中文提要:

在網路的應用上，目前在骨幹流量中佔最大宗的莫過於WWW (全球資訊網) 了。WWW速度慢的原因不外乎網站負擔沈重導致反應過慢與網路擁擠使得資料傳回花上太多時間。由於網頁並非每一回傳送的內容都不一樣，所以可以引入快取的觀念：在靠近使用者的位置加上快取伺服器，以便同時節省網站與網路兩者的負擔，減少使用者等待的時間。

本篇論文的主題在於有效發揮網路快取系統的效能。從目前網路快取伺服器的使用情況來看，在分工合作上缺乏有效率的設計，造成管理者偏好設定各自為政；即使合作建立了多層的架構，最頂層也容易成為瓶頸，拖慢整個系統的速度。我們應用集合關係，修改現有的協定，設計了一個與目前相容、以遞迴方式建立有效率的階層式架構，避免不必要的資料查詢；同時我們讓架構中底層直接到目的地索取資料，再將資料向上層回傳，藉此把資料流動的方向倒反過來，進一步減少最頂層的負擔。我們將整個過程的步驟逐項分析，對細部流程統計出結果，推導出整個過程的一般式。在分析了這個方法的特性後，我們發現這個架構主要是將最頂層抓資料的責任分散到底層，把整個系統工作量的分布往下壓。從演算法的分析中得知，在極端的假設條件之下，最頂層可以減少一半的工作量，雖然有些工作要轉嫁到系統底層。但是就系統的負載平衡角度而言，這樣的結果可以減少系統中瓶頸的發生，提高整體WWW傳送的效率。

For the past few years, WWW (World Wide Web) traffic has been tremendously growing on the Internet. However, it ironically becomes “World Wide Wait” due to overloaded server and/or seriously congested network. Almost any computer system that suffers from latency or bandwidth problems can benefit from caching. The introduction of cache concept to WWW server certainly reduce the waiting time at clients by efficiently relieving both the server and network load.

The purpose of this thesis is to design a hierarchical cache system so that it can work efficiently. The cache servers being used today have encountered the problems of lack of efficient collaboration due to different configuration flavors of management. In other words, the highest level of a hierarchical cache system may easily become the bottleneck, which in turn slow down the entire cache system. In our cache design, we apply inclusive and exclusive relationship to modify the ICP (Internet Cache Protocol) and use the recursive concept to build a hierarchical architecture to avoid the un-necessary information query. This implies that the lowest level can fetch the data from the destination server directly and perform the update recursively to its upper level. The up-side-down traffic flow, reffered to as reverse traffic flow in this thesis, can substantially release the load of upper levels in the hierarchy.

For the purpose of performance evaluation, we derive a general mathematical equation by analyzing the operation procedures step by step. The analytical results have shown that the highest level within this hierarchy can reduce almost 50% of the load under the worst-case assumption. Although the lower levels may slightly increase their work load, it does significantly increase the overall WWW efficiency and avoid the potential bottlenecks by balancing the loads among different levels.

摘要

緒論
1.1 研究動機
1.2 論文概要
1.3 論文組織

網路快取相關研究背景
2.1 頻寬使用研究的歷史
2.2 網路快取原理
2.3 Replacement Policy
2.4 負載平衡設計
2.5 階層架構原理與設計
2.6 常見支援階層架構的通訊協定

高效率與階層式架構的網路快取設計
3.1 建立高效率階層式架構
3.2 電腦元件效能趨勢分析
3.3 Reverse Traffic
3.4 建立流程與演算法

演算法分析與計算
4.1 物件搜尋時間
4.2 物件傳送時間
4.3 Reverse Traffic對頻寬使用的衝擊
4.4 快取空間對系統的影響
4.5 Sibling對系統的影響
4.6 命中率對系統的影響

結論與展望

參考文獻

附錄
A. 支援 Reverse Traffic 且與現有系統相容的通訊協定----ICP 的修改
B. Cache Digest 的原理
C. 動態選擇反應較快的 Sibling

索引

[1] K. hompson, G. J. Miller, and R. Wilder, "Wide Area Internet Traffic Patterns and Characteristics", IEEE Network, Vol 11, no 6, Nov./Dec. 1997, pp. 10-23.

