New Architecture and Algorithm for Webserver Cluster Based on Linux Virtual Server
New Architecture and Algorithm for Webserver Cluster Based on Linux Virtual Server
复制标题
DOI：
10.1109/isip.2008.143
复制
发表时间：
2008-05
期刊：
影响因子：
--
通讯作者：
Yanping Gao;Xiangjun Li;Yan-gang Che
中科院分区：
文献类型：
--
作者：
Yanping Gao;Xiangjun Li;Yan-gang Che
随着互联网的不断发展，许多网站都采用了集群技术来扩大其性能。目前，LVS （Linux Virtual Server）软件已被广泛用于构建web服务器集群，它是一种免费且性能良好的软件。但是，当请求增加时，LVS可能会引发瓶颈，使集群系统无法运行，因为它只有一个前端。在本文中，我们提出了一种基于LVS的web服务器集群的新架构，该架构具有多个前端可以同时作为后端。我们还提出了一种调度算法，在考虑服务器负载的情况下，将请求平均分配给服务器。本文提出的系统和算法消除了这一瓶颈，对中小型网站的建设具有实用价值。在本文提出的系统中，只有当服务器的负载不够大时，服务器才能直接响应客户端的请求。否则，服务器应该将请求重定向到负载最低的选定后端。我们希望具有多个前端的web服务器集群能够在成本和性能上适合和高效地构建中小型网站。
According to the growing of Internet, many websites have used clustering technologies to enlarge their performance. These days, LVS (Linux Virtual Server) software, which is free of charge and has good performance, has commonly been used to construct web server cluster. But when requests are increased, LVS can raise a bottleneck and can make the cluster system unable after all, because it has only single front-end. In this paper, we suggest a new architecture for webserver cluster based on LVS with multiple front-ends which can act as back-ends simultaneously. We also propose a scheduling algorithm to distribute requests equally to servers with considering their load. The proposed system and algorithm remove the bottleneck, and are useful in constructing small and middle-sized websites. In the proposed system, a server can make response directly to clientpsilas request only when its load is not large enough. Otherwise, the server should redirect the request to a selected back-end with the lowest load. We hope that the webserver cluster with multiple front-end will be suitable and efficient for constructing small and middle-sized websites at the point of cost and performance.