摘要
随着微处理器技术和高性能网络技术的飞速发展，集群计算逐渐成为一种有成本效益的并行／分布式计算资源。集群计算系统的核心问题是资源的共享及有效利用，只有平衡的负载才能达到最大的资源使用率。为了充分有效的利用整个集群资源，进程必须能够在集群中自由的迁移以达到有效的利用底层资源的能力和适应负载的动态变化。进程间的网络连接不仅直接影响迁移的透明性，带来了迁移限制，而且导致极大的剩余相关性，对迁移后的性能和系统的健壮性有很大的负面影响。因此，必须能够提供对网络连接迁移的支持，保持进程迁移前打开的网络连接的能力。我们将其称为套接字迁移。
本文研究了TCP／IP协议及其在LINUX系统下的实现，分析了网络连接的状态，给出了网络连接状态的提取、网络连接迁移和恢复的解决方案，解决了集群环境下进程迁移如何透明的保持TCP网络连接这一问题。本文还给出了套接字迁移在MOSIX集群系统上的一个原型实现。
本文对网络连接的迁移进行了有意义的探索，对集群系统设计和进程迁移机制研究有一定的参考价值。
Improvements in microprocessor and networking performance have made cluster system a very attractive platform for high-end parallel and distributed computing. The core problem in cluster system is how to share the cluster resources and fully utilize them. The best performance can not be achieved when loads among nodes are unbalanced. Achieving good performance requires that processes be able to migrate between cluster resources and efficiently adapt to the underlying resource capabilities. However,the network communication between processes not only affects the transparency of migration,but also puts limit on some processes which may cause them unfit for migration . Another drawback is that it may cause residual dependency on the previous host which will impacts on the performance and robustness of the system. Therefore we must provide a mechanism to support the migration of network connections and maintain network connectivity after process migration. We call this mechanism "Socket Migration".
In the thesis,the TCP/IP protocol and it's implementation on Linux is studied. The thesis presents how to retrieve the state of network connection and how to migrate established TCP connection. The problem that how to maintain network connection after process migration can be solved perfectly with our methods. We also implement a prototype of "Socket Migration" on the basis of MOSIX system.
The thesis makes a series of researches on migration of network connection and can help in the design of cluster system and study of process migration mechanism.
引文
[1]D. S Milojicic, Fred Douglis, Richard Wheeler: "Process Migration". 1999.
[2]M. R. Eskicioglu: "Design Issues of Process Migration Facilities in Distributed Systems", IEEE Technical Committee on Operating Systems Newsletter, vol. 4, no. 2, pp. 3-13, Winter, 1989.
[3]Douglis, Fred and Ousterhout, John: "Process Migration in Sprite: A Status Report", IEEE TCOS Newsletter, Vol. 3, No. 1, Winter 1989, pp. 8-10.
[4]Barak A, La' adan O, Shiloh A: "Scalable Cluster Computing with MOSIX for LINUN ", Proc. Linux Expo '99, pp. 95-100, Raleigh, N. C., May 1999.
[5]Rajkumar Buyya: "High Performance Cluster Computing: Architures and System, Volume 1" Prentice HALL PTR.
[6]B. Downey, M. Harchol-Balter: "A note on "The Limited Performance Benefits of Migrating Active Processes for Load Sharing ". Technical Report UCB/CSD-95-888, Computer Science Devision, University of California, Berkeley, California 94720, November 1995.
[7]Peter Smith, Norman C. Hutchinson: "Heterogeneous process migration: The tui system". Technical Report TR-96-04, University of British Columbia. Computer Science, February 1996.
[8]M. Litzkow, T. Tannenbaum, J. Basney: "Checkpoint and Migration of UNIX Processes in the Condor". Distributed Processing System, Technical Report #1346, Computer Sciences Department, University of Wisconsin, April 1997.
[9]F. Douglis: "Transparent Process Migration in the Sprite Operating System" .PhD thesis, University of California, Berkeley, CA 94720, September 1990.
[10]Martin O'Connor, Brendan Tangney: "Process Migration on the Chorus Micro-kernel" Technical Report TCD-CS-TR-93.02[INT], University of Dublin, Trinity College, February 1993.
[11]A. Barak, O. La'adan: "The MOSIX Multicomputer Operating System for High Performance Cluster Computing". Journal of Future Generation Computer Systems, April 1998.
[12][美]Scott Maxwell: 《Linux 内核源代码分析》．冯锐，邢飞，刘隆国等译．北京，机械工业出版社，2000．
[13]黄翊，蒋江，张民选：“MOSIX进程迁移机制研究”．《计算机工程》．
[14]Perkins, C. E: "IP mobility support RFC 2002"
[15]Irene Wu, Beichuan Zhang, Bing Zhang: "Extended Transmission Control Protocol Project". http://www. chem. ucla. edu/~beichuan/etcp/.
[16]A. C. Snoeren, m. G. Andersen, H. Balakrishnan: "Fine-Grained Failover Using Connection Migration". 3rd USENIX Symposium on Internet Technologies and Systems (USITS'01). San Francisco, CA, March 2001.
[17]Victor. C. Zandy, BartonP. Milller: "Reliable Socket"http: //citeseer. nj. nec. com/zandyO1reliable, html.
[18]Marian Bubak, Dariusz Zbik: "Portable Library of Migratable Sockets" Proceedings of the SGI Users' Conference 2000.
[19]J. Casas, D. L. Clark, R. Konuru: "MPVM: A Migration Transparent Version of PVM" Computing Systems 8, 2, Spring 1995, pp. 171-216.
