本文選題：集群管理 + 集群監(jiān)控��； 參考：《西安電子科技大學》2014年碩士論文

【摘要】：高性能計算的迅猛發(fā)展使其在醫(yī)學、航天、生物等領域中占有舉足輕重的地位。隨著問題復雜度的提升，高性能集群的規(guī)模也隨之增加。傳統(tǒng)的集群監(jiān)管方式已無法滿足用戶需求，命令行方式的操作為用戶操作帶來諸多不便。web應用不但可以為用戶提供交互友好的操作界面，也可以完成復雜的后臺處理邏輯，越來越多的管理系統(tǒng)也采用web的方式進行開發(fā)和設計。將集群的底層的數(shù)據(jù)采集與命令管理和web技術相結(jié)合，不但能直觀的顯示集群的運行狀態(tài)，同時為用戶屏蔽了命令行交互帶來的不便。 本論文設計并實現(xiàn)了基于web的集群管理系統(tǒng)——omni，，為用戶提供高效便捷的集群管理平臺。系統(tǒng)在邏輯上劃分為四個層次：集群監(jiān)控層，數(shù)據(jù)存儲層，集群管理層和GUI層。系統(tǒng)采用C/S和B/S混合架構(gòu)模式，并依托于集群操作系統(tǒng)rocks和調(diào)度軟件torque，omni提供了集群管理和監(jiān)控功能。本論文將介紹系統(tǒng)網(wǎng)絡架構(gòu)和軟件體系結(jié)構(gòu)，集群監(jiān)控以及數(shù)據(jù)存儲等方面的設計與實現(xiàn)，并將集群管理作為本文的重點進行詳細闡述。系統(tǒng)為用戶提供web接口，用戶無需使用任何命令便可實現(xiàn)對集群的管理和監(jiān)控操作。此外，論文對系統(tǒng)進行功能和性能的測試，實驗表明本系統(tǒng)可以較好的滿足集群監(jiān)管系統(tǒng)的需求，并且具有較高的可靠性，用戶體驗良好。
[Abstract]:With the rapid development of high-performance computing, it plays an important role in medicine, aerospace, biology and other fields. As the complexity of the problem increases, the scale of high performance clusters increases. The traditional cluster supervision mode can not meet the needs of users. Command line operation brings a lot of inconvenience to user operation. Web application can not only provide users with interactive and friendly operation interface, but also complete complex background processing logic. More and more management systems are also developed and designed by web. Combining the data collection with command management and web technology at the bottom of the cluster, not only can the running state of the cluster be displayed intuitively, but also the inconvenience caused by the command line interaction can be shielded for the user. This paper designs and implements a cluster management system based on web omnibus, which provides users with an efficient and convenient cluster management platform. The system is logically divided into four layers: cluster monitoring layer, data storage layer, cluster management layer and GUI layer. The system adopts the hybrid architecture of C / S and B / S, and relies on the cluster operating system rocks and the scheduling software torqueomni to provide cluster management and monitoring functions. This paper will introduce the system network architecture and software architecture, cluster monitoring and data storage and other aspects of the design and implementation, and cluster management as the focus of this paper is described in detail. The system provides the user with web interface, and the user can manage and monitor the cluster without any commands. In addition, the paper tests the function and performance of the system. The experiment shows that the system can meet the needs of the cluster supervision system, and has high reliability and good user experience.
【學位授予單位】：西安電子科技大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.09

【參考文獻】

相關期刊論文 前4條

1 李紅梅;;Linux高性能計算集群的設計與實現(xiàn)[J];電腦知識與技術;2008年14期

2 趙毅;朱鵬;遲學斌;牛鐵;曹宗雁;;淺析高性能計算應用的需求與發(fā)展[J];計算機研究與發(fā)展;2007年10期

3 張濤,于雪芹;基于XDR綱要的空間元數(shù)據(jù)存儲策略[J];計算機應用;2005年07期

4 張予倩;周健;翁紅明;韓靜;;Rocks高性能計算集群的建立和管理[J];實驗室研究與探索;2006年04期

本文編號：1905958