臺灣博碩士論文加值系統

　　當企業建置了叢集伺服架構，企業的營運狀況與叢集運作息息相關，一旦伺服器異常或者發生故障，將使企業無法順利作業而造成損失，為了維持數量龐大的伺服器叢集足以順利運作，叢集伺服器的管理的確有其迫切需求及重要性。
　　IPMI(Intelligent Platform Management Interface)稱為智慧平台管理介面，是一個智慧型管理系統的標準，藉由IPMI提供的跨平台標準介面可以監控系統內的狀況。透過IPMI標準介面，可以大幅減少伺服器管理成本，並且有效解決各種伺服器與周邊設備因介面不同而無法溝通的問題。
　　本研究在Embedded Linux 上建置建置符合IPMI v2.0規範的IPMI Firmware，藉以處理各項遠端輸入的IPMI standard command，利用IPMI over LAN的方式下達各個管理伺服器的指令來提供遠端叢集管理，為企業達到更高的營運成效。主要為伺服器叢集管理增加遠端監控伺服器硬體資訊的能力以及遠端伺服器開機的能力，克服傳統伺服器管理無法針對硬體監控以及當系統死當時無法進行遠端開機的困難。

After enterprises build their server clusters, the operation of enterprises will be highly related to that of server clusters. Once the server has a breakdown, enterprises will not be able to operate regularly, then making a loss. Therefore, the management of server clusters is necessary and plays an important role in maintaining regular operation of numerous server clusters.
IPMI (Intelligent Platform Management Interface) is a standard of intelligent management system. We can monitor the condition in the system through the cross-platform standard interface provided by IPMI. Furthermore, we can reduce the cost of server management and effectively solve a variety of problems resulting from that the interface of server cannot be compatible with that of its accessory appliances.
My research focuses primarily on establishing IPMI Firmware according with IPMI v2.0 spec on Embedded Linux. This IPMI Firmware helps us to deal with any IPMI standard commands from remote input. In addition, through IPMI over LAN we can transmit commands from management server to provide remote server cluster management, and this way helps enterprises to achieve better operating performance. As a result, server cluster management will enhance the ability to remotely monitor the hardware information in the server and the ability to turn on the remote server. Finally, server cluster management will overcome the difficulty that traditional server is unable to monitor computer hardware and remotely turn on the computer when computer system malfunctions.

第一章 緒論 1
1.1 研究背景 1
1.2 Intelligent Platform Management Interface 2
1.2.1 IPMI架構 2
1.2.2 IPMI功能 3
1.3 Cluster computing 5
1.4 研究動機及目的 6
第二章 文獻回顧 8
2.1 伺服器管理 8
2.2 叢集系統 12
2.2.1 High availability 14
2.2.2 Scalability 17
第三章 研究方法 19
3.1 研究架構 19
3.2 IPMI Core on BMC 21
3.3 Message handler 21
3.3.1 Implement Channel (通道) 22
3.3.2 Implement Session (連線) 24
3.3.3 Implement Message Bridging (橋接) 25
3.3.4 Other Commands 28
3.4.1 Get Device ID Command 29
3.4.2 Cold Reset Command 30
3.4.3 Get Self Test Results Command 31
3.4.4 Get Chassis Status Command 31
3.4.5 Chassis Control Command 32
3.4.6 Get System Restart Cause command 33
3.5.1 Get Sensor Reading Command 34
第四章 實驗過程與結果 39
4.1 開發環境 39
4.2 建立I/O table 42
4.3 SDR Generator 46
4.4 實作測試 46
4.4.1 可管理性實作 46
4.4.2 高可用性實作 52
4.4.3 可延展性實作 55
第五章 結論與建議 57
參考文獻 58
附錄一：IPMI Core Layout 61

