Maintainer: Sebastian Oeste, ZIH Team @ TU Dresden

There are several metrics to assess the I/O performance of parallel applications, from numerical metrics like bandwidth and IOPS to access pattern characterization.

The assessment of such metrics needs knowledge about the underlying hardware capacities but also about the I/O behavior of the application itself.

There are existing analysis tools to extract I/O performance metrics from application runs.

It is critical to know how to use these tools and how to assess the analysis results reported by the tools. By knowing which metrics affect I/O performance, it is possible to work out optimization strategies for applications I/O behavior and configuration settings of the used file systems.

• Assess metrics of I/O performance with regard to bandwidth, IOPS, latency, access pattern and file layout.
• Enable practioners to use tools for I/O performance analysis and characterization.
• Identify optimization strategies to improve I/O perfromance.

• Understand metrics that describe relevant I/O performance characteristics.
• Select tools to analyze I/O performance of applications.
• Apply the performance analysis tools to applications and workflows.
• Interpret the results of I/O analysis tools and distill optimization strategies for the application or system configurations.