RCD表达的网络缓存协作研究
|
摘要
网络缓存(Web Caching)把经常访问的网络对象在访问开销较少的网络节点上存一个副本，当用户再次访问这些对象时，可以直接从这个开销较少的网络节点上得到满足。从而在有限的带宽内，达到缩短用户等待时间、提高网络性能和网络可扩展性的目的，对于提高Internet访问效率具有重要意义。
网络缓存分单机缓存代理服务器缓存和协作缓存(Cooperative Caching)，三个层次。其中协作缓存是当前研究的热点。论文在对网络缓存技术的原理、现状和发展趋势综述的基础上，对RCD(Reduced Cache Digests)表达的网络缓存协作进行了研究，提出了精简的缓存摘要(RCD)网络缓存内容的表达方法。该方法在缓存摘要(CD)的基础上，将网络对象进一步分为最近访问的、最频繁访问的和很少访问的三类，那么其中最近访问的和最频繁访问的两类Web请求在将来最有可能再次发生。RCD只把前两类网络对象映射到缓存目录中，这样可减少缓存目录的存储空间，并且由于包含了网络对象的
太原理_仁大学硕士研究生学位论文
访问频率，所以特别适合于为具有相同访问特性的用户群提供
缓存服务。为此，论文讨论了在具有7层交换功能的交换机上，
建立、更新和访问RCD缓存目录的算法。最后在对同一局域网
上的不同用户组访问特性统计分析的基础上，对分布在本地交
换机上的RCD缓存协作进行了讨论。
理论分析和模拟实验表明，基于RCD的网络缓存协作方案，
克服了代理被所有的交换机共享和代理之间负载平衡的缺点，
并且充分的发挥7层交换机的内容感知功能，为本地用户提供
了一种快速有效的访问Internet的途径。
Web caching is a technique applied in Internet, with which the frequently accessed web objects are copied in a more easily accessed network node. Web caching can shorten clients' waiting time and improve Internet's performance without bandwidth increase. So it plays an important role in improving the efficiency of accessing Internet.
Web caching has three developing steps those are browser caching, proxy caching and cooperative caching. Among them, cooperative caching is the focus of current study. After recounting the theory, status quo and develop trend of web caching, this article puts out the RCD(Reduced Cache Digests) expression of the cached content and studies cooperative web caching using RCD expression. In RCD, the contents cached in proxy are classified into three groups: the recently accessed, the frequently accessed and the rarely accessed. Because the last group contributes nothing to the whole cache system, RCD only express the first two groups. Thus the dimension of the cached content's index is shortened. On the other hand, RCD especially adapts to servicing for client clusters with the same favoritism to web content, because it can store the accessed frequency of web objects. Based on RCD's frequency-storing characteristic, this article discusses the arithmetic of creating, renewing and accessing RCD index. Finally, cons
idering the users'
accessing-frequency, cooperative web caching using RCD index, which is distributed in local switches, is discussed.
Academic study and simulative experiment show that the problem tnat each proxy shared by all clients and balancing loading between proxies can't be achieved simultaneously is conquered by the cooperative caching plan based on RCD . In this plan, a L7 switch with RCD provides a fast and effective way for local user to access Internet, resulting from its content-awareness ability exerted fully by RCD.
网络缓存分单机缓存代理服务器缓存和协作缓存(Cooperative Caching)，三个层次。其中协作缓存是当前研究的热点。论文在对网络缓存技术的原理、现状和发展趋势综述的基础上，对RCD(Reduced Cache Digests)表达的网络缓存协作进行了研究，提出了精简的缓存摘要(RCD)网络缓存内容的表达方法。该方法在缓存摘要(CD)的基础上，将网络对象进一步分为最近访问的、最频繁访问的和很少访问的三类，那么其中最近访问的和最频繁访问的两类Web请求在将来最有可能再次发生。RCD只把前两类网络对象映射到缓存目录中，这样可减少缓存目录的存储空间，并且由于包含了网络对象的
太原理_仁大学硕士研究生学位论文
访问频率，所以特别适合于为具有相同访问特性的用户群提供
缓存服务。为此，论文讨论了在具有7层交换功能的交换机上，
建立、更新和访问RCD缓存目录的算法。最后在对同一局域网
上的不同用户组访问特性统计分析的基础上，对分布在本地交
换机上的RCD缓存协作进行了讨论。
理论分析和模拟实验表明，基于RCD的网络缓存协作方案，
克服了代理被所有的交换机共享和代理之间负载平衡的缺点，
并且充分的发挥7层交换机的内容感知功能，为本地用户提供
了一种快速有效的访问Internet的途径。
Web caching is a technique applied in Internet, with which the frequently accessed web objects are copied in a more easily accessed network node. Web caching can shorten clients' waiting time and improve Internet's performance without bandwidth increase. So it plays an important role in improving the efficiency of accessing Internet.
