【摘要】 随着互联网迅猛发展 ,研究Web请求的调度分配技术已成为国内外研究的热点。针对现有Web调度器存在着可扩展性差、系统开销大和无通用性诸多缺陷 ,设计和实现了一种新的、通用的、可扩展性好的基于内容的通用调度器Cuttle。该系统在内核TCP/IP层实现 ,采用伪服务器、捎带技术、截获、伪装三次握手、Mix LARD调度策略等技术。实验表明 ,Cuttle的可扩展性好 ,响应延迟小更多还原

【Abstract】 With the rapid development of the Internet, exploring the technology of scheduling Web requests has become the hot spot in the whole world. The present web schedulers have the deficiencies of poor scalability, high system overhead and non-generality. This paper introduces Cuttle—a new, general-purpose, and scalable content-aware scheduler implemented at the TCP/IP layer of the Linux kernel. Due to using the technologies of pseudo-server, piggy mechanism, interception and false 3-way handshaking, Mix-LARD scheduling strategy, we can see from our experiments that Cuttle has good scalability and low response latency.更多还原