[2] Chiang, C.-Y.; Liu, M.T.; Muller, M.E, "Caching neighborhood protocol: A foundation for building dynamic Web caching hierarchies with proxy servers", Parallel Processing, 1999. Proceedings.

[3] R. Fielding, J. Gettys, J. Mogul, H. Frystyk, L. Masinter, P. Leach, T. Berners-Lee, "Hypertext Transfer Protocol -- HTTP/1.1", RFC2616

[4] Web Cache Communication Protocol v2, http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/120newft/120t/120t3/wccp.htm

[5] Paul Vixie, Duane Wessels, "Hyper Text Caching Protocol (HTCP/0.0)", INTERNET-DRAFT <draft-vixie-htcp-proto-05.txt>

[6] Network Appliance - Products - NetCache, http://www.netcache.com/

[7] A Distributed Testbed for National Information Provisioning, http://ircache.nlanr.net/Cache/

[8] Duane Wessels, K. C. Claffy, “A Distributed Architecture for Global WWW Cache Integration”, http://www.ircache.net/Cache/Proposal.96/

[9] Hans-Werner Braun and K. Claffy. Web traffic characterization: an assessment of the impact of cach ing documents from NCSA's web server. In Second international world wide web conference, October 1994.

[10] C. Mic Bowman, Peter B. Danzig, Darren R. Hardy, Udi Manber, and Michael F. Schwartz. The harvest information discovery and access system. Proceedings of the Second International World Wide Web Conference, pages 763--771, October 1994.

[11] Peter B. Danzig, Richard S. Hall, and Michael F. Schwartz. A case for caching file objects inside internetworks. In Proceedings of ACM SIGCOMM '93, pages 239--248, September 1993.

[12] Andrei Z. Broder, Li Fan, Pei Cao and Jussara Almeida, "Summary Cache: A Scalable Wide-Area Web Cache Sharing Protocol", Proceedings of the ACM SIGCOMM '98 conference on Applications, technologies, architectures, and protocols for computer communication, pages 254--265.

[13] D. Wessels, K. Claffy, "Internet Cache Protocol (ICP), version 2", RFC 2186

[14] D. Wessels, K. Claffy, "Application of Internet Cache Protocol (ICP), version 2", RFC 2187

[15] R. Rivest, "The MD5 Message-Digest Algorithm", RFC 1321

[16] Michel, S.; Nguyen, K.; Rosenstein, A.; Zhang, L.; Floyd, S.; Jacobson, V., "Adaptive web caching: towards a new global caching architecture", Computer Networks And ISDN SystemsVol: 30, Issue: 22-23, pp. 2079-2282, November 25, 1998

[17] Adaptive Web Caching Project, http://irl.cs.ucla.edu/AWC/

[18] B. Duska, D. Marwood, and M. Feeley, "The Measured Access Characteristics of World-Wide-Web Proxy Caches", Proc. USENIX Symp. Internet Tech. and Sys., Monterey, CA, Dec., 1997, pp. 23-35

[19] Mahanti, C. Williamson, and D. Eager, "Traffic Analysis of a Web Proxy Caching Hierarchy", IEEE Network, Vol 14, no 3, May/June 2000, pp. 16-23.

[20] Burton Bloom, "Space/time trade-offs in hash coding with allowable errors" Communications of ACM, pages 13(7):422-426, July 1970.

