透過您的圖書館登入
IP:18.118.200.136
  • 學位論文

大量連線之分散式負載平衡架構

A Distributed Load Balancing Architecture for the Massive Connections

指導教授 : 湯政仁
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


隨者網際網路的普及化,很多的網路伺服器例如:線上遊戲、線上交易系統、大型網頁網站等都會面臨了有大量連線請求的問題。然而只使用一台伺服器是無法單獨提供這類的網路服務。分散式伺服器架構,提供這類網路服務能處理大量連線請求的能力。本篇論文採用weighted least connection演算法為負載平衡基礎建構一個分散式負載平衡架構,分散式架構把不同等級的電腦聚集起來,使之能應付短時間大量的連線服務。 而在熱門的網路服務,常常出現在短時間內有更大量的連線請求,當伺服器管理員發現這件事時,他會再加設新的伺服器來分攤整體網路架構的負載,因為分散式系統的負載平衡演算法是以weighted least connection演算法為基礎,在新伺服器的負載在所有伺服器裡是最小的這段時間裡,所有連線請求都會指派給這個新伺服器,假設短時間內連線請求的數量大於此新伺服器短時間所能服務的數量時,這就會照成某些連線請求失敗,為了面對這個問題,本篇論文修正了weighted least connection演算法。 而實驗的結果本篇論文的分散式負載平衡架構不僅具備處理大量連線的能力,也依照個別伺服器的能力分配負載,更能當有一個新的伺服器加入系統後,不會有因weighted least connection演算法的特性而造成的連線失敗。

並列摘要


With the popularization of the internet network, some network servers, such as the online game servers, the online trade system, and the website servers, often have to sustain the massive connection requests. However, a single server is not able to handle this kind of network service. The distributed server system provides this kind of network service, which can sustain this condition. This thesis adopts the weighted least connection algorithm to build a distributed load balancing architecture. The distributed server system collects the heterogeneous servers. Letting these heterogeneous servers are able to service the massive connection requests in a short period of time. In some popular network services, it is often occurring more massive connection requests in a short period of time. When the manager of the distributed server system discovers that this condition is happened, he will add a new server to share the other servers’ load. Because the load balancing algorithm of the distributed system is based on the weighted least connection scheduling algorithm, the connection requests are all assigned to the new server at the time during the new sub-server’s load is smallest in all servers. If the number of connection requests is larger then the number of the new server can serve in a short period of time, it occurs that some connections fail. In order to face this problem, this thesis modifies the weighted least connection algorithm. The experimental results show that this thesis proposed architecture can not only have the capability to process massive connection requests but also successfully dispatch the load to individual server. When a new server joins into the distributed system, the distributed system doesn’t occurring the connection failure which caused by the characteristic of the weighted least connection algorithm.

參考文獻


[2] Berkeley, http://www.books.com.tw
[5] F.Prefect, L.Doan, S.Gold, Th.Wicki, W.Wilick, “Performance Limiting Factors in Http Server Operations,” In COMPCOn’96, IEEE, 1996.
[7] M. Colajanni, P.S. Yu and D.M. Dias, “Analysis of Task Assignment Policies in Scalable Distributed Web-Servers Systems,” IEEE Transactions on Parallel and Distributed Systems, Vol. (, N). 6, June 1998.
[8] V. Cardellini, E. Casalicchio, M. Colajanni, and P.S., Yu, “The State of the Art in Locally Distributed Web-server systems,” ACM computing Surveys, vol. 34, no. 2, pp. 263-311, June 2002.
[10] E. Casalicchio and M. Colahanni, “A Client-Aware Dispatching Algorithm for Web Clusters Providing Multiple Services,” Proc. ACM World Wide Web Conf., pp. 535-544, May 2001.

延伸閱讀