Article

Article title USE OF SYSTEM MONITORING DATA TO DETERMINE FACTORS REDUSING APPLICATIONS SCALABILITY
Authors A.S. Antonov, A.M. Teplov
Section SECTION II. MATHEMATICAL AND SOFTWARE OF SUPERCOMPUTERS
Month, Year 12, 2014 @en
Index UDC 004.051
DOI
Abstract The paper describes an approach to finding the cause of limiting the scalability of parallel programs based on the use of data monitoring system. Using the approach described here can be prepared as data of a reason associated with the hardware of the computing system, and their impact on the program execution efficiency. The results of studies on the use of data monitoring system works parallel applications for the determination of the factors affecting the scalability of applications. Consider a set of test and custom applications, and reflected in the data system monitoring factors that affect scalability. To determine the effect of these factors is to use visual data analysis system monitoring and clustering methods of the data. To obtain the data used tool to generate reports on the program"s Job Digest, after which the data were analyzed using the analyzer WEKA. The results of these studies suggest the applicability of the proposed approach ble for research programs to analyze the impact of a large number of hardware factors limiting the scalability of parallel programs.

Download PDF

Keywords Scalability; efficiency; system monitoring; efficiency analysis; data clustering; parallel programming.
References 1. Weka 3: Data Mining Software in Java. Available at: http://www.cs.waikato.ac.nz/ ml/weka/.
2. Adinets A.B., Bryzgalov P.A., Voevodin Vad. Vl., Zhumatiy S.A., Nikitenko D.A., Ste-fanov K.S. Job Digest-podkhod k issledovaniyu dinamicheskikh svoystv zadach na superkomp'yuternykh sistemakh [Job Digest-approach to the study of the dynamic properties task on supercomputing systems], Vestnik Ufimskogo gos. aviatsionnogo tekhn. un-ta [Bulletin of the Ufa state aviation technical University], 2013, Vol. 17, No. 2(55), pp. 131-137.
3. Voevodin Vl.V., Zhumatiy S.A., Sobolev S.I., Antonov A.S., Bryzgalov P.I., Nikitenko D.A., Stefanov K.S., Voevodin V.V. Praktika superkomp'yutera «Lomonosov» [Practice supercomputer "Lomonosov"], Otkrytye sistemy [Open systems], 2012, No. 7, pp. 36-39.
4. Antonov A.S. SKIF MGU – osnova Superkomp'yuternogo kompleksa Moskovskogo universiteta [SKIF MSU - based Supercomputer of the University of Moscow], Vtoraya
Mezhdunarodnaya nauchnaya konferentsiya «Superkomp'yuternye sistemy i ikh primenenie» (SSA'2008): doklady konferentsii (27-29.10.2008 g., Minsk) [Second international scientific conference "Supercomputer systems and applications" (SSA'2008): report of the conference (27-29.10.2008, Minsk)]. Minsk: OIPI NAN Belarusi, 2008, pp. 7-10.
5. Antonov A.S., Teplov A.M. Issledovanie masshtabiruemosti programm s ispol'zovaniem instrumentov analiza parallel'nykh prilozheniy na primere modeli atmosfery NH3D [Study the scalability of programs using tools of analysis of parallel applications on the example of a model atmosphere NH3D], Vestnik Yuzhno-Ural'skogo gosudarstvennogo un-ta. Ser. Vychislitel'naya matematika i informatika [Bulletin of the South Ural state universitetet. A series of Computational mathematics and Informatics], 2013, Vol. 2, No. 1, pp. 5-16.
6. Antonov A.S., Zhumatiy S.A., Nikitenko D.A., Stefanov K.S., Teplov A.M., Shvets P.A. Issledovanie dinamicheskikh kharakteristik potoka zadach superkomp'yuternoy sistemy [Investigation of the dynamic characteristics of the task flow supercomputer system],
Vychislitel'nye metody i programmirovanie: Novye vychislitel'nye tekhnologii (Elektronnyy nauchnyy zhurn.) [Computational methods and programming: New computational technologies (Electronic scientific journal)], 2013, Vol. 14, pp. 104-108.
7. Nikitenko D.A., Stefanov K.S. Issledovanie effektivnosti parallel'nykh programm po dannym monitoringa [Study of the efficiency of parallel programs according to the monitoring],
Vychislitel'nye metody i programmirovanie: Novye vychislitel'nye tekhnologii (Elektronnyy nauchnyy zhurn.) [Computational methods and programming: New computational technologies (Electronic scientific journal)], 2012, Vol. 13, pp. 97-102.
8. Nikitenko D.A. Kompleksnyy analiz proizvoditel'nosti superkomp'yuternykh sistem, osnovannyy na dannykh sistemnogo monitoring [A comprehensive analysis of performance
supercomputer systems, based on the data monitoring system], Vychislitel'nye metody i programmirovanie: Novye vychislitel'nye tekhnologii (Elektronnyy nauchnyy zhurn.) [Computational methods and programming: New computational technologies (Electronic scientific journal)], 2014, Vol. 15, pp. 85-97.

Comments are closed.