Font Size: a A A

QoS Mechanisms On A Pipeline Web Server-KETA

Posted on:2006-06-01Degree:MasterType:Thesis
Country:ChinaCandidate:J Y LiuFull Text:PDF
GTID:2178360185963440Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With phenomenal growth of web applications on Internet, web fluxes take more than 60% of the total Internet fluxes. Exponential growth of HTTP requests makes plenty of web site facing problems of enhancing service ability and what to do when under overload. In additional, as the rise of electronic commerce, the service mod of Internet is under changing from traditional data communication and information viewing to electronic trade and service. Different from the traditional TCP/IP and HTTP to offer equal service, electronic commerce requires classification of the customers and requests and to supply differentiated services according to the profit. All of these make QoS become an important problem to be studied.A novel pipeline architecture multi-thread web server—KETA is designed to improve the web service ability of Kylin. KETA divides the total course to serve a request into four segments, and establishes four groups of threads to implement functions of four segments. Results of experiments show that KETA works well on improving threads availability to enhance service ability of Kylin system.This paper studies architecture of KETA and consults results of its test running to find out some lacks of KETA. Technologies of Web QoS methods are studied and brought forward to enhance stability and quality of KETA's services.Requests are classified and scheduled according to requesting files and information of network QoS. This can combine server QoS and network QoS to supply differentiated service. Requests are served by the theory of content-based adaptation when server is under overload. Server can serve more requests than just refusing requests when overloaded. Overload varies between segments of KETA's pipeline architecture. This paper combines feed-forward theory and feed-forward theory to schedule thread number of KETA's segments dynamically to get a better use of server's resource. Resources taken by KETA can also be limited according to the usage of web server. This makes server work stably.
Keywords/Search Tags:Web QoS, KETA, pipeline architecture, content-based adaptation, thread scheduling
PDF Full Text Request
Related items