摘要
大数据时代数据传输需求呈爆炸式增长,如何充分利用数据流的新特征,发挥不同交换方式的优势,是解决未来大数据传输的必由之路。为此,介绍了大数据传输在容量和能耗方面的挑战,并结合数据流的新特征,分析网络技术发展的新机遇。然后,回顾了近几年相关领域的研究进展,并简要介绍了笔者在混合交换和存储转发等方面正在开展的研究工作。
Explosive growth of traffic demand both inside and between data centers has resulted in a new wave of cutting edge research in the networking community. Inter-and Intra-data center research problems were identified and various solutions aiming to reduce networking cost or increase energy efficiency have been investigated. The growth in traffic demand in the big data era was accompanied by distinct characteristics in flow size distribution and delay tolerance. Data volume was dominated by large flows that are very few in number, and such flows are typically delay tolerant. How these characteristics may be used to tackle the challenge was discussed, followed by our recent work in big data transmission. The BLOC(blocking loss curve) framework aims to provide a framework with which hybrid switching systems combining packet and circuit switching can be studied. The time-shifted multi-layer graph is a generic tool that can be used to analyze the performance of circuit switched networks with bulk storage.
出处
《大数据》
2015年第2期56-65,共10页
Big Data Research
关键词
大数据网络
电路/分组混合交换
存储转发
数据中心
big data networking,hybrid circuit/packet switching,store and forward,data center