[1] Yung Chin Fang、Garima Kochhar、Randy Deroeck，運用智慧型平台管理介面(IPMI)管理與監控高效能運算叢集，http://tech.digitimes.com.tw/，九十四年八月
[2] 台灣微軟，「協助群益證券強化軟體安控、IT 資產清算及軟體部署，提昇企業競爭優勢」，http://www.microsoft.com/taiwan/smserver/evaluation/ casestudies/capital.htm，2004年2月。
[3] 李政諺，通用型IPMI KCS驅動程式之實作研究，世新大學，九十五年九月
[4] 美商甲骨文，運用Oracle Real Application Clusters建立高可用性資料庫伺服器White Paper，http://www.oracle.com，May 2001
[5] 張西亞、王順泰、周朝宜、陳德民、吳長興、李金泓、謝志偉，高效能電
腦叢集的發展與趨勢，物理雙月刊(廿九卷五期)，2007年十月
[6] 許致學，SQL Server 2005 高可用度技術探討，TechNet技術講座，2006年6月
[7] 劉仲怡，植基於IPMI的伺服器監控及管理效能提升之研究，世新大學，九十五年九月
[8] 陳彥佑，IPMI OEM智慧管理功能設計與實作，世新大學，九十五年九月
[9] 陳信諺，SMBIOS的研究與實作，世新大學，九十五年九月
[10] 戴伯愷，叢集電腦的主動式BMC資源管理，世新大學，九十五年九月
[11] 蕭富方，遠端伺服器監控管理系統設計與實現，世新大學，九十五年八月
[12] Bryan Thomas, Solutions for Highly Scalable Database Applications An analysis of architectures and technologies, Performance Tuning Corporation, Solutions for Highly Scalable Database Applications, May 2005.
[13] Chokchai Box Leangsuksun, Lixin Shen, Tong Liua, and Stephen L. Scott, Achieving high availability and performance computing with an HA-OSCAR cluster, Future Generation Computer Systems 21, 2005, Pages 597–606.
[14] D. Ridge, D. Becker, and P. Merkey, Beowulf: Harnessing the Power of Parallelism in a Pile-of-PCs, Proceedings of IEEE Aerospace, 1997.
[15] F. Gregory Pfister, In Search of Clusters, Prentice Hall, Apr 1998.
[16] Intel Developer Forum, Advances in Intelligent Platform Management: IPMI v2.0 In Action, http://www.intel.com/design/servers/ipmi/ipmi.htm , Sep. 2004.
[17] Intel Developer Forum, Intelligent Platform Management Futures: IPMI, CIM and Emerging Management Technologies, Spring 2005.
[18] Intel HP NEC Dell, Intelligent Platform Management Interface Specification Second Generation v2.0, http://www.intel.com/design/servers/ipmi/ipmi.htm, Feb 2006.
[19] J. Gray and D. Siewiorek, High-availability computer systems, IEEE Computer, 24(9):39–48, Sep 1991.
[20] Kanwardeep Singh Ahluwalia and Atul Jain, High availability design patterns, 2006 conference on Pattern languages of programs,October 2006.
[21] K. Hwang and Z.Xu, Scalable Parallel Computing: Technology, Architecture and Programming, WCB/McGraw-Hill, NY, 1998.
[22] Kiril Kamenev, Clustered Network Applications, Seminar on Internetworking, Apr. 2004
[23] Madhu Tumma, Overview of Oracle Real Application Clusters, http://www.ioug.org/client_files/members/select_pdf/aug03/IOUGRealApplicationCluster.pdf, 4th Qtr 2003.
[24] Meikel Poess and Raghunath K.Othayoth, Large Scale Data Warehouses on Grid:Oracle Database 10g and HP ProLiant Servers, Proceedings of the 31st VLDB Conference, pp.1055-1066, Jan 2005.
[25] M. Michael, J.E. Moreira, D. Shiloach, and R.W. Wisniewski, Scale-up x Scale-out: A Case Study using Nutch/Lucene, Parallel and Distributed Processing Symposium, 2007. (IPDPS 2007, IEEE International, Jan 2008)
[26] Oracle, An Oracle Technical White Paper, http://www.oracle.com, April 2007.
[27] Top500 List- November 2008, http://www.top500.org/lists/2008/11, Nov 2008.
[28] T. Sterling, D. Becker, D. Savarese et. al., BEOWULF: A PArallel Workstation for Scientific Computation, Proceedings of the 1995 International Conference on Parallel Processing(ICPP), August 1995.
[29] Yi-Hung Lin, Achieving High Availability and Performance Computing for ERP System Using Oracle Real Application Cluster 10g Architecture, Shih Hsin University, June 2007.
[30] Zafar Mahmood, Anthony Fernandez, Bert Scalzo, and Murali Vallath, Testing Oracle 10g RAC Scalability on Dell PowerEdge Servers and Dell/EMC Storage, DELL POWER SOLUTIONS, Feb. 2006.