Multipath Load-Balancing Routing Mechanism in Data Center Network

Nan Ya,Xingwei Wang,Min Huang
DOI: https://doi.org/10.1109/compcomm.2017.8322534
2017-01-01
Abstract:Data center is usually the center of data transmission, computation and storage. The traditional Data Center Networking (DCN) adopts hierarchical structure, so the bandwidth of the root node becomes the bottleneck of the system performance. To adapt to the new requirements of DCN, a variety of new network architectures have been proposed. These new structures have multiple paths in nature, and most topologies are fixed and well-regulated. Routing mechanisms can leverage multipath to provide higher network bandwidth and better reliability guarantee. However, most existing multipath routing mechanisms cannot be optimized for general use, because they are designed for specific network structures. Based on this, this paper proposes an Improved Ant Colony Optimization (IACO) algorithm, which uses the idea of data flow segmentation to improve the network throughput and resource utilization. The algorithm has the characteristics of distributed, and each ant can interact with each other quickly without knowing a complete network topology and find a number of better paths, which can save the cost and adapt to the rapid change of traffic in the DCN.
What problem does this paper attempt to address?