Distributed configurable application monitoring on SMP clusters

Karl Fürlinger; Michael Gerndt

doi:10.1007/978-3-540-39924-7_58

Distributed configurable application monitoring on SMP clusters

Karl Fürlinger, Michael Gerndt

Informatics 10 - Associate Professorship of Architecture of Parallel and Distributed Systems

Technical University of Munich

Research output: Contribution to journal › Article › peer-review

1 Scopus citations

Abstract

Performance analysis of applications on large clusters of SMPs requires a monitoring approach that supports tools realizing concepts like automation, distribution and on-line operations. Key goals are a minimization of the perturbation of the target application and flexibility and efficiency with respect to data pre-processing and filtering. To achieve these goals, our approach separates the monitor into a passive monitoring library linked to the application and an active 'runtime information producer' (RIP) which handles monitoring requests and performs pre-processing (e.g., aggregation) of performance data for individual cluster nodes. A directory service can be queried to discover which RIPs handle which nodes.

Original language	English
Pages (from-to)	429-437
Number of pages	9
Journal	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	2840
DOIs	https://doi.org/10.1007/978-3-540-39924-7_58
State	Published - 2003

Access to Document

10.1007/978-3-540-39924-7_58

Cite this

@article{841ce8b7b7aa42818e1791402a909077,

title = "Distributed configurable application monitoring on SMP clusters",

abstract = "Performance analysis of applications on large clusters of SMPs requires a monitoring approach that supports tools realizing concepts like automation, distribution and on-line operations. Key goals are a minimization of the perturbation of the target application and flexibility and efficiency with respect to data pre-processing and filtering. To achieve these goals, our approach separates the monitor into a passive monitoring library linked to the application and an active 'runtime information producer' (RIP) which handles monitoring requests and performs pre-processing (e.g., aggregation) of performance data for individual cluster nodes. A directory service can be queried to discover which RIPs handle which nodes.",

author = "Karl F{\"u}rlinger and Michael Gerndt",

year = "2003",

doi = "10.1007/978-3-540-39924-7\_58",

language = "English",

volume = "2840",

pages = "429--437",

journal = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

issn = "0302-9743",

publisher = "Springer Science and Business Media Deutschland GmbH",

}

TY - JOUR

T1 - Distributed configurable application monitoring on SMP clusters

AU - Fürlinger, Karl

AU - Gerndt, Michael

PY - 2003

Y1 - 2003

N2 - Performance analysis of applications on large clusters of SMPs requires a monitoring approach that supports tools realizing concepts like automation, distribution and on-line operations. Key goals are a minimization of the perturbation of the target application and flexibility and efficiency with respect to data pre-processing and filtering. To achieve these goals, our approach separates the monitor into a passive monitoring library linked to the application and an active 'runtime information producer' (RIP) which handles monitoring requests and performs pre-processing (e.g., aggregation) of performance data for individual cluster nodes. A directory service can be queried to discover which RIPs handle which nodes.

AB - Performance analysis of applications on large clusters of SMPs requires a monitoring approach that supports tools realizing concepts like automation, distribution and on-line operations. Key goals are a minimization of the perturbation of the target application and flexibility and efficiency with respect to data pre-processing and filtering. To achieve these goals, our approach separates the monitor into a passive monitoring library linked to the application and an active 'runtime information producer' (RIP) which handles monitoring requests and performs pre-processing (e.g., aggregation) of performance data for individual cluster nodes. A directory service can be queried to discover which RIPs handle which nodes.

UR - https://www.scopus.com/pages/publications/35248854595

U2 - 10.1007/978-3-540-39924-7_58

DO - 10.1007/978-3-540-39924-7_58

M3 - Article

AN - SCOPUS:35248854595

SN - 0302-9743

VL - 2840

SP - 429

EP - 437

JO - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

JF - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -

Distributed configurable application monitoring on SMP clusters

Abstract

Access to Document

Other files and links

Fingerprint

Cite this