TY - GEN
T1 - An experimental performance evaluation of touchstone delta concurrent file system
AU - Bordawekar, Rajesh R.
AU - Choudhary, Alok N.
AU - Del Rosario, Juan Miguel
N1 - Funding Information:
“ This work was sponsored in part by DARPA under contract # DABT63-91-C-0028 and NSF MIP-911081O, The content of the information does not necessarily reflect the position or the POIICY of the Government and no official endorsement should be ]nferred.
Publisher Copyright:
© 1993 ACM.
PY - 1993/8/1
Y1 - 1993/8/1
N2 - For a high-performance parallel machine to be a scalable system, it must also have a scalable parallel I/O system. This paper presents an experimental evaluation of the Intel Touchstone Delta's Concurrent File System (CFS). The main objective of the study is to determine the maximum file read/write rates for various configurations of I/O and compute nodes. In addition, we study the effects of file access modes, buffer sizes and file sizes on the system performance. In most cases, the result shows that performance of CFS scales as the number of disks is increased, but the sustained performance improvements are much lower than the system's peak capacity. We observe that the performance of CFS scales with the number of processors in the beginning, however, a plateu a quickly reached due to the I/O system bottleneck and enormous software overhead, especially that of synchronization. Finally we also show that the performance of the CFS can greatly vary for various data distributions commonly employed in scientific and engineering applications.
AB - For a high-performance parallel machine to be a scalable system, it must also have a scalable parallel I/O system. This paper presents an experimental evaluation of the Intel Touchstone Delta's Concurrent File System (CFS). The main objective of the study is to determine the maximum file read/write rates for various configurations of I/O and compute nodes. In addition, we study the effects of file access modes, buffer sizes and file sizes on the system performance. In most cases, the result shows that performance of CFS scales as the number of disks is increased, but the sustained performance improvements are much lower than the system's peak capacity. We observe that the performance of CFS scales with the number of processors in the beginning, however, a plateu a quickly reached due to the I/O system bottleneck and enormous software overhead, especially that of synchronization. Finally we also show that the performance of the CFS can greatly vary for various data distributions commonly employed in scientific and engineering applications.
UR - http://www.scopus.com/inward/record.url?scp=85031100675&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85031100675&partnerID=8YFLogxK
U2 - 10.1145/165939.166013
DO - 10.1145/165939.166013
M3 - Conference contribution
AN - SCOPUS:85031100675
T3 - Proceedings of the International Conference on Supercomputing
SP - 367
EP - 376
BT - Proceedings of the 7th International Conference on Supercomputing, ICS 1993
PB - Association for Computing Machinery
T2 - 7th International Conference on Supercomputing, ICS 1993
Y2 - 19 July 1993 through 23 July 1993
ER -