The Virtual Institute of I/O and the IO-500
Authors: Dr. Julian Kunkel (German Climate Computing Center)
Abstract: Due to the increasing complexity of HPC data management, activities in the storage research community have increased over the last few years. The general purpose of this BoF is to foster this community.
An important activity is the community driven development of an IO-500 benchmark. The speakers will briefly introduce the international Virtual Institute for I/O (VI4IO, http://vi4io.org) and focus on the status of the IO-500 development and the usage of the benchmark. A highlight is the presentation of the first IO-500 list. The direction of VI4IO and the standardization of the IO-500 is then discussed with the participants.
Long Description: Goals of the BoF are to 1) advertise the community hub but also discuss and steer the direction of the community effort; 2) to reveal the inaugural IO-500 list;
3) to discuss the benefit and direction of the efforts with the community.
Goals of the Virtual Institute for I/O are:
* Provide a platform for I/O researchers and enthusiasts for exchanging information
* Foster international collaboration in the field of high-performance I/O
* Track, and encourage, the deployment of large storage systems by hosting information about high-performance storage systems
VI4IO provides a resource for storage researchers to both document their systems and discuss best practices.
The IO-500 list offers a way to document relative performance for comparative purposes with a ranking to encourage competitive development and IO optimization for vendors and provides a management justification.
The IO-500 benchmark consists of data and metadata benchmarks to identify performance boundaries for optimized and suboptimal applications.
Together with comprehensive data from sites, supercomputers and storage, in-depth analysis of system characteristics are tracked by the list and can be analyzed in detail.
Tuning of storage is explicitly encouraged by the design of the IO-500 but tuning details are part of the submission process and must be revealed.
Thus, they help defining best practices.
We have run the benchmark on several sites and reveal the first list during supercomputing.
Expected HPC audience are 1) I/O experts from data centers and industry, 2) researchers/engineers working on high-performance I/O for data centers, 3) domain scientists and computer scientists interested in discussing I/O issues.
The outcome of this BoF will steer the direction of the community effort.
Conference Presentation: pdf
Birds of a Feather Index