網(wǎng)絡(luò)流量監(jiān)測(cè)算法的研究及其實(shí)現(xiàn).pdf

1、南京郵電大學(xué)碩士學(xué)位論文網(wǎng)絡(luò)流量監(jiān)測(cè)算法的研究及其實(shí)現(xiàn)姓名：王銀花申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)軟件與理論指導(dǎo)教師：陳春玲20090301南京郵電大學(xué)碩士研究生學(xué)位論文AbstractAbstractNetworktechnologyisalaborintensiveindustriesWithoutrealtimemonitoringandappropriatemanagementonnetworktraffic，networkwil

2、lbeeasilybusy，itcancausenetworkcongestionandotherconditionsEventually，usersareaffectedanddirecteconomiclossesOCCUrInordertoalleviatethepressure，anumberoflargenetworkequipmentmanufacturersbothathomeandabroadhavebeentrying

3、hardtofindthebestsolutionAtpresent，existingtrafficmonitoringtechnologieshavehighdemandonCPUandmemoryofhostfortrafficmonitoring，anddonotfacilitatecentralizedmanagementandtrafficmonitoringonmanylinksandlarge—scaleregionaln

4、etworkTheCPUloadandmemoryloadofMonitoringagentisrelativelylarge，theoutputinformationisnotperfect，networkinterfacestatisticalinformationcannotbeobtained。Informationoutputtedistoofrequenttotakemorenetworkbandwidthbetweenth

5、einformationcollectorandthemonitoringfacilitiesSo，wedesignandrealizetrafficmonitoringtechnologyNetStream，whichcanimprovesomeshortcomingsofexistingtrafficmonitoringtechnologyeffectivelyTheNetStream’Sessentialalgorithmisit

6、spackagesamplingalgorithmAlotofacademicrandomsamplingalgorithmsCannotbeachievedduetohardwarelimitationsCombindingthedevelopmentofNetStreamandhardware，wedesignafastpseudo—randomsamplingalgorithmandrealizethealgorithmThisa

7、lgorithmismoreeffectivetomitigateNP’SburdenonthepacketmirroringwithoutaffectingthedatastatisticsAnalysesofexperimentalresultprovethatthealgorithmiseffectiveKeywords：TrafficMonitoring，TrafficCollection，RandomSampling，NetS