The Communication Performance of BCDC Data Center Network

The BCDC network is a new server-centric data center network based on crossed cube. Its decentralized and recursively defined features solve the bandwidth bottleneck problem of the upper switch in the tree structure and make it more scalable. In addition, the degree of BCDC server is constant, which reduces connection cost and technical difficulty relative to DCell and BCube. In this paper, we study and analyze the data communication, fault tolerance, and node-disjoint paths of BCDC through practical experiments. The results show that BCDC is superior to DCell and Fat-Tree in data communication. Moreover, in terms of fault-tolerant routing and node-disjoint paths, the performance of BCDC is no worse than that of DCell and FatTree. We also propose a one-to-one communication algorithm in BCDC under 1-restricted connectivity, analyze time complexity of algorithm, and give an upper bound of the conditional faulty diameter of BCDC under 1-restricted connectivity. The work in this paper provides an important basis for the design and application of the new data center network.

[1]  Lei Shi,et al.  Dcell: a scalable and fault-tolerant network structure for data centers , 2008, SIGCOMM '08.

[2]  Eytan Modiano,et al.  A Hierarchical WDM-Based Scalable Data Center Network Architecture , 2019, ICC 2019 - 2019 IEEE International Conference on Communications (ICC).

[3]  Haitao Wu,et al.  BCube: a high performance, server-centric network architecture for modular data centers , 2009, SIGCOMM '09.

[4]  Amin Vahdat,et al.  A scalable, commodity data center network architecture , 2008, SIGCOMM '08.

[5]  Jia Guo,et al.  The g-good-neighbor conditional diagnosability of the crossed cubes under the PMC and MM* model , 2019, Theor. Comput. Sci..

[6]  Lih-Hsing Hsu,et al.  Edge Congestion and Topological Properties of Crossed Cubes , 2000, IEEE Trans. Parallel Distributed Syst..

[7]  Kemal Efe.  A Variation on the Hypercube with Lower Diameter , 1991, IEEE Trans. Computers.

[8]  Huaxi Gu,et al.  Mesh-of-Torus: a new topology for server-centric data center networks , 2018, The Journal of Supercomputing.

[9]  Mengjie Lv,et al.  The Conditional Reliability Evaluation of Data Center Network BCDC , 2020, Comput. J..

[10]  Charles E. Leiserson,et al.  Fat-trees: Universal networks for hardware-efficient supercomputing , 1985, IEEE Transactions on Computers.

[11]  Mengjie Lv,et al.  Reliability Evaluation of Data Center Network DCell , 2018, Parallel Process. Lett..

[12]  Cheng-Kuan Lin,et al.  Vertex-disjoint paths in DCell networks , 2016, J. Parallel Distributed Comput..

[13]  Farhad Faghani,et al.  Crystal: A scalable and fault-tolerant Archimedean-based server-centric cloud data center network architecture , 2019, Comput. Commun..