Analyzing Parallel I/O
Authors: Dr. Philip Carns (Argonne National Laboratory)
BP
Abstract: Parallel application I/O performance often fails to meet user expectations. In addition, subtle changes in access patterns may lead to significant changes in performance due to complex interactions between hardware and software. These challenges call for sophisticated tools to capture, analyze, understand, and tune application I/O.
In this BoF, we will highlight recent advances in monitoring and characterization tools to help address this problem. We will also encourage community discussion to compare best practices, identify gaps in measurement and analysis, and find ways to translate parallel I/O analysis into actionable outcomes for users, facility operators, and researchers.
Long Description: I/O behavior is increasingly complex and notoriously difficult to understand and tune. A variety of tools and techniques have been developed to help address this problem by instrumenting I/O behavior at the application, service, or hardware level. Some of them are designed to assist end users while others are oriented towards administrators. It is important to select the right tool (or combination of tools with correlated data) according to your use case.
The objectives of this BoF are to 1) highlight recent advances in tools and techniques for monitoring I/O activity in data centers, 2) to discuss experiences and limitations of current approaches, 3) to discuss and derive a roadmap for future I/O tools with the goal to capture, assess, predict and optimize I/O.
The technical presentations target end-users, I/O software developers, and facility administrators. We will collect surveys to assess the state of the field and post the results online along with all presentation materials.
This will be the 4th occurrence of this BoF (previously held at SC14, SC15, and SC16). The most recent event page can be found at https://wr.informatik.uni-hamburg.de/events/2016/bof-monitoring, and the final report can be found at https://wr.informatik.uni-hamburg.de/_media/events/2016/sc16-analyzing-parallel-io-bof-report.pdf. We estimate that roughly 100 people attended the SC16 event. The SC15 event was limited by room size and attendees were turned away.
Conference Presentation: pdf
Birds of a Feather Index