【摘要】：當(dāng)前互聯(lián)網(wǎng)技術(shù)發(fā)展迅速,新型網(wǎng)絡(luò)協(xié)議的不斷出現(xiàn),要求網(wǎng)絡(luò)轉(zhuǎn)發(fā)設(shè)備能夠及時(shí)提供對新協(xié)議的支持.目前,軟件定義網(wǎng)絡(luò)要兼顧可編程協(xié)議解析和數(shù)據(jù)轉(zhuǎn)發(fā)性能仍然面臨諸多困難.對此,本文提出了基于解析和執(zhí)行聯(lián)動結(jié)構(gòu)的可編程數(shù)據(jù)平面(CLIPE),通過在硬件的解析器上部署用戶可定義模塊,可實(shí)時(shí)更新硬件中解析邏輯中的協(xié)議多叉樹,從而實(shí)現(xiàn)協(xié)議解析的用戶定制性;并且,通過解析器和動作執(zhí)行器聯(lián)動的創(chuàng)新結(jié)構(gòu),減少了整個(gè)處理架構(gòu)的冗余性,從而減小動作執(zhí)行時(shí)延,提高了硬件資源利用率,與現(xiàn)有方案相比,節(jié)約了11%的邏輯資源和24%的BRAM資源.最后,本文基于NetFPGA-10G板卡完成了本方案的原型機(jī)實(shí)現(xiàn).
[Abstract]:With the rapid development of Internet technology and the emergence of new network protocols, network forwarding devices are required to provide timely support for the new protocols. At present, the software definition network still faces many difficulties to combine the performance of programmable protocol parsing and data forwarding. In this paper, a programmable data plane (CLIPE), based on parse and execution linkage structure is proposed. By deploying user-defined modules on the parser of the hardware, the protocol multitree in the parse logic can be updated in real time. In order to realize the customization of protocol parsing; In addition, through the innovative structure of the linkage of the parser and the action actuator, the redundancy of the whole processing architecture is reduced, and the delay of action execution is reduced, and the utilization rate of hardware resources is improved. It saves 11% logical resources and 24% BRAM resources. Finally, the prototype of this scheme is implemented based on NetFPGA-10G card.
【作者單位】： 國家數(shù)字交換系統(tǒng)工程技術(shù)研究中心;
【分類號】：TP393.0

【相似文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前1條

1 李健;基于OpenStack的可編程數(shù)據(jù)轉(zhuǎn)發(fā)平面構(gòu)建[D];北京郵電大學(xué);2015年

，

本文編號：2302372