### Abstract

In the context of generating efficient, contention free schedules for inter-node communication through a switch fabric in cluster computing or data center type environments, all-to-all scheduling with equal sized data transfer requests has been studied in the literature [1,3,4]. In this paper, we propose a communication scheduling module (CSM) towards generating contention free communication schedules for many-to-many communication with arbitrary sized data. Towards this end, we propose three approximation algorithms - PST, LDT and SDT. From time to time, the CSM first generates a bipartite graph from the set of received requests, then determines which of these three algorithms gives the best approximation factor on this graph and finally executes that algorithm to generate a contention free schedule. Algorithm PST has a worst case run time of O( max (Δ|E|, |E|log(|E|))) and guarantees an approximation factor of 2H _{2Δ-1}, where |E| is the number of edges in the bipartite graph, Δ is the maximum node degree of the bipartite graph and H _{2Δ-1} is the (2Δ-1)-th harmonic number. LDT runs in O(|E|^{2}) and has an approximation factor of 2(1+τ), where τ is a constant defined as a guard band or pause time to eliminate the possibility of contention (in an apparently contention free schedule) caused by system jitter and synchronization inaccuracies between the nodes. SDT gives an approximation factor of 4log(w _{max} ) and has a worst case run time of O(Δ|E|log(w _{max} )), where w _{max} represents the longest communication time in a set of received requests.

Original language | English |
---|---|

Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |

Pages | 150-161 |

Number of pages | 12 |

Volume | 6536 LNCS |

DOIs | |

Publication status | Published - 21 Feb 2011 |

Externally published | Yes |

Event | 7th International Conference on Distributed Computing and Internet Technology, ICDCIT 2011 - Bhubaneshwar, India Duration: 9 Feb 2011 → 12 Feb 2011 |

### Publication series

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

Volume | 6536 LNCS |

ISSN (Print) | 03029743 |

ISSN (Electronic) | 16113349 |

### Other

Other | 7th International Conference on Distributed Computing and Internet Technology, ICDCIT 2011 |
---|---|

Country | India |

City | Bhubaneshwar |

Period | 9/2/11 → 12/2/11 |

### Fingerprint

### Keywords

- approximation algorithms
- cluster computing
- contention free schedule
- data centers
- Many-to-many scheduling
- switch fabric
- switch scheduling

### ASJC Scopus subject areas

- Computer Science(all)
- Theoretical Computer Science

### Cite this

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)*(Vol. 6536 LNCS, pp. 150-161). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6536 LNCS). https://doi.org/10.1007/978-3-642-19056-8_10

**Contention-free many-to-many communication scheduling for high performance clusters.** / Banerjee, Satyajit; Datta Chowdhury, Atish; Sinha, Koushik; Ghosh, Subhas Kumar.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics).*vol. 6536 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6536 LNCS, pp. 150-161, 7th International Conference on Distributed Computing and Internet Technology, ICDCIT 2011, Bhubaneshwar, India, 9/2/11. https://doi.org/10.1007/978-3-642-19056-8_10

}

TY - GEN

T1 - Contention-free many-to-many communication scheduling for high performance clusters

AU - Banerjee, Satyajit

AU - Datta Chowdhury, Atish

AU - Sinha, Koushik

AU - Ghosh, Subhas Kumar

PY - 2011/2/21

Y1 - 2011/2/21

N2 - In the context of generating efficient, contention free schedules for inter-node communication through a switch fabric in cluster computing or data center type environments, all-to-all scheduling with equal sized data transfer requests has been studied in the literature [1,3,4]. In this paper, we propose a communication scheduling module (CSM) towards generating contention free communication schedules for many-to-many communication with arbitrary sized data. Towards this end, we propose three approximation algorithms - PST, LDT and SDT. From time to time, the CSM first generates a bipartite graph from the set of received requests, then determines which of these three algorithms gives the best approximation factor on this graph and finally executes that algorithm to generate a contention free schedule. Algorithm PST has a worst case run time of O( max (Δ|E|, |E|log(|E|))) and guarantees an approximation factor of 2H 2Δ-1, where |E| is the number of edges in the bipartite graph, Δ is the maximum node degree of the bipartite graph and H 2Δ-1 is the (2Δ-1)-th harmonic number. LDT runs in O(|E|2) and has an approximation factor of 2(1+τ), where τ is a constant defined as a guard band or pause time to eliminate the possibility of contention (in an apparently contention free schedule) caused by system jitter and synchronization inaccuracies between the nodes. SDT gives an approximation factor of 4log(w max ) and has a worst case run time of O(Δ|E|log(w max )), where w max represents the longest communication time in a set of received requests.

AB - In the context of generating efficient, contention free schedules for inter-node communication through a switch fabric in cluster computing or data center type environments, all-to-all scheduling with equal sized data transfer requests has been studied in the literature [1,3,4]. In this paper, we propose a communication scheduling module (CSM) towards generating contention free communication schedules for many-to-many communication with arbitrary sized data. Towards this end, we propose three approximation algorithms - PST, LDT and SDT. From time to time, the CSM first generates a bipartite graph from the set of received requests, then determines which of these three algorithms gives the best approximation factor on this graph and finally executes that algorithm to generate a contention free schedule. Algorithm PST has a worst case run time of O( max (Δ|E|, |E|log(|E|))) and guarantees an approximation factor of 2H 2Δ-1, where |E| is the number of edges in the bipartite graph, Δ is the maximum node degree of the bipartite graph and H 2Δ-1 is the (2Δ-1)-th harmonic number. LDT runs in O(|E|2) and has an approximation factor of 2(1+τ), where τ is a constant defined as a guard band or pause time to eliminate the possibility of contention (in an apparently contention free schedule) caused by system jitter and synchronization inaccuracies between the nodes. SDT gives an approximation factor of 4log(w max ) and has a worst case run time of O(Δ|E|log(w max )), where w max represents the longest communication time in a set of received requests.

KW - approximation algorithms

KW - cluster computing

KW - contention free schedule

KW - data centers

KW - Many-to-many scheduling

KW - switch fabric

KW - switch scheduling

UR - http://www.scopus.com/inward/record.url?scp=79951638887&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79951638887&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-19056-8_10

DO - 10.1007/978-3-642-19056-8_10

M3 - Conference contribution

SN - 9783642190551

VL - 6536 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 150

EP - 161

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

ER -