Research On Performance Optimization Of WEB Cluster System Based On Ordered Sessionid Model

With the rapid development of information technology, computer network, and mobile phones, etc., human society has entered a rapid Internet era. However, with the increasingly expanding of the scale of China’s internet and the deepening of various Internet applications, the Web application providers are confronted with a greater challenge, which put forward a higher demand toward the availability and processing ability of the platform. Web cluster technology is an effective method to resolve how to provide high performance service. So how to improve the processing ability of the Web cluster has become an important research topic.At present, the research to improve the processing ability of the Web cluster focuses on promoting the load balancer algorithm efficiency by using various technologies. However, the paper conducts the research from the aspect of session management.In view of the current Web cluster service environment, the Session database is extensively applied to store the important information of Session users. Sessionid is the only identification to distinguish different Session. In order for safety, Sessionid is generated by using patterns randomly while the Session reading by Web cluster depends on the location of Sessionid, which affects the query speed resulting from the disorder and the response ability of the Web cluster. Based on this, a new formation mechanism is proposed in the paper, which makes disorderly generation of Sessionid become a mode of order plus disorder, namely the first orderly 9 characters plus another 13 disorderly characters so as to improve the query speed by means of the orderly indexing and disorder to ensure safety. The mode, which is realized by the key method of rewriting, has an advantage in reducing indexing time with the in-depth research on the Session class in ASP.NET kernel.The massive query of Session database has low efficiency because of the disorderly storage of Sessionid while using the Web cluster. It is put forward that the Session data fast query based on the use of two fast search algorithms can be fulfilled by the user-defined storage procedure combining database cursor technology and the orderly storage in the Sessionid database, which accordingly improve the processing ability of the Web cluster. Lastly, experiments demonstrate that compared with the ordinary Web cluster in the high concurrent access, the binary research query algorithm achieves better results.