本文關(guān)鍵詞：云環(huán)境下基于冷點(diǎn)虛擬機(jī)遷移的熱點(diǎn)消除問題研究　出處：《東北大學(xué)》2013年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 云計(jì)算 虛擬機(jī)遷移 熱點(diǎn)消除 冷點(diǎn) 層次分析法 多目標(biāo)決策

【摘要】：虛擬化是實(shí)現(xiàn)云環(huán)境下“計(jì)算即服務(wù)”解決方案的關(guān)鍵技術(shù)。作為虛擬化的產(chǎn)物,虛擬機(jī)具有可以靈活部署資源、實(shí)現(xiàn)對(duì)物理機(jī)狀態(tài)的隔離以及動(dòng)態(tài)遷移等特點(diǎn),在云環(huán)境下可以有效提高資源的利用率和提升數(shù)據(jù)中心的動(dòng)態(tài)資源管理能力。動(dòng)態(tài)遷移可以將虛擬機(jī)的狀態(tài)從一臺(tái)物理機(jī)轉(zhuǎn)移到另一臺(tái)物理機(jī),以此來緩解節(jié)點(diǎn)的超負(fù)載狀態(tài),進(jìn)而消除熱點(diǎn)。熱點(diǎn)是指由于資源不足無法保證服務(wù)SLA的情形,通過遷移虛擬機(jī)對(duì)數(shù)據(jù)中心進(jìn)行動(dòng)態(tài)資源調(diào)配是消除熱點(diǎn)的有效手段,也是當(dāng)今云計(jì)算的一個(gè)研究熱點(diǎn)。本文就關(guān)注虛擬環(huán)境下借助虛擬機(jī)遷移手段的熱點(diǎn)消除問題。傳統(tǒng)的熱點(diǎn)消除方法在遷移點(diǎn)的選擇問題上缺乏對(duì)熱點(diǎn)虛擬機(jī)上服務(wù)保持問題的考慮,而且缺少對(duì)熱點(diǎn)的一個(gè)系統(tǒng)化判定過程。針對(duì)該兩點(diǎn)問題,本文引入熱度和冷點(diǎn)虛擬機(jī)的概念,提出基于冷點(diǎn)虛擬機(jī)遷移的熱點(diǎn)消除策略,并特別針對(duì)熱度評(píng)估、冷熱點(diǎn)判定和冷點(diǎn)遷移點(diǎn)的選擇問題給出了基于冷點(diǎn)虛擬機(jī)遷移的熱點(diǎn)消除過程和方法。第一,對(duì)云服務(wù)質(zhì)量的相關(guān)參數(shù)指標(biāo)進(jìn)行分析,制定熱度綜合評(píng)估指標(biāo)體系,并結(jié)合該指標(biāo)體系給出一種將模糊層次分析法與多目標(biāo)決策法相結(jié)合的熱度綜合評(píng)估方法,計(jì)算得到虛擬機(jī)的熱度等級(jí)排序,進(jìn)而完成對(duì)其冷熱點(diǎn)情況的判定。第二,本文從熱點(diǎn)虛擬機(jī)服務(wù)保持的角度出發(fā),給出基于冷點(diǎn)虛擬機(jī)遷移的遷移點(diǎn)選擇決策方法,該方法綜合了虛擬機(jī)遷移代價(jià)、資源適配性和優(yōu)先等級(jí)三個(gè)目標(biāo)因素建立冷點(diǎn)虛擬機(jī)遷移點(diǎn)選擇的多目標(biāo)決策模型,為遷移點(diǎn)的選擇提供決策依據(jù)。本文借助Xen虛擬化平臺(tái)部署一個(gè)多資源混合類型應(yīng)用,實(shí)現(xiàn)對(duì)虛擬機(jī)的熱度綜合評(píng)估和熱點(diǎn)消除,驗(yàn)證本文的熱點(diǎn)消除方法。實(shí)驗(yàn)結(jié)果表明：應(yīng)用本文的熱度評(píng)估方法可以比較全面地評(píng)估虛擬機(jī)的真實(shí)熱度情況,進(jìn)而給冷、熱點(diǎn)的判定提供有力依據(jù)；本文提出的冷點(diǎn)遷移點(diǎn)選擇決策方法不僅可以綜合多方面因素選擇相對(duì)最合適的冷點(diǎn)虛擬機(jī)進(jìn)行遷移,有效消除熱點(diǎn),在該過程中熱點(diǎn)虛擬機(jī)的服務(wù)也得到了很好的運(yùn)行保持效果。
[Abstract]:Virtualization is a key technology for implementing "computing as a service" solution in a cloud environment. As a result of virtualization, virtual machines have the flexibility to deploy resources. The physical machine state isolation and dynamic migration are realized. Dynamic migration can transfer the state of virtual machine from one physical machine to another. In order to alleviate the overloading state of the node and eliminate the hot spot, the hot spot refers to the situation that the service SLA can not be guaranteed because of the lack of resources. Dynamic resource allocation of data center by migration virtual machine is an effective method to eliminate hot spots. This paper focuses on the hot spot elimination with the help of virtual machine migration in virtual environment. The traditional hot spot elimination method is lack of the hot spot virtual machine in the problem of migration point selection. Service maintenance considerations. For these two problems, this paper introduces the concepts of heat and cold point virtual machine, and proposes a hot spot elimination strategy based on cold point virtual machine migration. Especially for heat evaluation, cold hot spot decision and cold point migration point selection problem, this paper gives the hot spot elimination process and method based on cold point virtual machine migration. First, the related parameters of cloud service quality are analyzed. The thermal comprehensive evaluation index system is established, and a thermal comprehensive evaluation method which combines fuzzy analytic hierarchy process and multi-objective decision making method is presented, and the ranking of thermal grade of virtual machine is obtained. Then the cold hot spot is judged. Secondly, from the point of view of hot spot virtual machine service maintenance, this paper gives a migration point selection decision method based on cold point virtual machine migration. This method synthesizes three objective factors of virtual machine migration cost resource adaptation and priority level to establish a multi-objective decision model for cold point virtual machine migration point selection. This paper uses Xen virtualization platform to deploy a multi-resource hybrid application to realize the thermal comprehensive evaluation and hot spot elimination of virtual machine. The experimental results show that the heat evaluation method can be used to evaluate the real heat of virtual machine comprehensively and provide a powerful basis for the determination of cold and hot spots. The method proposed in this paper can not only integrate many factors to select the most suitable cold point virtual machine for migration, but also effectively eliminate hot spots. In this process, the service of hot virtual machine also gets very good performance.
【學(xué)位授予單位】：東北大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類號(hào)】：TP302
，

本文編號(hào)：1410626