本文選題：以太網(wǎng)光纖通道 + 融合網(wǎng)絡(luò)適配器�。� 參考：《電子科技大學(xué)》2013年碩士論文

【摘要】：在傳統(tǒng)數(shù)據(jù)中心中存在光纖通道網(wǎng)絡(luò)和以太網(wǎng)，光纖通道網(wǎng)絡(luò)用于提供存儲(chǔ)服務(wù)，以太網(wǎng)用于提供用戶數(shù)據(jù)服務(wù)。這兩種網(wǎng)絡(luò)的共存使數(shù)據(jù)中心的維護(hù)變得日益復(fù)雜，不符合當(dāng)今數(shù)據(jù)中心網(wǎng)絡(luò)融合的趨勢(shì)。以太網(wǎng)光纖通道(FCoE: FibreChannel over Ethernet)是一種網(wǎng)絡(luò)融合技術(shù)標(biāo)準(zhǔn)。融合網(wǎng)絡(luò)適配器（ConvergedNetwork Adapter）是一種基于FCoE技術(shù)的數(shù)據(jù)中心網(wǎng)絡(luò)接入設(shè)備。融合網(wǎng)絡(luò)適配器的使用，使得以太網(wǎng)流量和FCoE流量可以在一條物理鏈路上傳輸，從而減少服務(wù)器中網(wǎng)絡(luò)適配器和線纜連接數(shù)量，并有效降低數(shù)據(jù)中心的管理維護(hù)成本。 融合網(wǎng)絡(luò)適配器如今主要運(yùn)用于數(shù)據(jù)中心網(wǎng)絡(luò)中實(shí)現(xiàn)將服務(wù)器接入到10G以太網(wǎng)光纖通道中。國(guó)外一些大型廠商如Intel、Brocade等公司已經(jīng)生產(chǎn)出了相關(guān)產(chǎn)品，在國(guó)內(nèi)融合網(wǎng)絡(luò)適配器的研究剛剛起步，還沒有自主知識(shí)產(chǎn)權(quán)的產(chǎn)品推出。為了在自身科研環(huán)境中使用高速的10G以太網(wǎng)光纖通道，國(guó)內(nèi)某些科研機(jī)構(gòu)，如研究航電網(wǎng)絡(luò)的單位對(duì)融合網(wǎng)絡(luò)適配器產(chǎn)生了相應(yīng)的需求，在這些關(guān)鍵性領(lǐng)域需要有國(guó)內(nèi)自主知識(shí)產(chǎn)權(quán)的產(chǎn)品，能夠收發(fā)自己定制的各種幀。 基于上述需求和教研室研究現(xiàn)狀，本文主要研究了10G以太網(wǎng)光纖通道相關(guān)協(xié)議（FC，F(xiàn)CoE，F(xiàn)IP等），在此基礎(chǔ)上實(shí)現(xiàn)了融合網(wǎng)絡(luò)適配器驅(qū)動(dòng)程序，并進(jìn)行了測(cè)試。本文中的融合網(wǎng)絡(luò)適配器硬件平臺(tái)采用了Altera Strative IV530開發(fā)板，軟件平臺(tái)采用了Linux。本文對(duì)FC協(xié)議，F(xiàn)IP（FCoE Initialization Protocol）協(xié)議，，F(xiàn)CoE協(xié)議進(jìn)行了分析，編寫代碼實(shí)現(xiàn)了各個(gè)協(xié)議模塊，在此基礎(chǔ)上實(shí)現(xiàn)了驅(qū)動(dòng)程序。整個(gè)驅(qū)動(dòng)程序主要包括FIP模塊、DCBx模塊、登錄模塊和SCSI-FCP模塊。其中FIP模塊完成融合網(wǎng)絡(luò)適配器登錄到交換機(jī)的過(guò)程，DCBx模塊完成與FCoE交換機(jī)協(xié)商流量控制等參數(shù)，登錄模塊實(shí)現(xiàn)融合網(wǎng)絡(luò)適配器通過(guò)FCoE交換機(jī)登錄到遠(yuǎn)端磁盤陣列，SCSI-FCP模塊完成將遠(yuǎn)端磁盤陣列接入到Linux內(nèi)核SCSI子系統(tǒng)。本文對(duì)各個(gè)模塊進(jìn)行了測(cè)試，F(xiàn)IP模塊實(shí)現(xiàn)了FCoE交換機(jī)的登陸，DCBx模塊實(shí)現(xiàn)了與交換機(jī)進(jìn)行參數(shù)協(xié)商，登錄模塊實(shí)現(xiàn)了磁盤陣列登陸，SCSI-FCP模塊實(shí)現(xiàn)了遠(yuǎn)端磁盤陣列接入到Linux內(nèi)核SCSI子系統(tǒng)。
[Abstract]:There are fiber channel networks and Ethernet in traditional data centers. Optical fiber channel networks are used to provide storage services and Ethernet is used to provide user data services. The coexistence of these two networks makes the maintenance of data center become more and more complicated, which does not conform to the trend of data center network convergence. FibreChannel over Ethernet is a network convergence technology standard. Converged Network Adapter is a data center network access device based on FCoE technology. By integrating the use of network adapters, Ethernet traffic and FCoE traffic can be transmitted over a physical link, thus reducing the number of network adapters and cable connections in the server, and effectively reducing the management and maintenance costs of the data center. Converged network adapters are now mainly used in data center networks to connect servers to 10G Ethernet fiber channels. Some large foreign manufacturers, such as Intel Brocade, have already produced related products. The research on converged network adapters in China has just started, and no independent intellectual property products have been launched. In order to use high speed 10G Ethernet fiber channel in their own scientific research environment, some domestic scientific research institutions, such as avionics network research units, have a corresponding demand for converged network adapters. In these critical areas there is a need for domestically owned intellectual property products capable of sending and receiving their own customized frames. Based on the above requirements and the current research situation of the teaching and research department, this paper mainly studies the 10G Ethernet fiber channel related protocol, such as FCU FCoEN FIP and so on. On this basis, the integrated network adapter driver is implemented and tested. In this paper, the hardware platform of the fusion network adapter adopts Altera Strative IV530 development board, and the software platform adopts Linux. In this paper, the FCoE Initialization protocol of FC protocol is analyzed, and the code is written to realize each protocol module. On this basis, the driver is implemented. The driver includes FIP module, login module and SCSI-FCP module. Among them, the FIP module completes the process of integrating the network adapter to log on to the switch, and the DCBx module completes the negotiation with the FCoE switch on parameters such as flow control, etc. The login module realizes the integration of the network adapter to the remote disk array SCSI-FCP module through the FCoE switch to complete the remote disk array access to the Linux kernel SCSI subsystem. In this paper, the FIP module of each module is tested, the FCoE switch login module is implemented and the parameters are negotiated with the switch. The login module implements the disk array landing pad SCSI-FCP module and implements the remote disk array access to the Linux kernel SCSI subsystem.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類號(hào)】：TN929.11

【參考文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前2條

1 徐靜;基于iSCSI磁盤陣列在線擴(kuò)容的研究[D];武漢理工大學(xué);2009年

2 陳航;基于Linux的無(wú)線通信網(wǎng)關(guān)的設(shè)計(jì)與實(shí)現(xiàn)[D];北京郵電大學(xué);2010年

本文編號(hào)：1800722