JIOPAT

JIOPAT is a toolkit including many tools/libs/systems for I/O Pattern study. It includes the following:

• LibPatternDetector: This is a library for pattern detection. The pattern detection algorithm is from I/O Acceleration with Pattern Detection (HPDC'2013).
• RIOSEA: RIOSEA project lets you use R to visualize I/O behaviors and performance in motion. The I/O behaviors and performance is synchronized so that you can easily relate them.
• FUSE-IO-Tracer: FUSE-Tracer records fine-granularity I/O requests with information such as hostname, offset, length, PID, operation start time, operation end time. It is very easy to use and it does not require recompilation of the application to be traced or any libraries. To use FUSE-Tracer, all you need to do is mount FUSE-Tracer on a directory and file operations performed on the mount will be recorded to a file.
• Trace-Replayer: Trace-Replayer is a tool that replays the traces collected by FUSE-Trace; it also provides some extra functionalities such as adding fixed computational delays be-tween I/O operations.
• PAP-FS: it is Pattern-Aware Prefetching File System based on FUSE. The prefetching uses I/O pattern detection algorithm proposed in I/O Acceleration with Pattern Detection (HPDC'2013). Some other prefetching techniques are also implemented. (The system is not very user-friendly at this time since it is still a prototype for the paper. It will be put here.)
• Pattern-PLFS: A version of PLFS using pattern detection algorithm to reduce the amount of indices.
• MapReplayer: It can replay traces previously collected by PLFS (generated by plfs_map) and show the performance.

Please be aware that all projects above are not production-ready yet. You take your own risk using them.

Please cite the following paper in your work if you use anything from JIOPAT.

Bibtex:

@inproceedings{he2013acceleration,
 title={I/O Acceleration with Pattern Detection},
 author={He, Jun and Bent, John and Torres, Aaron 
         and Grider, Gary and Gibson, Garth 
         and Maltzahn, Carlos and Sun, Xian-He},
 booktitle={the 22th International ACM Symposium 
            on High Performance Distributed 
            Computing (HPDC'13)},
 year={2013},
 organization={ACM}
}

Pure Text:

Jun He, John Bent, Aaron Torres, Gary Grider, Garth Gibson, Carlos Maltzahn, and Xian-He Sun, "I/O Acceleration with Pattern Detection", accepted to appear in Proc. of the 22th International ACM Symposium on High Performance Distributed Computing (HPDC'13), New York City, NY, June 2013.

Find a proper citation format here if you do not use bibtex.

Contact: Jun He (junnhe at gmail.com)

LA-UR-13-22371