一种新的实现区分服务的可扩展缓存管理算法
A Novel Scalable Buffer Management Algorithm for DiffServ
-
摘要: 目前Internet网络中采用的缓存管理策略大多为丢尾(Drop tail)算法,并不能适应区分服务模型的要求;而RED及其改进算法CHOKe也不能提供公平性和相对优先级的保证。该文简要地分析了目前常见的缓存管理算法RED以及CHOKe所存在的不足,提出了一种基于区分服务模型的新型缓存管理算法D-CHOKe。该算法提供了公平性和相对优先级的保证,并且保持下现有IP网的可扩展性。仿真试验表明,该算法能够较好地满足区分服务的要求,为不同优先级的流分配不同的带宽。Abstract: Current Internet uses typically drop tail as its buffer management scheme. This scheme does not meet the need of DiffServ model, while RED and its improved algorithms also suffer from shortcoming in term of fairness and relative drop precedence. This paper analyzes several popular packet scheduling algorithms, proposes a novel scalable algorithm named D-CHOKe(Difference-CHOKe) based on DiffServ model. This algorithm achieves fairness and relative drop precedence, and keep the scalability of the IP network. Simulation demonstrates that D-CHOKe algorithm well meets the requirements of DiffServ.
计量
- 文章访问数: 2297
- HTML全文浏览量: 122
- PDF下载量: 610
- 被引次数: 0