[21] Proxy Client Autoconfig File Format, http://home.netscape.com/eng/mozilla/2.0/relnotes/demo/proxy-live.html

[22] Cache Array Routing Protocol (CARP) - Microsoft ISN, http://www.microsoft.com/ISN/whitepapers/cache_array_routing.asp
[23] Miyoshi, T., "Consideration about the cache server optimization and stable network access", Knowledge-Based Intelligent Electronic Systems, 1998. Proceedings KES '98. 1998 Second International Conference on Pages: 21 - 25 vol.2

[24] Markatos, E.P., “Main memory caching of Web documents, Computer Networks And ISDN Systems” , Vol: 28, Issue: 11, pp. 893-1580, May 1996 , pp. 893-905

[25] Nishikawa, N.; Hosokawa, T.; Mori, Y.; Yoshida, K.; Tsuji, H., “Memory-based architecture for distributed WWW caching proxy”, Computer Networks And ISDN Systems , Vol: 30, Issue: 1-7, pp. 1-776, April 1998 , pp. 205-214

[26] Web Cache Communications Protocol v2, http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/120newft/120t/120t3/wccp.htm

[27] Ilhwan Kim, Yeom H.Y., Joonwon Lee, "Analysis of buffer replacement policies for WWW proxy", Information Networking, 1998. (ICOIN-12).

[28] Ying Shi, Watson E., Ye-sho Chen, "Model-driven Simulation Of World-wide-web Cache Policies", Simulation Conference, 1997., Proceedings of the 1997 Winter.

[29] Bolot J.-C., Hoschka, P., "Performance engineering of the World Wide Web: Application to dimensioning and cache design", Computer Networks And ISDN Systems , Vol: 28, Issue: 11, pp. 893-1580, May 1996 , pp. 1397-1405.

[30] Krishnamurthy, B.; Wills, C.E., "Proxy cache coherency and replacement-towards a more complete picture", Distributed Computing Systems, 1999. Proceedings. 19th IEEE International Conference on Pages: 332 - 339

[31] Belloum, A.; Hertzberger, L.O., "Dealing with one-timer-documents in Web caching", Euromicro Conference, 1998. Proceedings. 24th on Pages: 544 - 550 vol.2

[32] Belloum, A.; Hertzberger, L.O., "Document replacement policies dedicated to Web caching", Intelligent Control (ISIC), 1998. Proceedings of the 1998 IEEE International Symposium on Pages: 576 - 581
[33] Aggarwal, C.; Wolf, J.L.; Yu, P.S., "Caching on the World Wide Web", Knowledge and Data Engineering, IEEE Transactions on Pages: 94 - 107

[34] Scheuermann, P.; Shim, J.; Vingralek, R., "A case for delay-conscious caching of Web documents", Computer Networks And ISDN SystemsVol: 29, Issue: 8-13, pp. 865-1554, September 1997 , pp. 997-1005

[35] Zhang, J.; Izmailov, R.; Reininger, D.; Ott, M., "Web caching framework: analytical models and beyond", Internet Applications, 1999. IEEE Workshop on Pages: 132 - 141

[36] 國立中山大學代理伺服器首頁, http://proxy.nsysu.edu.tw/

[37] 交大 Proxy 服務, http://proxy.nctu.edu.tw/

[38] 教育部 Proxy 伺服器統計分析, http://cache.moe.edu.tw/

[39] NTU Proxy Caching Service, http://proxy.ntu.edu.tw/

[40] Squid Web Proxy Cache, http://www.squid-cache.org/

[41] Martin Hamilton, Alex Rousskov, Duane Wessels, “Cache Digest specification - version 5”, http://www.squid-cache.org/CacheDigest/cache-digest-v5.txt

[42] Foundry Networks - Products - ServerIron Internet Traffic Management Switch Datasheet, http://www.foundrynetworks.com/datasheets/serverironspec.html

[43] CacheMgr@proxy.ntu.edu.tw: menu, http://proxy.ntu.edu.tw/cgi-bin/cachemgr.cgi?host=proxy.ntu.edu.tw&port=3128&user_name=&operation=menu&auth=

[44] TANet 骨幹流量統計, http://www.edu.tw/tanet/backbone/

[45] R. Tewari et al., "Design Considerations for Distributed Caching on the Internet", Proc. 19th IEEE int'l. Conf Dist. Comp. Sys., Austin, TX, May 1999, pp. 273-84