计算机工程
計算機工程
계산궤공정
COMPUTER ENGINEERING
2014年
2期
284-288
,共5页
戴薇薇%奚宏生%曾威龙
戴薇薇%奚宏生%曾威龍
대미미%해굉생%증위룡
资源池化%KAD 网络%租借调度%马尔科夫链预测法%三网融合%视频业务网
資源池化%KAD 網絡%租藉調度%馬爾科伕鏈預測法%三網融閤%視頻業務網
자원지화%KAD 망락%조차조도%마이과부련예측법%삼망융합%시빈업무망
resource pooling%KAD network%loan scheduling%Markov chain prediction method%tri-networks integration%video service network
对现有 KAD 协议在高并发、高实时性需求环境中的优缺点进行分析。为减少系统信息的冗余和延迟时间,提出一种在三网融合背景下,基于服务节点资源池划分的改进 KAD 网络视频业务系统的调度管理策略。采用基于网络连通性的准则将视频业务网的服务节点划分为若干资源池,各池内部网络采用 KAD 方式进行组织。根据用户请求量的预测,采用池间资源租借的方式解决单个资源池由于业务量增大而导致的服务过载问题。实验结果表明,对于典型的视频业务请求,马尔科夫链预测法能够实现过载漏报率为0的精确预警,从而根据预测结果,并利用 KAD 网络灵活的扩展性及时地补充服务资源。
對現有 KAD 協議在高併髮、高實時性需求環境中的優缺點進行分析。為減少繫統信息的冗餘和延遲時間,提齣一種在三網融閤揹景下,基于服務節點資源池劃分的改進 KAD 網絡視頻業務繫統的調度管理策略。採用基于網絡連通性的準則將視頻業務網的服務節點劃分為若榦資源池,各池內部網絡採用 KAD 方式進行組織。根據用戶請求量的預測,採用池間資源租藉的方式解決單箇資源池由于業務量增大而導緻的服務過載問題。實驗結果錶明,對于典型的視頻業務請求,馬爾科伕鏈預測法能夠實現過載漏報率為0的精確預警,從而根據預測結果,併利用 KAD 網絡靈活的擴展性及時地補充服務資源。
대현유 KAD 협의재고병발、고실시성수구배경중적우결점진행분석。위감소계통신식적용여화연지시간,제출일충재삼망융합배경하,기우복무절점자원지화분적개진 KAD 망락시빈업무계통적조도관리책략。채용기우망락련통성적준칙장시빈업무망적복무절점화분위약간자원지,각지내부망락채용 KAD 방식진행조직。근거용호청구량적예측,채용지간자원조차적방식해결단개자원지유우업무량증대이도치적복무과재문제。실험결과표명,대우전형적시빈업무청구,마이과부련예측법능구실현과재루보솔위0적정학예경,종이근거예측결과,병이용 KAD 망락령활적확전성급시지보충복무자원。
In order to meet the demands of the network management and control of video service in the tri-networks integration background, the advantages and disadvantages of using KAD protocol in high concurrent and real-time environment are analyzed and a scheduling management strategy based on loaning video service nodes between resource pools is proposed to minimize redundancy information and delay time of system. Divide the video service nodes into a number of resource pools based on the connectivity between video service node and different networks and each pool is organized by KAD protocol. Adopt a method of loaning available video service nodes from other pools to avoid overload or denial of service when a large number of video requests are predicted. Experimental result shows that the Markov chain prediction method can warn the overload state effectively and false negative rate achieves zero. The resource of video service nodes can be balanced based on the mechanism.