本文選題：P2P　切入點(diǎn)：失效檢測(cè)　出處：《哈爾濱工業(yè)大學(xué)》2015年碩士論文

【摘要】：對(duì)等網(wǎng)絡(luò)(peer-to-peer,簡(jiǎn)稱P2P),是一種沒(méi)有中央服務(wù)器控制和層次管理的分布式網(wǎng)絡(luò),節(jié)點(diǎn)隨時(shí)會(huì)不加通知就離開網(wǎng)絡(luò),必須使用失效檢測(cè)技術(shù)周期性地探測(cè)鄰居節(jié)點(diǎn)的狀態(tài),快速地發(fā)現(xiàn)失效節(jié)點(diǎn),從而維護(hù)鄰居路由表以及整個(gè)網(wǎng)絡(luò)的拓?fù)涞恼_性。失效檢測(cè)技術(shù)給對(duì)等網(wǎng)絡(luò)節(jié)點(diǎn)的鄰居表維護(hù)提供有力的支持,但P2P節(jié)點(diǎn)的檢測(cè)關(guān)系動(dòng)態(tài)性強(qiáng),鏈路關(guān)系復(fù)雜多變,導(dǎo)致許多適合靜態(tài)檢測(cè)關(guān)系的檢測(cè)算法并不適合用到P2P網(wǎng)絡(luò)中。本文提出了一個(gè)拉模式的自適應(yīng)于Qo S檢測(cè)需求的失效檢測(cè)算法,算法運(yùn)行并不需要對(duì)鏈路傳輸狀況有任何提前的假設(shè),非常適用于P2P網(wǎng)絡(luò)。目前P2P失效檢測(cè)的算法都是建立單一的檢測(cè)關(guān)系,同一個(gè)節(jié)點(diǎn)會(huì)被多個(gè)節(jié)點(diǎn)同時(shí)檢測(cè),存在大量的重復(fù)檢測(cè)關(guān)系,缺乏檢測(cè)結(jié)果的共享機(jī)制。本文從整個(gè)失效檢測(cè)服務(wù)的架構(gòu)出發(fā),研究如何將結(jié)果共享機(jī)制算法sed_FD與自適應(yīng)失效檢測(cè)器相結(jié)合,從而減少大量的重復(fù)檢測(cè)關(guān)系,有效降低檢測(cè)負(fù)載,并能識(shí)別節(jié)點(diǎn)失效以及鏈路失效,最終提高檢測(cè)的準(zhǔn)確性。本文最后著力研究對(duì)等網(wǎng)絡(luò)中失效檢測(cè)實(shí)現(xiàn)所面臨的關(guān)鍵問(wèn)題,結(jié)合層次式檢測(cè)框架、結(jié)果共享算法、自適應(yīng)失效檢測(cè)算法實(shí)現(xiàn)了一個(gè)獨(dú)立于P2P應(yīng)用的失效檢測(cè)服務(wù)系統(tǒng),利用該系統(tǒng),P2P網(wǎng)絡(luò)應(yīng)用可以輕易地對(duì)感興趣的節(jié)點(diǎn)發(fā)起失效檢測(cè),提升P2P應(yīng)用在失效檢測(cè)方面的速度、精度以及降低檢測(cè)的負(fù)載。
[Abstract]:Peer-to-Peer (P2PN) is a distributed network with no central server control and hierarchical management. Nodes will leave the network without notice at any time. The failure detection technology must be used to periodically detect the status of neighbor nodes. In order to maintain the correctness of the neighbor routing table and the topology of the whole network, the failure detection technology provides the powerful support for the maintenance of the neighbor table of the peer-to-peer network, but the detection relationship of the P2P node is very dynamic. Because of the complex and changeable link relationship, many detection algorithms suitable for static detection are not suitable for P2P networks. In this paper, a pull-mode fault detection algorithm adaptive to QoS detection requirements is proposed. The algorithm does not need to have any presupposition about the link transmission condition, so it is very suitable for P2P network. At present, P2P failure detection algorithms establish a single detection relationship, the same node will be detected by multiple nodes at the same time. There are a lot of repeated detection relationships and there is no shared mechanism of detection results. This paper studies how to combine the result sharing mechanism (sed_FD) algorithm with adaptive failure detection algorithm (sed_FD) from the architecture of the whole failure detection service. In order to reduce a large number of repeated detection relationships, effectively reduce the detection load, and can identify node failures and link failures, and ultimately improve the accuracy of detection. Finally, this paper focuses on the key problems faced by the realization of failure detection in peer-to-peer networks. Combined with hierarchical detection framework, result sharing algorithm and adaptive failure detection algorithm, a failure detection service system independent of P2P applications is implemented. Using this system, P2P network applications can easily initiate failure detection to interested nodes. Improve the speed and accuracy of P2P application in failure detection, and reduce the load of detection.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TP393.02

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 張宇翔;楊冬;張宏科;;P2P網(wǎng)絡(luò)中Churn問(wèn)題研究[J];軟件學(xué)報(bào);2009年05期

，

本文編號(hào)：1685589