首页> 外文期刊>International Journal of Information and Communication Technology Research >Favoring Short Flows Without Punishing Long Flows for the Internet Traffic Performance Improvement
【24h】

Favoring Short Flows Without Punishing Long Flows for the Internet Traffic Performance Improvement

机译:在不影响长流量的情况下支持短流量,以提高Internet流量性能

获取原文
           

摘要

An explosive growth in business applications using the Internet have resulted in a strong demand for some notion of reliability or quality of service. During periods of congestion or failure, the quality of service of all flows is degraded. As a result, a strong need for service differentiation in the flows to provide service guarantees. Today the internet carries different types of traffic with the increased use of peer to peer, HTTP, FTP applications. Web traffic comprises short flows which requires low response time and long flows seen in the internet originate form peer to peer file sharing applications TCP is the dominating protocol that carries majority of the total internet traffic. Recent internet traffic measurement shows most of the TCP flows are short lived. The performance improvement in the internet traffic can be achieved by the advantages of scheduling algorithms to favor short TCP flows first. However long TCP flows competing against short TCP flows starve at some point. Hence, the proposed Dynamic Queue Scheduling (DQS) favors short TCP flows without penalizing the performance of long flows using Dynamic packet scheduling ratio. We observed that the mean transmission time of flows and packet loss significantly decreases in comparison with FIFO and RuN2C.
机译:使用Internet的业务应用程序的爆炸性增长导致对可靠性或服务质量的某些概念的强烈需求。在拥塞或故障期间,所有流的服务质量都会下降。结果,强烈需要在流中区分服务以提供服务保证。今天,随着对等,HTTP,FTP应用程序的使用越来越多,Internet承载了不同类型的流量。 Web流量包括短流,它需要很短的响应时间,而长流则是从Internet看到的,是从对等文件共享应用程序发起的,TCP是承载大部分总Internet流量的主导协议。最近的互联网流量测量显示,大多数TCP流都是短暂的。可以通过调度算法的优势实现互联网流量的性能提升,该算法首先支持短TCP流。但是,长TCP流与短TCP流竞争有时会出现饥饿。因此,提出的动态队列调度(DQS)支持短TCP流,而不会使用动态包调度比率来影响长流的性能。我们观察到,与FIFO和RuN2C相比,流的平均传输时间和数据包丢失显着减少。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号