DSM系統在多人同時使用的情況下，DSM各節點可能會因為某些執行節點產生重疊現象而負載過重，某些節點因無人使用而處於閒置狀態。在傳統的執行環境之下，因無法取得系統整體資訊，評估各DSM應用程式間相互干擾的程度，所以並無法避免或改善此系統負載不平衡的現象。故在此我們加入一個獨立的資源管理者而成為三層式遠端執行管理架構，藉由完整的節點資訊收集，對資源做出最有效的管理與分配。另外可透過DSM應用程式執行時收集的狀態資訊，以分析其效能的瓶頸，並更進一步經由動態負載重分配機制，藉由適當地執行緒搬移使工作量動態地從負載重的節點搬移到負載輕的節點，進而達到負載平衡的目的。最後從我們的實驗結果裡可以發現，在經由動態負載重分配機制調整各節點上的工作量之後，在DSM應用程式的執行效能與總生產率上，都有相當明顯地提升。

When people execute applications on the same DSM system group, some nodes maybe have overlapping phenomenon which more than one applications executing on the same node. These overlapping nodes may be overloaded, but some will be in the status of idle at the same time. In traditional DSM environments, there is no global information about all the executing application clusters. It will be very difficult to appraise the interference among different DSM applications without the global information. Therefore, it can not efficiently improve the load imbalance phenomenon of each DSM application. Based on the above description, we insert a remote manager into the original DSM system architecture and make it become a three-tier executing and managing architecture, including the DSM group, manager, and consoles. With the global information, the manager can manage and dispatch the resources of DSM group more efficiently. Besides, the manager can efficiently detect the performance bottleneck and lunch the dynamic load balance mechanism automatically to improve the overall performance of the DSM group. In the experimental results, we can see that the application performance and system throughput will be promoted efficiently with our dynamic load redistributing scheme.

[1] James Coomer and Charu Chaubal; “Introduction to the Cluster Grid-Part1”; Sun BluePrintsTM Online; August 2002.

[2] James Coomer and Charu Chaubal; “Introduction to the Cluster Grid-Part2”;Sun BluePrintsTM OnLine; September 2002.

[3] Jim Basney and Miron Livny; “Deploying a High Throughput Computing Cluster”;Department of Computer Sciences University of Wisconsin-Madison Wisconsin;USA.

[4] Niranjan G. Shivaratri, Phillip Krueger, Mukesh Singhal; "Load Distributing for Locally Distributed Systems"; Computer, Vol. 25, No. 12, pp. 33-44;December 1992.

[5] An-Chow Lai, Ce-Kuen Shieh, Yih-Tzye Kok; “Load balancing in distributed shared memory systems”;Proceeding of IEEE International Performance,Computing,and communications Conference, Arizona, U.S.A., pp152-158;February 1997.

[6] Alex Dubrovski, Roy Friedman and Assaf Schuster, “Load Balancing in Distributed Shared Memory Systems”;International Journal of Applied Software Technology, vol 3, pp.167-202, March 1998.

[7] Jyh-Chang Ueng, Ce-Kuen Shieh, Wu-Hwa Liu, and Chi-Chang Lin, “Proteus: a runtime reconfigurable distributed shared memory system”; High Performance Distributed Computing, 1999. Proceedings. The Eighth International Symposium on Page(s):347-348, 1999.

[8] K. Thitikamol and P. Keleher; “Thread migration and load balancing in nodedicated environments”; Proc. Of the 14th Intl. Parallel and Distributed Processing Symp.,pages 583-588, May 2000.

[9] Weisong Shi and Zhimin Tang; “Dynamic Computation Scheduling for Load Balancing in Home-based Software DSMs”; Proceedings of the 1999 International Symposium on Parallel Architectures, Algorithms and Networks(I-SPAN'99), IEEE Computer Press, Perth, Australia, June, 1999.

[10]Weisong Shi, Zhimin Tang; “Load Balancing in Home-Based Software DSMs”; International Journal of Foundaions of Computer Science 12(3):307-324,2001.

[11]Vinod G.J. Peris, Mark S. Squillante, and Vijay K. Naik; “Analysis of the Impact of Memory in Distributed Parallel Processing Systems”; Proceedings of the 1994 ACM SIGMETRICS Conference, page 5-18; February 1994.

[12]Jim Mauro and Richard McDougall; "Solaris Internals: Core Kernel Components"; Sun Microsystems Press, 2001, ISBN: 0-13-022496-0

[13]J. B. Chang and C. K. Shieh; "Teamster: A Transparent Distributed Shared Memory for Clustered Symmetric Multiprocessors"; Workshop on Distributed Shared Memory on Clusters at IEEE CCGrid’2001, Brisbane, Australia. Page 508-513; May 2001

[14]Chi-Chung Liao; "Communication Minimization of Progressive Multi-layer Reconfiguration on Teamster"; Master thesis, Department of Electrical Engineering, National Cheng Kung University, Tainan, Twiwan, R.O.C, 2001

[15]Adrian Cockcroft and Richard Pettit, "Sun Performance and Tuning: Java and the Internet 2/e," Sun Microsystems Press, 1998, ISBN: 0-13-095249-4