本文選題：服務(wù)器 + SOPC　； 參考：《上海交通大學(xué)》2014年碩士論文

【摘要】：隨著大數(shù)據(jù)時(shí)代的到來(lái)，用戶(hù)對(duì)互聯(lián)網(wǎng)的高效和穩(wěn)定提出了更高的要求。內(nèi)容分發(fā)網(wǎng)絡(luò)（CDN）可以盡量避開(kāi)互聯(lián)網(wǎng)的瓶頸，通過(guò)在各地放置節(jié)點(diǎn)服務(wù)器，以代理緩存的方式減少客戶(hù)訪問(wèn)的延遲，越來(lái)越受到內(nèi)容提供商的青睞。由于內(nèi)容分發(fā)網(wǎng)絡(luò)需要大量分布在各地的節(jié)點(diǎn)服務(wù)器來(lái)提供服務(wù)，服務(wù)器能源的消耗，使其運(yùn)營(yíng)成本一直居高不下。雖然國(guó)內(nèi)外學(xué)者對(duì)內(nèi)容分發(fā)網(wǎng)絡(luò)的結(jié)構(gòu)有不少的研究，但對(duì)如何提高節(jié)點(diǎn)服務(wù)器能效比方面的研究還比較少。現(xiàn)有的內(nèi)容分發(fā)網(wǎng)絡(luò)節(jié)點(diǎn)服務(wù)器是基于通用處理器的服務(wù)器，由于硬件結(jié)構(gòu)的限制，很難在提高性能的同時(shí)降低功耗。根據(jù)內(nèi)容分發(fā)節(jié)點(diǎn)服務(wù)器的特點(diǎn)，本文提出了一種采用了主設(shè)備SOPC結(jié)構(gòu)的基于TCP/IP硬件棧的大文件亂序下載服務(wù)器架構(gòu)來(lái)提高節(jié)點(diǎn)服務(wù)器的效能，將TCP/IP協(xié)議棧的處理流程從通用處理器中分離出來(lái)，采用專(zhuān)門(mén)的硬件電路實(shí)現(xiàn)，大幅度降低功耗，并通過(guò)文件的亂序下載，提高了系統(tǒng)緩存命中率。最后使用FPGA搭建了原型機(jī)，，并與IBM通用服務(wù)器進(jìn)行了能效比的對(duì)比測(cè)試。
[Abstract]:With the arrival of big data era, users put forward higher requirements for the efficiency and stability of the Internet. Content distribution network (CDN) can avoid the bottleneck of Internet as far as possible, and reduce the delay of client access by proxy cache by placing node servers in different places, so it is more and more popular among content providers. Because the content distribution network needs a large number of node servers distributed in various places to provide services, the energy consumption of the server makes its operating cost remain high. Although scholars at home and abroad have a lot of research on the structure of content distribution network, there is little research on how to improve the efficiency ratio of node server. The existing content distribution network node server is a server based on universal processor. Due to the limitation of hardware structure, it is difficult to improve performance and reduce power consumption. According to the characteristics of the content distribution node server, this paper proposes a large file disorderly download server architecture based on TCP/IP hardware stack, which adopts the main device SOPC structure to improve the efficiency of the node server. The processing flow of the TCP/IP protocol stack is separated from the general purpose processor and realized by a special hardware circuit. The power consumption is greatly reduced and the cache hit rate is improved by downloading the files out of order. Finally, the prototype machine is built with FPGA, and compared with the IBM universal server.
【學(xué)位授予單位】：上海交通大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類(lèi)號(hào)】：TP393.05

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 王楓,羅家融;Linux下多線程Socket通訊的研究與應(yīng)用[J];計(jì)算機(jī)工程與應(yīng)用;2004年16期

2 穆榮均;;消息隊(duì)列漫談[J];程序員;2010年11期

3 彭湘凱;CDN網(wǎng)絡(luò)及其應(yīng)用[J];微計(jì)算機(jī)信息;2005年02期

4 樓樺;不同監(jiān)控環(huán)境下服務(wù)器監(jiān)控系統(tǒng)部署策略設(shè)計(jì)[J];鄭州輕工業(yè)學(xué)院學(xué)報(bào);2005年01期

5 聶秀英;;CDN業(yè)務(wù)發(fā)展現(xiàn)狀及發(fā)展趨勢(shì)[J];現(xiàn)代電信科技;2012年04期

本文編號(hào)：1860041