Adaptive routing for layer-2 load balancing in data center networks
被引:0
作者:
Kanagavelu, Renuga
论文数: 0引用数: 0
h-index: 0
机构:
Data Storage Institute, SingaporeData Storage Institute, Singapore
Kanagavelu, Renuga
[1
]
论文数: 引用数:
h-index:
机构:
Francis, Bu-Sung Lee
[2
]
Ragavendran, Vasanth
论文数: 0引用数: 0
h-index: 0
机构:
National university of Singapore, SingaporeData Storage Institute, Singapore
Ragavendran, Vasanth
[3
]
Aung, Khin Mi Mi
论文数: 0引用数: 0
h-index: 0
机构:
Data Storage Institute, SingaporeData Storage Institute, Singapore
Aung, Khin Mi Mi
[4
]
机构:
[1] Data Storage Institute, Singapore
[2] Nanyang Technological University, Singapore
[3] National university of Singapore, Singapore
[4] Data Storage Institute, Singapore
来源:
Advances in Information Sciences and Service Sciences
|
2012年
/
4卷
/
15期
关键词:
Ethernet;
-;
Trees;
(mathematics);
D O I:
10.4156/AISS.vol4.issue15.1
中图分类号:
学科分类号:
摘要:
With the Internet boom over the last decade, large scale Data Centers are rapidly increasing in capacity and numbers to meet the ever increasing demand for bandwidth. There is a trend of deploying a large number of commodity layer-2 Ethernet switches in Data Centers. However, the existing Spanning Tree Protocol (STP) used in the traditional Ethernet networks becomes inefficient for Data Centers as it underutilizes the resources due to the lack of multipath capability leading to poor bandwidth utilization. In this paper we develop a layer-2 load balancing framework with a multi-path forwarding mechanism which balances the load across the network, thereby reducing the occurrence of congestion that leads to efficient utilization of the bi-section bandwidth in Data Centers. The proposed framework has several important features: 1) It provides adaptive multi path forwarding compatibility in layer-2 Ethernet switched networks to efficiently spread the load in adaption to changing load conditions. 2) It improves the bi-section bandwidth utilization in Data Center networks. 3) It is capable of achieving load balancing at the layer-2 level and 4) It ensures effective traffic redistribution upon link failures. We demonstrate the effectiveness of the proposed mechanism through simulation results.