Automatically configuring the network layer of data centers for cloud computing

  • Authors:
  • C. Hu;M. Yang;K. Zheng;K. Chen;X. Zhang;B. Liu;X. Guan

  • Affiliations:
  • MOE Key Lab for Intelligent Networks and Network Security, Department of Computer Science and Technology, School of Electronic and Information Engineering, Xi'an Jiaotong University, Xi'An, China ...;Department of Computer Science and Technology, Tsinghua University, Beijing, China;IBM Research Division, China Research Laboratory, Beijing, China;Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, IL;Computer Science Department, Carnegie Mellon University, Pittsburgh, PA;Department of Computer Science and Technology, Tsinghua University, Beijing, China;MOE Key Lab for Intelligent Networks and Network Security, School of Electronic and Information Engineering, Xi'an Jiaotong University, Xi'an, China

  • Venue:
  • IBM Journal of Research and Development
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

With the requirement of very large data centers for cloud computing, the challenge lies in how to produce a scalable and inexpensive design to interconnect a large number of servers in data centers while providing fault tolerance and high network capacity. Internet Protocol (IP)-based data center networks (DCNs) can support a sufficiently large number of machines and dynamic multi-path routing. However, the complex and error-prone manual configuration of IP-based DCNs hinders its deployment. In this paper, we propose DCZeroconf, a fully automatic IP address configuration mechanism, to eliminate the burden of manual configurations of the IP addresses of servers and switches in data centers. We have implemented a prototype system to validate the effectiveness of DCZeroconf via extensive experiments and simulations. The evaluation results demonstrate that DCZeroconf supports different topologies, and the assigned IP addresses can be automatically adjusted upon dynamic topology changes. In addition, the entire automatic process can be completed in seconds or less.