资源描述:
《clos网络可重排算法分析》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、摘要可降低连接阻塞概率的重排算法采用的思想是使中间级交换单元的选取尽可能的集中在已经被占用的少数几个中,使空闲的中间级交换单元的利用性提高,从而降低网络的阻塞特性。本文通过程序仿真验证了此算法能使连接阻塞概率降低20%至30%,从而说明此算法对于降低阻塞概率有很大的帮助。关键词:CLOS网络,无阻塞,可重排算法,阻塞概率IIABSTRACTABSTRACTDistributionframeisthemostbasicpartofanetworkinformationsystem.Itscorefunctionmodulesisswitchingnetwork.Theperformanceo
2、ftheentirenetworkisdecidedbytheabilitytodealwithconnectionrequestsinswitchingnetwork.Ifswitchingnetworkcouldwellhandlemultipleconnectionrequests,andtherequestsarenotinconflict,thenthelossrateoftheentirenetworkwillgreatlyreduced,theperformanceoftheentirenetworkiswell.Otherwise,switchingnetworkcould
3、n’tdoitexcellent,couldn’tpromptlyandeffectivelyfindthecorrectwayfortherequest,thentheentirenetworkwillalwaysbeastateofcongestion,itsperformanceisbad.Atlastitwilldeeplyaffecttheuseofnetworkusers.Consideringtheblockingproperty,switchingnetworkcanbedividedintonon-blockingnetworkandblockingnetwork.Non
4、-blockingnetworkcanalsobedividedintothreecategories:strictlynon-blockingnetwork,wide-sensenon-blockingnetworkandrearrangeablenon-blockingnetwork.Thehardwarerequirementsinstrictlynon-blockingnetworkarerelativelyhigh,soit’sdifficulttorealizeinproject.Astowide-sensenon-blockingnetwork,thereislessrese
5、archonitatpresent.Becausetherequirementsinrearrangeablenon-blockingarelow,it’seasytorealizeit.Thekeytorearrangeablenon-blockingisitsroutingalgorithms.Andtheperformanceofblockingnetworkismeasuredbyitsblockingprobability.ThethesisselectsCLOSnetworkasitsresearchbackground,studiesvariousrearrangeablea
6、lgorithmsovertherecentyears,andproposestworearrangeablenon-blockingalgorithms:blockingandrearrangingalgorithmsandreducingblockingprobabilityrearrangeablealgorithms.Blockingandrearrangingalgorithmsimprovestheoldwaythatrerouteifexisinganumberofblockings,proposesanewmethodthatifthereisoneblocking,imm
7、ediatelyrearrangeexistedconnections.Itcutstheprocessforcallingman-madeconnections,andraisestheefficiencyofalgorithms.Moreoveritisprovedconvergenceintheory,reducingblockingprobabilityrearrangeablealgorithmsintrodu