An In-Depth I/O Pattern Analysis in HPC Systems

被引：5

作者：

Bang, Jiwoo ^{[1
]}

Kim, Chungyong ^{[1
]}

Wu, Kesheng ^{[2
]}

Sim, Alex ^{[2
]}

Byna, Suren ^{[2
]}

Sung, Hanul ^{[3
]}

Eom, Hyeonsang ^{[1
]}

机构：

[1] Seoul Natl Univ, Dept Comp Sci & Engn, Seoul, South Korea

[2] Lawrence Berkeley Natl Lab, Computat Res Div, Berkeley, CA USA

[3] Sangmyung Univ, Dept Game Design & Dev, Seoul, South Korea

来源：

2021 IEEE 28TH INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING, DATA, AND ANALYTICS (HIPC 2021) | 2021年

基金：

新加坡国家研究基金会;

关键词：

I/O characteristic; Unsupervised learning; Feature selection; Clustering; Prediction model; High performance computing;

D O I：

10.1109/HiPC53243.2021.00056

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

High-performance computing (HPC) systems consist of thousands of compute nodes, storage systems and high-speed networks, providing multiple layers of I/O stack with high complexity. By adjusting the diverse configuration settings that HPC systems provide, the I/O performance of applications can be improved. However, it is challenging to identify the optimal configuration settings without a thorough knowledge of the system, as each of the different I/O characteristics of applications can be an important factor for parameter decision. In this paper, we use multiple machine learning approaches to perform an in-depth analysis on I/O behaviors of HPC applications and to search for the optimal configuration settings for jobs sharing similar I/O characteristics. Improved by maximum 0.07 H-squared score, our results in overall show that jobs run on the HPC systems can obtain the predicted I/O performance for different configuration parameters with a high accuracy, using the proposed machine learning-based prediction models.

引用

页码：400 / 405

页数：6

共 22 条

[1]

Ali-Eldin A., 2013, Workload classification for efficient auto-scaling of cloud resources

[2]

[Anonymous], 2017, P 26 INT S HIGH PERF, DOI DOI 10.1145/3078597.3078614

[3] HPAS: An HPC Performance Anomaly Suite for Reproducing Performance Variations [J].

Ates, Emre ;

Zhang, Yijia ;

Aksar, Burak ;

Brandt, Jim ;

Leung, Vitus J. ;

Egele, Manuel ;

Coskun, Ayse K. .

PROCEEDINGS OF THE 48TH INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING (ICPP 2019), 2019,

[4] SMOTE: Synthetic minority over-sampling technique [J].

Chawla, Nitesh V. ;

Bowyer, Kevin W. ;

Hall, Lawrence O. ;

Kegelmeyer, W. Philip .

2002, American Association for Artificial Intelligence (16)

[5] CLUSTER SEPARATION MEASURE [J].

DAVIES, DL ;

BOULDIN, DW .

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1979, 1 (02) :224-227

[6] Gauge: An Interactive Data-Driven Visualization Tool for HPC Application I/O Performance Analysis [J].

del Rosario, Eliakin ;

Currier, Mikaela ;

Isakov, Mihailo ;

Madireddy, Sandeep ;

Balaprakash, Prasanna ;

Carns, Philip ;

Ross, Robert B. ;

Harms, Kevin ;

Snyder, Shane ;

Kinsy, Michel A. .

PROCEEDINGS OF 2020 IEEE/ACM FIFTH INTERNATIONAL PARALLEL DATA SYSTEMS WORKSHOP (PDSW 2020), 2020, :15-21

[7]

Li DY, 2019, 2019 INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE BIG DATA AND INTELLIGENT SYSTEMS (HPBD&IS), P14, DOI [10.1109/HPBDIS.2019.8735467, 10.1109/hpbdis.2019.8735467]

[8]

Lux T., 2018, PREDICTIVE MODELING

[9]

Lv YR, 2018, 2018 51ST ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE (MICRO), P613, DOI [10.1109/MICR0.2018.00056, 10.1109/MICRO.2018.00056]

[10] Machine Learning Based Parallel I/O Predictive Modeling: A Case Study on Lustre File Systems [J].

Madireddy, Sandeep ;

Balaprakash, Prasanna ;

Carns, Philip ;

Latham, Robert ;

Ross, Robert ;

Snyder, Shane ;

Wild, Stefan M. .

HIGH PERFORMANCE COMPUTING, ISC HIGH PERFORMANCE 2018, 2018, 10876 :184-204

← 1 2 3 →