多層級網際網路應用中服務不中斷機制之效能評估

Performance Study for the Service Continuity for Multi-Tier Client-Server Internet Applications

林威言(Wei-Yen Lin)

指導教授 ： 林風

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.10447

摘要

隨著網際網路的快速發展，現代社會大量使用網際網路應用來存取資訊以及進行電子商務。然而大部分的網際網路應用提供者都會因為服務的中斷而影響其服務品質與客戶滿意度。因此，服務不中斷 (Service Continuity) 技術被提出用以維持服務品質與使用者滿意度。目前現存的服務不中斷機制往往需要複雜的實作架構和有著大量的效能負擔。在本篇論文中，我們提出一個服務不中斷機制 (Trap and Recover Mechanism)，此服務不中斷機制在不更動網際網路應用的系統架構下加入兩種代理元件，攔截代理元件以及復原代理元件。此兩種代理元件使用了訊息錄誌以及檢查點兩種功能來實現服務不中斷。我們提出數學分析模型以及電腦模擬來驗證服務不中斷機制的效能。研究顯示此服務不中斷機制可以較小的效能負擔達成服務不中斷。

關鍵字

服務不中斷 ； 效能評估

並列摘要

With the fast development of the Internet, modern society increasingly depends on Internet applications to retrieve information and conduct business. Most of Internet application providers suffer the revenue loss and customer dissatisfaction due to service interruption of Internet applications. While considering the QoS (Quality of Service) for Internet applications, the Service Continuity (i.e., the uninterrupted delivery of a service from an end user's perspective) is proposed to increase the satisfaction to users. Although several Service Continuity mechanisms have been proposed, these existing mechanisms are usually too complicated to implement or have tremendous overhead. In this paper, we propose a Service Continuity mechanism, namely the Trap and Recover Mechanism (TRM). In TRM, two types of online lightweight agents are deployed in the Internet application servers without modifying the underlying architecture, namely, the Trap Agent and the Recover Agent. These agents adopt the message logging and the checkpointing to achieve the Service Continuity. We derive the analytic models and conduct simulation experiments to investigate the performance of the proposed mechanism. Our study shows that TRM can achieve the Service Continuity with lower overhead.

並列關鍵字

Service Continuity ； Performance evaluation

參考文獻

[1] Alvisi, L. and Marzullo, K. Message Logging: Pessimistic, Optimistic, Causal, and Optimal. IEEE Transactions on Software Engineering, 24(2):149–159, Febuary 1998.

[2] Bellavista, P., Corradi, A., and Giannelli, C. Mobility Prediction for Mobile Agentbased Service Continuity in the Wireless Internet. In First International Workshop on Mobility Aware Technologies and Applications, pages 1–12, October 2004.

[7] Elnozahy, E. N., Alvisi, L.,Wang, Y.-M., and Johnson, D. B. A Survey of Rollbackrecovery Protocols in Message-passing Systems. ACM Computing Surveys (CSUR), 34(3):375–408, September 2002.

[8] Elnozahy, E. N. and Zwaenepoel, W. On the Use and Implementation of Message Logging. In 24th International Symposium on Fault-Tolerant Computing, pages 298–307, June 1994.

[9] Jalote, P. Fault Tolerance in Distributed Systems. Prentice Hall, 1998.