SC is the International Conference for
 High Performnance Computing, Networking, Storage and Analysis

SCHEDULE: NOV 13-19, 2010

IOrchestrator: Improving the Performance of Multi-node I/O Systems via Inter-Server Coordination

SESSION: System I/O Optimization


TIME: 2:30PM - 3:00PM


AUTHOR(S):Xuechen Zhang, Kei Davis, Song Jiang


To exploit I/O parallelism, parallel file systems usually stripe file data across multiple data servers. While a program's spatial locality is critical to its I/O performance, existing disk schedulers, such as anticipatory scheduling, are incapable of exploiting this locality when the data servers concurrently serve synchronous requests from different programs. This problem becomes especially acute when many parallel programs rely on techniques such as collective I/O to increase spatial locality and thereby improve I/O performance. We propose a scheme, IOrchestrator, to recover the lost I/O performance in multi-server I/O systems by orchestrating I/O requests from different programs across the data servers when such coordination is deemed cost effective. We have implemented IOrchestrator in the PVFS2 parallel file system and tested it with representative parallel benchmarks. Our experiments show that IOrchestrator can significantly improve I/O performance delivered by a cluster of data servers servicing concurrently-running parallel programs.

Chair/Author Details:

Tevfik Kosar (Chair) - Louisiana State University

Xuechen Zhang - Wayne State University

Kei Davis - Los Alamos National Laboratory

Song Jiang - Wayne State University

Add to iCal  Click here to download .ics calendar file

Add to Outlook  Click here to download .vcs calendar file

Add to Google Calendarss  Click here to add event to your Google Calendar

The full paper can be found in the ACM Digital Library and IEEE Computer Society

   Sponsors    IEEE    ACM