Web caching has three developing steps those are browser caching, proxy caching and cooperative caching. Among them, cooperative caching is the focus of current study. After recounting the theory, status quo and develop trend of web caching, this article puts out the RCD(Reduced Cache Digests) expression of the cached content and studies cooperative web caching using RCD expression. In RCD, the contents cached in proxy are classified into three groups: the recently accessed, the frequently accessed and the rarely accessed. Because the last group contributes nothing to the whole cache system, RCD only express the first two groups. Thus the dimension of the cached content's index is shortened. On the other hand, RCD especially adapts to servicing for client clusters with the same favoritism to web content, because it can store the accessed frequency of web objects. Based on RCD's frequency-storing characteristic, this article discusses the arithmetic of creating, renewing and accessing RCD index. Finally, cons
idering the users'
accessing-frequency, cooperative web caching using RCD index, which is distributed in local switches, is discussed.
Academic study and simulative experiment show that the problem tnat each proxy shared by all clients and balancing loading between proxies can't be achieved simultaneously is conquered by the cooperative caching plan based on RCD . In this plan, a L7 switch with RCD provides a fast and effective way for local user to access Internet, resulting from its content-awareness ability exerted fully by RCD.
引文
[1]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page ⅩⅩⅠ -ⅩⅩⅩ
[2]Jacobson, V. How to kill the Internet. In a presentation at SIGCOMM '95, Middleware Workshop.
[3]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 47-54
[4]Kristol, D., and Montulli, L.RFC2109: HTTP state management mechanism.(1997).
[5]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 122-133
[6]Another MD5 http://www. yicong.com/study/printpage.asp? ArticleID=52
[7]Bloom Filter http://www.cs.wisc.edu/～cao/papers/summary-cache/node8.html
[8]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 99-111
[9]The introduction of IR Cache http://www.ircache.net/
[10]曾新平 刘美华 韩真 等 预取技术研究进展 计算机科学．2003．030(008)．-28-30，55
[11]L.Breslau, Pei Cao, Li Fan, G.Phillips, and S.Shenker. Web caching and Zipf-like Distributions: Evidence and implications. In Proceeding of IEEE INFOCM, April 1999.
[12]Carlos Cunha, Azer Bestavros and M. Crovella. Characteristics of WWW client-based Traces. Technical Report BU-Cs-95-010, Computer Science Department, Boston .University, April 1999.
[13]刘美华 古志民等 web缓存系统浅析 北京理工大学学报．2003．023(001)．-78-82
[14]Shudong Jin and Azer Bestavros. Sources and Characteristics of web temporal locality.
[15]Marting Ariltt, Rich Friedrich and Tai Jin. Workload Characterization of a web proxy in A cable modem environment. Technical Report HPL-1999-48, Hewlett Packard Labs.
[16]Chankhunthod A, Danzig P B,Neerdaels C,et al. A Hierarchical Internet Object Cache.Usenix96,1996-012
[17]Rodriguez P, Spanner C, Biersack E W.Web Caching Architectures:Hierarchical and Distributed Caching.Proceedings of WCW'99,1993
[18]郭琳 李欣等 网络缓存技术计算机系统应用．2002．000(008)．-44-45
[19]Li B, Golin M J, Italiano G F, et aI.On the Optimal Placement of Web Proxies in the Internet.Proceedings of Infocom'99, 1996
[20]Aggarwal C,Wolf J L,Yu P S.Caching on the World Wide Web.IEEE Transactions on
Knowledge and Data Engineering, 1999,11 (1)
[21]Crovella ME, Carter RL. Dynamic server selection in the Internet. In: Proceedings of the 3rd IEEE Workshop on the Architecture and Implementation of High Performance Communication Subsystems (HPCS?5). Piscataway: IEEE Communication Society Press, 1995. 158～162.
