A visual network analysis method for large-scale parallel I/O systems

Carmen Sigovan, Chris Muelder, Kwan-Liu Ma, Jason Cope, Kamil Iskra, Robert Ross

Research output: Contribution to conferencePaper

13 Citations (Scopus)

Abstract

Parallel applications rely on I/O to load data, store end results, and protect partial results from being lost to system failure. Parallel I/O performance thus has a direct and significant impact on application performance. Because supercomputer I/O systems are large and complex, one cannot directly analyze their activity traces. While several visual or automated analysis tools for large-scale HPC log data exist, analysis research in the high-performance computing field is geared toward computation performance rather than I/O performance. Additionally, existing methods usually do not capture the network characteristics of HPC I/O systems. We present a visual analysis method for I/O trace data that takes into account the fact that HPC I/O systems can be represented as networks. We illustrate performance metrics in a way that facilitates the identification of abnormal behavior or performance problems. We demonstrate our approach on I/O traces collected from existing systems at different scales.

Original languageEnglish (US)
Pages308-319
Number of pages12
DOIs
StatePublished - Oct 7 2013
Event27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013 - Boston, MA, United States
Duration: May 20 2013May 24 2013

Other

Other27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013
CountryUnited States
CityBoston, MA
Period5/20/135/24/13

Fingerprint

Electric network analysis
Supercomputers

Keywords

  • Graph
  • Parallel I/O
  • Performance Analysis
  • Visualization

ASJC Scopus subject areas

  • Software

Cite this

Sigovan, C., Muelder, C., Ma, K-L., Cope, J., Iskra, K., & Ross, R. (2013). A visual network analysis method for large-scale parallel I/O systems. 308-319. Paper presented at 27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013, Boston, MA, United States. https://doi.org/10.1109/IPDPS.2013.96

A visual network analysis method for large-scale parallel I/O systems. / Sigovan, Carmen; Muelder, Chris; Ma, Kwan-Liu; Cope, Jason; Iskra, Kamil; Ross, Robert.

2013. 308-319 Paper presented at 27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013, Boston, MA, United States.

Research output: Contribution to conferencePaper

Sigovan, C, Muelder, C, Ma, K-L, Cope, J, Iskra, K & Ross, R 2013, 'A visual network analysis method for large-scale parallel I/O systems' Paper presented at 27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013, Boston, MA, United States, 5/20/13 - 5/24/13, pp. 308-319. https://doi.org/10.1109/IPDPS.2013.96
Sigovan C, Muelder C, Ma K-L, Cope J, Iskra K, Ross R. A visual network analysis method for large-scale parallel I/O systems. 2013. Paper presented at 27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013, Boston, MA, United States. https://doi.org/10.1109/IPDPS.2013.96
Sigovan, Carmen ; Muelder, Chris ; Ma, Kwan-Liu ; Cope, Jason ; Iskra, Kamil ; Ross, Robert. / A visual network analysis method for large-scale parallel I/O systems. Paper presented at 27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013, Boston, MA, United States.12 p.
@conference{f0a9ee1ad14d42dd8682df38011f0c0b,
title = "A visual network analysis method for large-scale parallel I/O systems",
abstract = "Parallel applications rely on I/O to load data, store end results, and protect partial results from being lost to system failure. Parallel I/O performance thus has a direct and significant impact on application performance. Because supercomputer I/O systems are large and complex, one cannot directly analyze their activity traces. While several visual or automated analysis tools for large-scale HPC log data exist, analysis research in the high-performance computing field is geared toward computation performance rather than I/O performance. Additionally, existing methods usually do not capture the network characteristics of HPC I/O systems. We present a visual analysis method for I/O trace data that takes into account the fact that HPC I/O systems can be represented as networks. We illustrate performance metrics in a way that facilitates the identification of abnormal behavior or performance problems. We demonstrate our approach on I/O traces collected from existing systems at different scales.",
keywords = "Graph, Parallel I/O, Performance Analysis, Visualization",
author = "Carmen Sigovan and Chris Muelder and Kwan-Liu Ma and Jason Cope and Kamil Iskra and Robert Ross",
year = "2013",
month = "10",
day = "7",
doi = "10.1109/IPDPS.2013.96",
language = "English (US)",
pages = "308--319",
note = "27th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2013 ; Conference date: 20-05-2013 Through 24-05-2013",

}

TY - CONF

T1 - A visual network analysis method for large-scale parallel I/O systems

AU - Sigovan, Carmen

AU - Muelder, Chris

AU - Ma, Kwan-Liu

AU - Cope, Jason

AU - Iskra, Kamil

AU - Ross, Robert

PY - 2013/10/7

Y1 - 2013/10/7

N2 - Parallel applications rely on I/O to load data, store end results, and protect partial results from being lost to system failure. Parallel I/O performance thus has a direct and significant impact on application performance. Because supercomputer I/O systems are large and complex, one cannot directly analyze their activity traces. While several visual or automated analysis tools for large-scale HPC log data exist, analysis research in the high-performance computing field is geared toward computation performance rather than I/O performance. Additionally, existing methods usually do not capture the network characteristics of HPC I/O systems. We present a visual analysis method for I/O trace data that takes into account the fact that HPC I/O systems can be represented as networks. We illustrate performance metrics in a way that facilitates the identification of abnormal behavior or performance problems. We demonstrate our approach on I/O traces collected from existing systems at different scales.

AB - Parallel applications rely on I/O to load data, store end results, and protect partial results from being lost to system failure. Parallel I/O performance thus has a direct and significant impact on application performance. Because supercomputer I/O systems are large and complex, one cannot directly analyze their activity traces. While several visual or automated analysis tools for large-scale HPC log data exist, analysis research in the high-performance computing field is geared toward computation performance rather than I/O performance. Additionally, existing methods usually do not capture the network characteristics of HPC I/O systems. We present a visual analysis method for I/O trace data that takes into account the fact that HPC I/O systems can be represented as networks. We illustrate performance metrics in a way that facilitates the identification of abnormal behavior or performance problems. We demonstrate our approach on I/O traces collected from existing systems at different scales.

KW - Graph

KW - Parallel I/O

KW - Performance Analysis

KW - Visualization

UR - http://www.scopus.com/inward/record.url?scp=84884868039&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84884868039&partnerID=8YFLogxK

U2 - 10.1109/IPDPS.2013.96

DO - 10.1109/IPDPS.2013.96

M3 - Paper

SP - 308

EP - 319

ER -