WWW server's performance.