[22]Myers A, Dinda P, Zhang H. Performance characteristics of mirror servers on the Internet. In: Proceedings of the IEEE INFOCOM?9. Piscataway: IEEE Press, 1999. 304～312.
[23]Cardellini V, Colajanni M, Yu KS. Dynamic load balancing on Web-server system. IEEE Internet Computing, 1999,3(3):28～39.
[24]Iyengar A, Challenger J, Dantzig P. High-Performance Web site design techniques. IEEE Internet Computing, 2000,4(2):17～26.
[25]Schroeder T, Goddard S, Ramamurthy B. Scalable Web server clustering technologies. IEEE Network, 2000,14(3):38～45.
[26]Andresen D, Yang T, Holmedahl V, Ibarra OH. SWEB: towards a scalable World Wide Web server on multicomputers. In: Proceedings of the IPPS?6. Los Alamitos: IEEE Computer Society Press, 1996. 850～856. http://computer.org/proceedings/ipps/7255/7255toc.htm.
[27]Baker SM, Moon B. Distributed cooperative Web servers. Computer Networks and ISDN Systems, 1999,31(11-16):1215～1229.
[28]Lin YW, Zhang DJ, Qian HL. A cooperative Web caching system based on concentrated management. Journal of Computer Research and Development, 2001,38(1):68～73 (in Chinese with English Abstract).
[29]Pai VS, Aron M, Banga G, et al. Locality-Aware request distribution in cluster-based network servers. In: Proceedings of the ASPLOS-Ⅷ. New York: ACM Press, 1998. 205～216. http://computer.org/proceedings/ipps/7255/7255toc.htm.
[30]Aron M, Sanders D, Druschel P, et al. Scalable content-aware request distribution in cluster-based network servers. In: Proceedings of the USENIX 2000 Annual Technical Conference. Berkeley, CA: USENIX Association, 2000. 323～336. http://www.usenix.org/events/usenix2000/program.pdf.
[31]Vinod Valloppillil and Keith W.Ross. Cache Array Routing Protocol, v 1.0. http://ircache.nlanr.net/cache/ICP/draft-vinod-carp-v1-02.txt, 1997.
[32]Li Fan, Pei Cao and Jussara Almeida. Summary cache : A scalable wide-area web cache Sharing protocol. Proceedings of SIGCOMM98,1998.
[33]Alex Rousskov and Duane Wessels. Cache Digest .http://ircache.nlanr.net
[34]Grimm C. And Neitzner M. Request Routing in cache meshes. Computer Networks and ISDN System, 1998.
[2]Jacobson, V. How to kill the Internet. In a presentation at SIGCOMM '95, Middleware Workshop.
[3]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 47-54
[4]Kristol, D., and Montulli, L.RFC2109: HTTP state management mechanism.(1997).
[5]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 122-133
[6]Another MD5 http://www. yicong.com/study/printpage.asp? ArticleID=52
[7]Bloom Filter http://www.cs.wisc.edu/～cao/papers/summary-cache/node8.html
[8]Michael, R. and Oliver, S. WEB Caching AND Replication. Prentice hall, Addison-Wesley. (2001).Page 99-111
[9]The introduction of IR Cache http://www.ircache.net/
[10]曾新平 刘美华 韩真 等 预取技术研究进展 计算机科学．2003．030(008)．-28-30，55
[11]L.Breslau, Pei Cao, Li Fan, G.Phillips, and S.Shenker. Web caching and Zipf-like Distributions: Evidence and implications. In Proceeding of IEEE INFOCM, April 1999.
