简介:Offlinenetworktrafficanalysisisveryimportantforanin-depthstudyupontheunderstandingofnetworkconditionsandcharacteristics,suchasuserbehaviorandabnormaltraffic.WiththerapidgrowthoftheamountofinformationontheInternet,thetraditionalstand-aloneanalysistoolsfacegreatchallengesinstoragecapacityandcomputingefficiency,butwhichistheadvantagesforHadoopcluster.Inthispaper,wedesignedanofflinetrafficanalysissystembasedonHadoop(OTASH),andproposedaMapReduce-basedalgorithmforTopNuserstatistics.Inaddition,westudiedthecomputingperformanceandfailuretoleranceinOTASH.FromtheexperimentswedrewtheconclusionthatOTASHissuitableforhandlinglargeamountsofflowdata,andarecompetenttocalculateinthecaseofsinglenodefailure.