摘要

Clustering multiple devices to form a single powerful device is a common method for improving performance. Most designs of the clustering schemes in the current literature are deploying a traffic splitter in front of devices in the cluster which acts as a centralized job dispatcher splitting workloads to backend devices. In this paper, we propose a decentralized clustering scheme, with no traffic splitter deployed, as an alternative solution on building a cluster system for those devices configured in transparent mode, such as bandwidth controllers, NIPSs, and traffic monitors. Devices in the cluster process the network traffic in parallel in a decentralized manner to scale the throughput. A device can also migrate its workload to others for the purpose of load balance or fault tolerance. Experiment results suggest that the proposed scheme can effectively improve performance of transparent mode devices in terms of throughput, load balance, and fault tolerance.

  • 出版日期2012-9

全文