發(fā)布時(shí)間：2018-11-18 08:36

【摘要】：服務(wù)器在銀行、電信、電子商務(wù)等一些特殊領(lǐng)域的應(yīng)用，要求其同時(shí)具備很高的性能和可用性。這類(lèi)服務(wù)器需要采用分布式結(jié)構(gòu)的高可用性設(shè)計(jì)方案以達(dá)到其要求。本文立足課題背景，以面向自主計(jì)算的Linux平臺(tái)高可用服務(wù)器系統(tǒng)為目標(biāo)系統(tǒng)，研究了分布式結(jié)構(gòu)高可用服務(wù)器系統(tǒng)管理軟件的設(shè)計(jì)方案，該軟件實(shí)現(xiàn)對(duì)該系統(tǒng)復(fù)雜結(jié)構(gòu)自動(dòng)化的運(yùn)行管理，保證系統(tǒng)各部件各司其職、相互配合、有條不紊地運(yùn)作。 本文首先對(duì)課題的來(lái)源和背景做了介紹，，闡述了相關(guān)領(lǐng)域的發(fā)展概況，包括自主計(jì)算技術(shù)概述、高可用技術(shù)概況和高可用服務(wù)器管理軟件的研究現(xiàn)狀。之后描述了本文目標(biāo)系統(tǒng)的體系結(jié)構(gòu)，并對(duì)各主要部件在系統(tǒng)中的作用和地位做了說(shuō)明。在對(duì)系統(tǒng)全面細(xì)致的需求分析的基礎(chǔ)上，本文給出了面向自主計(jì)算的高可用服務(wù)器系統(tǒng)管理軟件的概要設(shè)計(jì)。最后，文章詳細(xì)闡述了系統(tǒng)管理軟件的詳細(xì)設(shè)計(jì)，包括管理軟件的架構(gòu)設(shè)計(jì)，各主要部件的詳細(xì)設(shè)計(jì)和核心算法的具體描述。 本文著重進(jìn)行了系統(tǒng)管理軟件的架構(gòu)設(shè)計(jì)、分布式結(jié)構(gòu)高可用系統(tǒng)的任務(wù)負(fù)載均衡調(diào)度策略的設(shè)計(jì)、虛擬網(wǎng)絡(luò)管理方案的設(shè)計(jì)以及分布式高可用系統(tǒng)雙機(jī)子系統(tǒng)的高可用管理方案設(shè)計(jì)等工作。關(guān)于軟件的架構(gòu)設(shè)計(jì)，本文通過(guò)對(duì)Linux共享庫(kù)技術(shù)、C++對(duì)象動(dòng)態(tài)創(chuàng)建技術(shù)的使用和一個(gè)經(jīng)過(guò)特殊設(shè)計(jì)的接口類(lèi)，使軟件可運(yùn)行時(shí)動(dòng)態(tài)地加載、卸載和替換功能模塊。這使得軟件具備了可在線升級(jí)、在線擴(kuò)展功能的特征。關(guān)于任務(wù)的負(fù)載均衡調(diào)度，本文設(shè)計(jì)實(shí)現(xiàn)了4種常見(jiàn)的調(diào)度策略，并提出與實(shí)現(xiàn)了一種新的基于動(dòng)態(tài)反饋機(jī)制的新策略。這一策略在保證良好的負(fù)載均衡性的同時(shí)，可降低算法的時(shí)間復(fù)雜度，有效提高調(diào)度執(zhí)行效率。對(duì)于虛擬網(wǎng)絡(luò)管理，本文在設(shè)計(jì)實(shí)現(xiàn)一個(gè)內(nèi)核網(wǎng)絡(luò)地址轉(zhuǎn)換模塊的基礎(chǔ)上加入了增量連接備份的機(jī)制，借以實(shí)現(xiàn)對(duì)虛擬網(wǎng)絡(luò)連接的快速備份，為L(zhǎng)eader失效切換后的無(wú)間斷服務(wù)提供了基礎(chǔ)。對(duì)于雙機(jī)子系統(tǒng)的高可用設(shè)計(jì)方案，本文是基于主動(dòng)故障管理系統(tǒng)做出的。Leader節(jié)點(diǎn)的失效由該故障管理系統(tǒng)檢測(cè)并通知管理軟件進(jìn)行Leader的功能切換。
[Abstract]:The application of server in banking, telecommunication, e-commerce and other special fields requires high performance and availability. This kind of server needs to adopt the high availability design scheme of distributed structure to meet its requirement. Based on the background of the subject, this paper studies the design scheme of the distributed high availability server system management software, which is based on the high availability server system of Linux platform, which is oriented to autonomous computing. The software realizes the operation and management of the automation of the complex structure of the system, and ensures that the various parts of the system perform their respective duties, cooperate with each other, and operate methodically. In this paper, the source and background of the subject are introduced, and the development of related fields is described, including the overview of autonomous computing technology, the general situation of high availability technology and the research status of high availability server management software. Then, the architecture of the target system is described, and the function and position of the main components in the system are explained. Based on the detailed analysis of the requirements of the system, this paper presents the outline design of the high availability server system management software for autonomous computing. Finally, the detailed design of the system management software is described in detail, including the architecture design of the management software, the detailed design of the main components and the detailed description of the core algorithm. This paper focuses on the architecture design of the system management software and the design of the task load balancing scheduling strategy for the distributed high availability system. The design of virtual network management scheme and the design of high availability management scheme of distributed high availability system. About the software architecture design, this paper makes the software dynamically load, uninstall and replace the function module by using the Linux shared library technology, the C object dynamic creation technology and a specially designed interface class. This enables the software to have online upgrade, online extension features. For the load balancing scheduling of tasks, this paper designs and implements four common scheduling strategies, and proposes and implements a new strategy based on dynamic feedback mechanism. This strategy not only ensures good load balance, but also reduces the time complexity of the algorithm and effectively improves the efficiency of scheduling execution. For virtual network management, an incremental connection backup mechanism is added on the basis of designing and implementing a kernel network address translation module, so as to realize the fast backup of virtual network connection. It provides the foundation for the uninterrupted service after the Leader failover. For the high-availability design scheme of dual-machine sub-system, this paper is based on the active fault management system. The failure of the Leader node is detected by the fault management system and the management software is notified to switch the functions of the Leader.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2012
【分類(lèi)號(hào)】：TP368.5

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 周集良,彭小寧,王正華;基于集群的負(fù)載平衡調(diào)度算法研究與實(shí)現(xiàn)[J];計(jì)算機(jī)工程;2005年12期

2 王晉鵬,潘龍法,李降龍;LVS集群中的動(dòng)態(tài)反饋調(diào)度算法[J];計(jì)算機(jī)工程;2005年19期

3 張�？�;史忠植;;自主計(jì)算環(huán)境[J];計(jì)算機(jī)工程;2006年07期

4 李健;王玲;李俊;閻保平;;基于Linux內(nèi)核Netfilter框架的P2P管理[J];計(jì)算機(jī)工程;2007年12期

5 劉蕾,王欣;Petri網(wǎng)在高可用集群性能分析中的建模[J];計(jì)算機(jī)工程與設(shè)計(jì);2004年10期

本文編號(hào)：2339493