ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization

Eric R. Schendel; Saurabh V. Pendse; John Jenkins; David A. Boyuka; Zhenhuan Gong; Sriram Lakshminarasimhan; Qing Liu; Hemanth Kolla; Jackie Chen; Scott Klasky; Robert Ross; Nagiza F. Samatova

doi:10.1145/2287076.2287086

ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization

Eric R. Schendel
, Saurabh V. Pendse
, John Jenkins
, David A. Boyuka
, Zhenhuan Gong
, Sriram Lakshminarasimhan
, Qing Liu
, Hemanth Kolla
, Jackie Chen
, Scott Klasky
, Robert Ross
, Nagiza F. Samatova

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

32 Scopus citations

Abstract

Current peta-scale data analytics frameworks suffer from a significant performance bottleneck due to an imbalance between their enormous computational power and limited I/O bandwidth. Using data compression schemes to reduce the amount of I/O activity is a promising approach to addressing this problem. In this paper, we propose a hybrid framework for interleaving I/O with data compression to achieve improved I/O throughput side-by-side with reduced dataset size. We evaluate several interleaving strategies, present theoretical models, and evaluate the efficiency and scalability of our approach through comparative analysis. With our theoretical model, considering 19 real-world scientific datasets both from the public domain and peta-scale simulations, we estimate that the hybrid method can result in a 12 to 46% increase in throughput on hard-to-compress scientific datasets. At the reported peak bandwidth of 60 GB/s of uncompressed data for a current, leadership-class parallel I/O system, this translates into an effective gain of 7 to 28 GB/s in aggregate throughput.

Original language	English (US)
Title of host publication	HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing
Pages	61-72
Number of pages	12
DOIs	https://doi.org/10.1145/2287076.2287086
State	Published - 2012
Externally published	Yes
Event	21st ACM Symposium on High-Performance Parallel and Distributed Computing, HPDC '12 - Delft, Netherlands Duration: Jun 18 2012 → Jun 22 2012

Publication series

Name	HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing

Other

Other	21st ACM Symposium on High-Performance Parallel and Distributed Computing, HPDC '12
Country/Territory	Netherlands
City	Delft
Period	6/18/12 → 6/22/12

All Science Journal Classification (ASJC) codes

Software

Keywords

High Performance Computing
Hybrid Interleaving
I/O
ISOBAR
Lossless Compression
Staging

Access to Document

10.1145/2287076.2287086

Cite this

Schendel, E. R., Pendse, S. V., Jenkins, J., Boyuka, D. A., Gong, Z., Lakshminarasimhan, S., Liu, Q., Kolla, H., Chen, J., Klasky, S., Ross, R., & Samatova, N. F. (2012). ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization. In HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing (pp. 61-72). (HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing). https://doi.org/10.1145/2287076.2287086

Schendel, Eric R. ; Pendse, Saurabh V. ; Jenkins, John et al. / ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization. HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing. 2012. pp. 61-72 (HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing).

@inproceedings{6f0663d25b4d4cd98df3dcfb8ee062d7,

title = "ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization",

abstract = "Current peta-scale data analytics frameworks suffer from a significant performance bottleneck due to an imbalance between their enormous computational power and limited I/O bandwidth. Using data compression schemes to reduce the amount of I/O activity is a promising approach to addressing this problem. In this paper, we propose a hybrid framework for interleaving I/O with data compression to achieve improved I/O throughput side-by-side with reduced dataset size. We evaluate several interleaving strategies, present theoretical models, and evaluate the efficiency and scalability of our approach through comparative analysis. With our theoretical model, considering 19 real-world scientific datasets both from the public domain and peta-scale simulations, we estimate that the hybrid method can result in a 12 to 46\% increase in throughput on hard-to-compress scientific datasets. At the reported peak bandwidth of 60 GB/s of uncompressed data for a current, leadership-class parallel I/O system, this translates into an effective gain of 7 to 28 GB/s in aggregate throughput.",

keywords = "High Performance Computing, Hybrid Interleaving, I/O, ISOBAR, Lossless Compression, Staging",

author = "Schendel, \{Eric R.\} and Pendse, \{Saurabh V.\} and John Jenkins and Boyuka, \{David A.\} and Zhenhuan Gong and Sriram Lakshminarasimhan and Qing Liu and Hemanth Kolla and Jackie Chen and Scott Klasky and Robert Ross and Samatova, \{Nagiza F.\}",

year = "2012",

doi = "10.1145/2287076.2287086",

language = "English (US)",

isbn = "9781450308052",

series = "HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing",

pages = "61--72",

booktitle = "HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing",

note = "21st ACM Symposium on High-Performance Parallel and Distributed Computing, HPDC '12 ; Conference date: 18-06-2012 Through 22-06-2012",

