Monitoring and evaluating I/O performance of HPC systems

Manos Farsarakis; Michele Weiland; William Jackson; Mark Parsons

Abstract / Description of output

In this extended abstract, we describe our work, as part of the NEXTGenIO project, around system-wide monitoring of workloads and I/O on large HPC systems. Several factors mean that monitoring at this level is a difficult problem: there should be little or no performance limiting overhead associated with monitoring; the system consists of a multitude of components, which each have different monitoring interfaces; and the large number of users, jobs and applications means that the amount of data that is collected from monitoring is not trivial. In this abstract, we describe how we aim to address this challenge in NEXTGenIO.

Original language	English
Publication status	Published - 29 Apr 2016
Event	4th International Exascale Applications and Software (EASC) - Stockholm City Conference Centre, Stockholm, Sweden Duration: 25 Apr 2016 → 29 Apr 2016

Conference

Conference	4th International Exascale Applications and Software (EASC)
Country/Territory	Sweden
City	Stockholm
Period	25/04/16 → 29/04/16

Access to Document

https://drive.google.com/file/d/0B1Smvx6Sn4rzTU16UmFRREtRblE/viewLicence: Unspecified

NEXTGenIO: Next Generation I/O for Exascale
Parsons, M.
EU government bodies
1/10/15 → 30/09/19
Project: Research

Cite this

@conference{86e57fa04d084f1499210e3eeaf3560d,

title = "Monitoring and evaluating I/O performance of HPC systems",

abstract = "In this extended abstract, we describe our work, as part of the NEXTGenIO project, around system-wide monitoring of workloads and I/O on large HPC systems. Several factors mean that monitoring at this level is a difficult problem: there should be little or no performance limiting overhead associated with monitoring; the system consists of a multitude of components, which each have different monitoring interfaces; and the large number of users, jobs and applications means that the amount of data that is collected from monitoring is not trivial. In this abstract, we describe how we aim to address this challenge in NEXTGenIO.",

author = "Manos Farsarakis and Michele Weiland and William Jackson and Mark Parsons",

year = "2016",

month = apr,

day = "29",

language = "English",

note = "4th International Exascale Applications and Software (EASC) ; Conference date: 25-04-2016 Through 29-04-2016",

}

TY - CONF

T1 - Monitoring and evaluating I/O performance of HPC systems

AU - Farsarakis, Manos

AU - Weiland, Michele

AU - Jackson, William

AU - Parsons, Mark

PY - 2016/4/29

Y1 - 2016/4/29

N2 - In this extended abstract, we describe our work, as part of the NEXTGenIO project, around system-wide monitoring of workloads and I/O on large HPC systems. Several factors mean that monitoring at this level is a difficult problem: there should be little or no performance limiting overhead associated with monitoring; the system consists of a multitude of components, which each have different monitoring interfaces; and the large number of users, jobs and applications means that the amount of data that is collected from monitoring is not trivial. In this abstract, we describe how we aim to address this challenge in NEXTGenIO.

AB - In this extended abstract, we describe our work, as part of the NEXTGenIO project, around system-wide monitoring of workloads and I/O on large HPC systems. Several factors mean that monitoring at this level is a difficult problem: there should be little or no performance limiting overhead associated with monitoring; the system consists of a multitude of components, which each have different monitoring interfaces; and the large number of users, jobs and applications means that the amount of data that is collected from monitoring is not trivial. In this abstract, we describe how we aim to address this challenge in NEXTGenIO.

M3 - Abstract

T2 - 4th International Exascale Applications and Software (EASC)

Y2 - 25 April 2016 through 29 April 2016

ER -

Monitoring and evaluating I/O performance of HPC systems

Abstract / Description of output

Conference

Access to Document

Fingerprint

Projects

NEXTGenIO: Next Generation I/O for Exascale

Cite this