[12]Carlos Cunha, Azer Bestavros and M. Crovella. Characteristics of WWW client-based Traces. Technical Report BU-Cs-95-010, Computer Science Department, Boston .University, April 1999.
[13]刘美华 古志民等 web缓存系统浅析 北京理工大学学报．2003．023(001)．-78-82
[14]Shudong Jin and Azer Bestavros. Sources and Characteristics of web temporal locality.
[15]Marting Ariltt, Rich Friedrich and Tai Jin. Workload Characterization of a web proxy in A cable modem environment. Technical Report HPL-1999-48, Hewlett Packard Labs.
[16]Chankhunthod A, Danzig P B,Neerdaels C,et al. A Hierarchical Internet Object Cache.Usenix96,1996-012
[17]Rodriguez P, Spanner C, Biersack E W.Web Caching Architectures:Hierarchical and Distributed Caching.Proceedings of WCW'99,1993
[18]郭琳 李欣等 网络缓存技术计算机系统应用．2002．000(008)．-44-45
[19]Li B, Golin M J, Italiano G F, et aI.On the Optimal Placement of Web Proxies in the Internet.Proceedings of Infocom'99, 1996
[20]Aggarwal C,Wolf J L,Yu P S.Caching on the World Wide Web.IEEE Transactions on
Knowledge and Data Engineering, 1999,11 (1)
[21]Crovella ME, Carter RL. Dynamic server selection in the Internet. In: Proceedings of the 3rd IEEE Workshop on the Architecture and Implementation of High Performance Communication Subsystems (HPCS?5). Piscataway: IEEE Communication Society Press, 1995. 158～162.
[22]Myers A, Dinda P, Zhang H. Performance characteristics of mirror servers on the Internet. In: Proceedings of the IEEE INFOCOM?9. Piscataway: IEEE Press, 1999. 304～312.
[23]Cardellini V, Colajanni M, Yu KS. Dynamic load balancing on Web-server system. IEEE Internet Computing, 1999,3(3):28～39.
[24]Iyengar A, Challenger J, Dantzig P. High-Performance Web site design techniques. IEEE Internet Computing, 2000,4(2):17～26.
[25]Schroeder T, Goddard S, Ramamurthy B. Scalable Web server clustering technologies. IEEE Network, 2000,14(3):38～45.
[26]Andresen D, Yang T, Holmedahl V, Ibarra OH. SWEB: towards a scalable World Wide Web server on multicomputers. In: Proceedings of the IPPS?6. Los Alamitos: IEEE Computer Society Press, 1996. 850～856. http://computer.org/proceedings/ipps/7255/7255toc.htm.
[27]Baker SM, Moon B. Distributed cooperative Web servers. Computer Networks and ISDN Systems, 1999,31(11-16):1215～1229.
[28]Lin YW, Zhang DJ, Qian HL. A cooperative Web caching system based on concentrated management. Journal of Computer Research and Development, 2001,38(1):68～73 (in Chinese with English Abstract).
[29]Pai VS, Aron M, Banga G, et al. Locality-Aware request distribution in cluster-based network servers. In: Proceedings of the ASPLOS-Ⅷ. New York: ACM Press, 1998. 205～216. http://computer.org/proceedings/ipps/7255/7255toc.htm.
[30]Aron M, Sanders D, Druschel P, et al. Scalable content-aware request distribution in cluster-based network servers. In: Proceedings of the USENIX 2000 Annual Technical Conference. Berkeley, CA: USENIX Association, 2000. 323～336. http://www.usenix.org/events/usenix2000/program.pdf.
[31]Vinod Valloppillil and Keith W.Ross. Cache Array Routing Protocol, v 1.0. http://ircache.nlanr.net/cache/ICP/draft-vinod-carp-v1-02.txt, 1997.
[32]Li Fan, Pei Cao and Jussara Almeida. Summary cache : A scalable wide-area web cache Sharing protocol. Proceedings of SIGCOMM98,1998.
[33]Alex Rousskov and Duane Wessels. Cache Digest .http://ircache.nlanr.net
[34]Grimm C. And Neitzner M. Request Routing in cache meshes. Computer Networks and ISDN System, 1998.