分布式计算机系统节点运行分析的随机RA网络(IJCNIS-V8-N6-1)

I.J.ComputerNetworkandInformationSecurity,2016,6,1-8PublishedOnlineJune2016inMECS()DOI:10.5815/ijcnis.2016.06.01Copyright©2016MECSI.J.ComputerNetworkandInformationSecurity,2016,6,1-8StochasticRA-NetworkfortheNodesFunctioningAnalysisintheDistributedComputerSystemsZhengbingHuSchoolofEducationalInformationTechnology,CentralChinaNormalUniversity,Wuhan,ChinaE-mail:[e-mail],HeorhiiLoutskii,YaroslavKornagaNationalTechnicalUniversityofUkraine―KievPolytechnicInstitute‖Kiev,03056,UkraineE-mail:[e-mail]—InthepaperisdescribedthesimulatingprocessforthesituationsanalysisandthedecisionsmakingaboutthefunctioningoftheDistributedComputerSystems(DCS)nodesonthebasisofspecialstochasticRA-networksmechanism.TherearepresentedthemainproblemsintheestimationsoftheDCSnodesfunctioningparametersandthereareshownthatthesuggestedRA-networksmechanismallowssimulatethedataflowwiththedifferent,includingthesignificantlydifferentintensities,whatisparticularlyimportantinforthesituationsanalysisandthedecisionsmakingintheDCSnodesparametersdynamicscontrol.IndexTerms—Distributedcomputersystems,parametersanalysis,decisionsmaking,RA-networks.I.INTRODUCTIONOneofthemostimportantaspectsinthedesignofthedistributedcomputersystems(DCS)isananalysisoftheirfunctionalparameters.TheDCShasaspecialjobscheduler,which,inparticular,performsthedistributionoftheresourcesbetweenthetasks,includingthetasksrelatedtoDCSsecurityensuring[1-3].Themalfunctionofjobscheduler,inparticular,thedelaysinthecriticalsituationsanalysisandinthedecisionsmakingasresponsetothem,especiallyinthecasewhenthereareprocessingtheconfidentialinformation,potentiallyresultingininadequatereactionoreventoacompletecessationofDCSfunctions[1-6].Infact,theproblemofthesupportthesustainablefunctioningoftheDCSisassociatedwithanestimateofthetimeofthesituationanalysisanddecision-makinginDCSjobscheduler[7-11].Theprobabilityofexceedingofthemaximumallowedtimeforthesituationanalysisanddecision-makingisfunctionallylinkedwiththevaluesoftheupperandlowerboundsoftheallowableperiodforthesituationanalysisanddecisionmaking[10,11].Toanalyzetheprobabilityofexceedingofthecriticaltimeofdecision-making,i.e.,thereactionofthejobschedulertotheemergencysituationsintheconditionsofdynamicchangesoftheDCSsettings,theeffectiveapproach,asitseems,istoperformtheformalizationofprocessesintheDCSjobscheduler,inparticular,onthebasisofspecialnetworkingmechanismsandmodels.Thesimulationresults,wasreceivedwiththeclassicalstochasticPetrinetworksmechanisms[12-14]fortheevaluationofcriticaltimeinthesituationsanalysisandthedecision-makingintheDCSjobschedulerhaveshownthatthesenetworksallowcorrectlyprocessonlythosecaseswhentheeventsarearrivingwiththeintensitieswhicharepracticallyequivalent.Inthecase,whentheeventsarearrivingwiththedifferentintensities,theDCScomponentwiththesmallestintensityofeventsflowactuallyisdelayingthedecisiononthemodificationoftheallDCSparameters.So,inordertoperformthefiringofthesynchronizedtransition,whichlaunchingtheprocessofsituationanalyzingandthedecisionmakingontheDCSfunctioning,thereisrequiredatokensintheallinputpositions,duetotheneedtotakeintoaccountthecurrentstateoftheallconsideredDCScomponentstoproduceacompleteandobjectiveconclusionaboutthesituationtoperformthesubsequenteffectiveanalysis.Asaresult,thissynchronizedtransitionimplementsthesynchronizationoftheprocessesofdataaccumulationfortheanalysis,andthisleadstotheabovementionedproblems.However,inpracticalapplicationsitishappenoftenenoughthataviolationoftheDCSfunctionswillbecausedbyasinglecomponentorasubsetofthecomponents.Thenumberofevents,arrivingfromthiscomponentissignificantlyexceedsthesimilarparameterincomparisontotheothercomponents.Atthesametime,themechanismsforthesustainablesupportingoftheDCSfunctioningshouldquicklyrespondtothesesituations[7-11,15-17].Thus,thereistheneedtodevelopandusethespecificnetworkingmechanismsthatwillworkcorrectlywiththedifferentialintensitiesofthearrivingeventsrelatedtotheDCScomponentsfunctions[18].2StochasticRA-NetworkfortheNodesFunctioningAnalysisintheDistributedComputerSystemsCopyright©2016MECSI.J.ComputerNetworkandInformationSecurity,2016,6,1-8II.THEMAINCONCEPTSOFTHESTOCHASTICRA-NETWORKSTHEORYTosolvethisproblem,wesuggestthemechanismofthestochasticRA-networks,whichisthemodificationoftheclassicPetri-networks[12–14],andwhichallowhandleproperlytheeventsflowwiththedifferentintensity,evenwithsubstantiallydifferent.LetusconsiderthemaindefinitionsandconceptsofthestochasticRA-networkstheory.Statement.RA-networksisatuple(S,M,L,W),where[18]:S–thestructureofRA-networks,N=(P,T,I,O),where:P={P1,P2,...,Pn}-setofplaces;T={T1,T2,...,Tm}-asetoftransitions;I={I(T1),I(T2),...,I(Tm)}-setofinputtransitions,O={O(T1),O(T2),...,O(Tm)}-setofoutputtransitions;M={M1,M2,...,Mn}-marking,Mi-numberoftokensattheplacesofRA-networks.L={λ1(T),λ2(T),...,λm(T)}–thesetofintensitiestransitionsfire;W={ω1(T),ω2(T),...,ωm(T)}-setofweightsoftheinputtransitions.Asthegraph,thestructureofRA-networksisabipartite-orientedgraph