【摘要】：隨著網(wǎng)絡(luò)的快速發(fā)展,網(wǎng)絡(luò)的拓?fù)浣Y(jié)構(gòu)和新型業(yè)務(wù)不斷出現(xiàn)。為了保證網(wǎng)絡(luò)的正常運(yùn)行,提供全面可靠的網(wǎng)絡(luò)管理,需要研究可行性的網(wǎng)絡(luò)監(jiān)測(cè)方法。大規(guī)模網(wǎng)絡(luò)作為一種常見(jiàn)的網(wǎng)絡(luò)形式,具有維數(shù)多、層級(jí)多、結(jié)構(gòu)復(fù)雜等特點(diǎn),對(duì)大規(guī)模網(wǎng)絡(luò)進(jìn)行有效的管理并提供準(zhǔn)確的網(wǎng)絡(luò)數(shù)據(jù)有著重要的意義。虛擬化監(jiān)測(cè)方法根據(jù)網(wǎng)絡(luò)的拓?fù)湫畔⒁约肮?jié)點(diǎn)屬性提供多種測(cè)量模式,完成對(duì)網(wǎng)絡(luò)的動(dòng)態(tài)監(jiān)測(cè)。 大規(guī)模網(wǎng)絡(luò)結(jié)構(gòu)復(fù)雜,需要研究有針對(duì)性的監(jiān)測(cè)方法。本文介紹了云監(jiān)測(cè)系統(tǒng),包括云監(jiān)測(cè)平臺(tái)和部署到網(wǎng)絡(luò)中的探針。對(duì)于監(jiān)測(cè)點(diǎn)的選取,探針的部署必須在代價(jià)和覆蓋范圍之間進(jìn)行權(quán)衡。該文以最小化監(jiān)測(cè)探針數(shù)目為目標(biāo),同時(shí)覆蓋盡可能廣的網(wǎng)絡(luò)鏈路,采用了一種基于蟻群算法的網(wǎng)絡(luò)監(jiān)測(cè)探針部署方法。如果測(cè)量集合中的監(jiān)測(cè)點(diǎn)沒(méi)有部署探針,系統(tǒng)進(jìn)行自動(dòng)部署探針程序。根據(jù)網(wǎng)絡(luò)情況以及節(jié)點(diǎn)的狀態(tài),可以采用虛擬機(jī)鏡像方式和軟件包部署方式。系統(tǒng)提供三種網(wǎng)絡(luò)監(jiān)測(cè)方法滿足不同用戶的需求,分別為指定鏈路模式、自動(dòng)監(jiān)測(cè)模式、用戶制定模式。在任務(wù)執(zhí)行過(guò)程中系統(tǒng)監(jiān)測(cè)節(jié)點(diǎn)的資源狀態(tài),動(dòng)態(tài)調(diào)整探針,靈活調(diào)整監(jiān)測(cè)點(diǎn)位置。系統(tǒng)部署在全球性大規(guī)模網(wǎng)絡(luò)PlanetLab上進(jìn)行部署測(cè)試,下發(fā)大量長(zhǎng)時(shí)間任務(wù)收集測(cè)量數(shù)據(jù),驗(yàn)證了系統(tǒng)對(duì)網(wǎng)絡(luò)監(jiān)測(cè)結(jié)果的準(zhǔn)確性和運(yùn)行的可靠性。 在介紹虛擬化監(jiān)測(cè)方法的過(guò)程中,首先介紹了本文的選題背景及意義,說(shuō)明了對(duì)大規(guī)模虛擬化監(jiān)測(cè)方法提出的背景。隨后對(duì)相關(guān)技術(shù)背景,包括網(wǎng)絡(luò)測(cè)量發(fā)展現(xiàn)狀及性能指標(biāo)等進(jìn)行了闡述。通過(guò)對(duì)虛擬化監(jiān)測(cè)方法的需求分析,對(duì)系統(tǒng)進(jìn)行總體設(shè)計(jì)并對(duì)各個(gè)子功能進(jìn)行詳細(xì)設(shè)計(jì)。隨后對(duì)監(jiān)測(cè)點(diǎn)選取方法、自動(dòng)部署、靈活調(diào)整監(jiān)測(cè)點(diǎn)實(shí)現(xiàn)過(guò)程進(jìn)行了說(shuō)明。系統(tǒng)最后在大規(guī)模網(wǎng)絡(luò)P1anetLab上進(jìn)行實(shí)驗(yàn)測(cè)試,并對(duì)測(cè)量結(jié)果進(jìn)行了分析。測(cè)試數(shù)據(jù)驗(yàn)證了虛擬化監(jiān)測(cè)方法的有效性以及穩(wěn)定性。
[Abstract]:With the rapid development of network, network topology and new services are emerging. In order to ensure the normal operation of the network and provide comprehensive and reliable network management, it is necessary to study the feasible network monitoring method. As a common network form, large-scale network has many characteristics, such as multi-dimension, multi-level, complex structure, etc. It is of great significance to manage large-scale network effectively and provide accurate network data. Virtualization monitoring method provides a variety of measurement modes according to the network topology information and node properties to complete the dynamic monitoring of the network. Because of the complexity of large-scale network structure, targeted monitoring methods need to be studied. This paper introduces cloud monitoring system, including cloud monitoring platform and probe deployed to the network. For the selection of monitoring points, the deployment of probes must be balanced between cost and coverage. Aiming at minimizing the number of monitoring probes and covering as wide a network link as possible, an ant colony algorithm based approach for the deployment of network monitoring probes is proposed in this paper. If the probe is not deployed at the monitoring point in the measurement set, the system performs an automatic probe deployment program. According to the network situation and the status of nodes, virtual machine mirror and package deployment can be adopted. The system provides three network monitoring methods to meet the needs of different users, namely, designated link mode, automatic monitoring mode, and user development mode. In the process of task execution, the system monitors the resource status of nodes, adjusts the probe dynamically, and adjusts the position of monitoring points flexibly. The system is deployed on the global large-scale network PlanetLab for deployment test, and a large number of long time tasks are sent to collect the measurement data, which verifies the accuracy of the system to the network monitoring results and the reliability of the operation. In the process of introducing virtualization monitoring method, the background and significance of this paper are introduced, and the background of large-scale virtualization monitoring method is explained. Then the related technical background, including the development of network measurement and performance indicators are described. By analyzing the requirements of the virtualization monitoring method, the system is designed in detail and the sub-functions are designed in detail. Then, the method of monitoring point selection, automatic deployment and flexible adjustment of monitoring point realization process are explained. Finally, the system is tested on the large-scale network P1anetLab, and the measurement results are analyzed. The test data verify the effectiveness and stability of the virtualization monitoring method.
【學(xué)位授予單位】：北京郵電大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TP393.06

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 錢(qián)進(jìn);賀貴明;;分布式網(wǎng)絡(luò)性能監(jiān)測(cè)的探針部署方法研究[J];計(jì)算機(jī)工程;2007年07期

，

本文編號(hào)：2400139