}

Schendel, ER, Pendse, SV, Jenkins, J, Boyuka, DA, Gong, Z, Lakshminarasimhan, S, Liu, Q, Kolla, H, Chen, J, Klasky, S, Ross, R & Samatova, NF 2012, ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization. in HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing. HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing, pp. 61-72, 21st ACM Symposium on High-Performance Parallel and Distributed Computing, HPDC '12, Delft, Netherlands, 6/18/12. https://doi.org/10.1145/2287076.2287086

ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization. / Schendel, Eric R.; Pendse, Saurabh V.; Jenkins, John et al.
HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing. 2012. p. 61-72 (HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization

AU - Schendel, Eric R.

AU - Pendse, Saurabh V.

AU - Jenkins, John

AU - Boyuka, David A.

AU - Gong, Zhenhuan

AU - Lakshminarasimhan, Sriram

AU - Liu, Qing

AU - Kolla, Hemanth

AU - Chen, Jackie

AU - Klasky, Scott

AU - Ross, Robert

AU - Samatova, Nagiza F.

PY - 2012

Y1 - 2012

N2 - Current peta-scale data analytics frameworks suffer from a significant performance bottleneck due to an imbalance between their enormous computational power and limited I/O bandwidth. Using data compression schemes to reduce the amount of I/O activity is a promising approach to addressing this problem. In this paper, we propose a hybrid framework for interleaving I/O with data compression to achieve improved I/O throughput side-by-side with reduced dataset size. We evaluate several interleaving strategies, present theoretical models, and evaluate the efficiency and scalability of our approach through comparative analysis. With our theoretical model, considering 19 real-world scientific datasets both from the public domain and peta-scale simulations, we estimate that the hybrid method can result in a 12 to 46% increase in throughput on hard-to-compress scientific datasets. At the reported peak bandwidth of 60 GB/s of uncompressed data for a current, leadership-class parallel I/O system, this translates into an effective gain of 7 to 28 GB/s in aggregate throughput.

AB - Current peta-scale data analytics frameworks suffer from a significant performance bottleneck due to an imbalance between their enormous computational power and limited I/O bandwidth. Using data compression schemes to reduce the amount of I/O activity is a promising approach to addressing this problem. In this paper, we propose a hybrid framework for interleaving I/O with data compression to achieve improved I/O throughput side-by-side with reduced dataset size. We evaluate several interleaving strategies, present theoretical models, and evaluate the efficiency and scalability of our approach through comparative analysis. With our theoretical model, considering 19 real-world scientific datasets both from the public domain and peta-scale simulations, we estimate that the hybrid method can result in a 12 to 46% increase in throughput on hard-to-compress scientific datasets. At the reported peak bandwidth of 60 GB/s of uncompressed data for a current, leadership-class parallel I/O system, this translates into an effective gain of 7 to 28 GB/s in aggregate throughput.

KW - High Performance Computing

KW - Hybrid Interleaving

KW - I/O

KW - ISOBAR

KW - Lossless Compression

KW - Staging

UR - https://www.scopus.com/pages/publications/84863890077

UR - https://www.scopus.com/pages/publications/84863890077#tab=citedBy

U2 - 10.1145/2287076.2287086

DO - 10.1145/2287076.2287086

M3 - Conference contribution

AN - SCOPUS:84863890077

SN - 9781450308052

T3 - HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing

SP - 61

EP - 72

BT - HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing

T2 - 21st ACM Symposium on High-Performance Parallel and Distributed Computing, HPDC '12

Y2 - 18 June 2012 through 22 June 2012

ER -

Schendel ER, Pendse SV, Jenkins J, Boyuka DA, Gong Z, Lakshminarasimhan S et al. ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization. In HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing. 2012. p. 61-72. (HPDC '12 - Proceedings of the 21st ACM Symposium on High-Performance Parallel and Distributed Computing). doi: 10.1145/2287076.2287086

ISOBAR hybrid compression-I/O interleaving for large-scale parallel I/O optimization

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Keywords

Access to Document

Other files and links

Fingerprint

Cite this