For more than three decades, the slowest and most limiting component in computer systems has been the disk-based file system. For that reason, file systems have consistently been a critical area of computer systems research. One of the most useful techniques for studying file systems is to collect, analyze, and replay detailed activity traces. In the past, traces have been used for retrospective analysis and as a source of workloads for both simulation and live experiments. However, many file-system studies have been handicapped by a lack of publicly available trace data, often causing experiments to be performed using inadequate data sets. In many cases, those data sets are not published for use by other researchers, making comparative studies difficult or impossible.
In this work, the PIs are operating and expanding IOTTA Trace Repository (after the Storage Networking Industry Association's I/O Tools, Traces, and Analysis Working Group) that makes standardized, high-quality data sets available to researchers worldwide. This repository is widely known in the research community as the most reliable source of file system and I/O trace data, and has proven its worth by providing data sets to hundreds of research projects worldwide each year. The project involves defining standard formats and then creating tools to make it easy to work with traces, despite their immense size. Because the trace repository is inherently a community effort, it is supervised by an advisory panel of experienced researchers.
