Analysis of delays structure of interconnections in supercomputer by means of DBScan and divisive clustering algorithms
https://doi.org/10.23947/2587-8999-2018-2-1-33-43
Abstract
In this paper we propose method for estimating and analysis measurements of delays in the computational cluster interconnection subsystem. Delays are combined into the set of pairs (source, destination). We have measurements of delays extracted by network_test2 utility from interconnections of following supercomputers: BlueGene/P, Lomonosov-1, Lomonosov-2 (Lomonosov MSU) and Jurope (Julich). We have clustered pairs of delays by DBscan and Divisive algorithms. Results of clusterisation revealed that DBScan is more accurate algorithm then divisive and allows to extract clusters, which correspond to the actual features in the supercomputer interconnections. Clusters gather near the same components of supercomputer network infrastructure. Gained clusters were visualized in 2-D by special tool, developed by authors.
About the Authors
Alexey Nikolaevich Salnikov
Lomonosov Moscow State University
(Leninskie Gory 1, Moscow, Russian Federation)
Russian Federation
Russian Federation
Salnikov Alexey Nikolaevich, Lomonosov Moscow State University (Leninskie Gory 1, Moscow, Russian Federation), Senior Researcher, Candidate of Science in Physics and Maths
Artur Andreevich Begaev
Lomonosov Moscow State University
(Leninskie Gory 1, Moscow, Russian Federation)
Russian Federation
Russian Federation
Begaev Artur Andreevich, Lomonosov Moscow State University (Leninskie Gory 1, Moscow, Russian Federation)
Archil Iverievich Maysuradze
Lomonosov Moscow State University
(Leninskie Gory 1, Moscow, Russian Federation)
Russian Federation
Russian Federation
Maysuradze Archil Iverievich, Lomonosov Moscow State University (Leninskie Gory 1, Moscow, Russian Federation), Associate Professor, Candidate of Science in Physics and Maths
References
1. Brian S. Everitt, Landau S., Leese M., Stahl D., Cluster Analysis, 5th edition // Wiley, 2011, pp. 49-53, 84-88, 220 – 222.
2. Charu C. Aggarwal, Chandan K. Reddy, Data Clustering. Algorithms and Applications // CRC Press, 2014, pp. 5 – 8.
3. A. Peiravi, A fast algorithm for connectivity graph approximation using modified Manhattan distance in dynamic networks // 2008 – Applied Mathematics and Computation, pp. 319-332
4. Salnikov A.N., Andreev D.Yu., Lebedev D.Yu., Toolkit for analyzing the communication environment characteristics of a computational cluster based on MPI standard functions // 2012. ‒ Moscow University Computational Mathematics and Cybernetics, Vol. 36, no. 1, Moscow, publishing house Moscow State University, pp. 41-49, DOI: 10.3103/S0278641912010074.
5. Salnikov A.N., Maysuradze A.I., Andreev D.Yu., Kostin G.A., Klasterizatsiya rezul'tatov testirovaniya kommunikatsionnoy sredy mnogoprotsessornykh sistem: edenicy analiza, issledovaniye metodov, vizualizatsiya rezul'tatov // 2012. ‒ Vestnik of UGATU, Vol. 16, no. 6, pp. 149-157.
6. Ester, Martin; Kriegel, Hans-Peter; Sander, Jörg; Xu, Xiaowei and other. A density-based algorithm for discovering clusters in large spatial databases with noise. // 1996. ‒ Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96), AAAI Press. pp. 226–231
7. Clustbench benchmarking suite: https://github.com/clustbench.
Review
For citations:
Salnikov A.N., Begaev A.A., Maysuradze A.I. Analysis of delays structure of interconnections in supercomputer by means of DBScan and divisive clustering algorithms. Computational Mathematics and Information Technologies. 2018;2(1). https://doi.org/10.23947/2587-8999-2018-2-1-33-43
Views:
128
Email this article 