A Kernel Mechanism for Efficiently Supporting Quality of Service in Web Cluster System

• Main Authors: Chun-Hung Wu, 吳俊泓
• Other Authors: Mei-Ling Chiang
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/11219267940914069613

碩士 === 國立暨南國際大學 === 資訊管理學系 === 96 === Due to the explosive growth of the Internet and increasing service demands from all around the world, the cluster-based system that consists of one request-dispatching server and several request-handling servers has become a cost-effective way to serve the huge amount of service demands. Nowadays, Web-based services have occupied a great proportion of the Internet services. In order to support end-to-end Quality of Service in Web clusters, it’s essential to design a Web cluster that can provide differentiated services to various requests and treat important requests such as billing and shipping requests with preferred order to avoid losing of business revenue. Besides, Web servers also have to handle more complex types of requests since requests from clients may be mixed with dynamic Web pages, database processing, or multimedia stream data. Therefore, a Web cluster should be enhanced with some mechanisms to provide end-to-end QoS and designed with intelligent request dispatching policies for supporting various types of service requests. In this thesis, we have designed and implemented a kernel-level Web-based Quality of Service (WQoS) mechanism that could efficiently support differentiated services when serving various types of Web requests in our LVS-CAD Web cluster. We have also proposed two new content-aware request distribution policies named Locality-Aware Request Distribution with Replication and Classification (LARD/RC) and Grouped Client-Aware Policy (GCAP) to dispatch requests efficiently in Web clusters providing multiple types of services and running in homogeneous or heterogeneous environments. Experimental results demonstrate that the LVS-CAD Web cluster with our proposed Web-based QoS mechanism can ensure all requests with high priority to conform to Service Level Agreement (SLA) by dropping acceptable percentage of requests with minor importance during the situation of system overloaded. Besides, our proposed LARD/RC and GCAP policies can perform 110.43% and 83.54% better than the well-known content-aware LARD/R policy and CAP policy respectively in a homogeneous environment and outperform the LARD/R policy and CAP policy by 123.27% and 47.34% respectively in a heterogeneous environment.