200821852 九、發明說明： 【發明所屬之技術領域】 本發明係關於-種網路儲存管理裝置與方法，尤指雙通道的網路 儲存管理裝置與方法。 【先前技術】 隨著網路體速度日触速、錄，使得在晴讀輸數位影音 也變得絲越普及’又加上數姆彡音本身雜別、斜分龐^ 而同時帶動了數位儲存媒體大型化的發展趨勢。 隨著數位舰大魏_勢，單—硬碟，甚至是單—硬碟插都不 足以應付這種大量儲存需求，而發展出AqE(ata 與iSCSI這兩種特殊的通訊協定。 ATA over Ethernet (AoE)是由 Br贈ley c〇ile 所提創的—種網 路通訊協^ ’且此齡可以在乙太網路上存取愚標準的硬碟褒 置’進而達到能以平價且鮮的技絲實現—麵存區域網路環 境。 Α〇Ε不倚賴乙太網路中網路層以上的協定，包括ιρ、卿，等 都棄捨不用。這麵AqE不能麵在區域網路上進行路由、繞徑， 所以AoE確實僅打算做為儲存區域網路之用。 需要注意的是，AoE與iSCSI均_自身是遠比光纖通道（Fibre 200821852200821852 IX. Description of the Invention: [Technical Field] The present invention relates to a network storage management apparatus and method, and more particularly to a dual channel network storage management apparatus and method. [Prior Art] With the speed of the network body, the speed of recording, so that the digital audio and video in the clear reading has become more and more popular, and the number of voices of the voice itself is miscellaneous, obliquely divided and simultaneously led to the digital The trend of large-scale storage media. With the digital ship, the single-hard disk, even the single-hard disk plug is not enough to cope with this large storage demand, and developed AqE (ata and iSCSI two special communication protocols. ATA over Ethernet (AoE) is a kind of network communication protocol created by Br ley c〇ile. And this age can access the standard hard disk device on the Ethernet network, and then it can be cheap and fresh. Technology implementation - the local area network environment. Α〇Ε Do not rely on the network layer above the Ethernet network, including ιρ, qing, etc. are abandoned. This AqE can not be routed on the regional network A, the AoE is really only intended to be used as a storage area network. It should be noted that both AoE and iSCSI are much larger than Fibre Channel (Fibre 200821852).
Channel)低廉的儲存區域網路佈建方案，不過A〇E標榜比幻 更ώ單、更低廉，單就標準規範的技術文件而言，A〇E規格僅有8 頁，而iSCSI卻為257頁。 •在支板的硬體方面，目前已有c〇raid公司推出撼用的硬碟樞， •被歉為EthedMve。錄财自，s_eFQrge·贼所提供的 ’ vblade，能使安裝有vblade的Unux伺服器，會透過網路去尋找Channel) Low-cost storage area network deployment solution, but A〇E is more succinct and cheaper than the illusion. For standard technical documents, the A〇E specification is only 8 pages, while the iSCSI is 257. page. • In terms of the hardware of the support board, c〇raid has already launched a hard disk hub, and is apologized to EtedMve. Recording money, s_eFQrge·thief provides ’ vblade, which enables Unux server with vblade to be searched through the Internet.
AoE硬碟樞’並在找尋到後便可向硬碟概出儲存服務的請求，進 而使用硬碟櫃内的儲存空間、資源。vblade軟體有兩種型態與執 行法，-較贿赌郎（userspaee)的執行程式（屬&⑽灿 工具程式包的-部份）來運作，另—種是以匕丨麵的核心模組程 式（kernel module)來運作。 雖然AoE是簡單的網路協定，但在儲存的功效機制運作上卻也增 高了複雜性，因此底下針對A〇E的細節作進一部說明。 在AoE協定中，主要採用了區塊性的儲存傳輸。在這種情況下， ΑΤΑ在AoE中被當成一種硬碟用的線路協定，資料寫入硬碟後若要 •進行讀出，則會以一個或一個以上的塊狀型態進行回傳，且每塊 ，的谷里尺寸大小（Block Size)固定一致，此可稱為傳輸區塊 (Block) 〇 接著，Α〇Ε^定直接、簡單地將ΑΤΑ指令（ATA command)送入低 I1白、’周路封包内，使知乙太網路線cawe)能有效取代 ΑΤΑ介面的接線（或稱··排線，ribb〇n cable)，並可以在乙太網 6 200821852 路所及的各端點(例如鄕（硬碟櫃）或賴）進行㈣ 運與傳輸。 對準中的ata排線而言，它不用去理會自身所傳輸的是何 種貝料區塊’讀ΑοΕ標準而言也是相同，Α〇Ε標準所用的乙太網 路線不用去瞭解它所傳輸的是何师料區塊，用戶可以讀出、寫 入任何想要祕塊該，科過在錄叙下m統將比傳 輸區塊更適合絲進行資料的組織與結構化運用。 就傳統而§ ’制者是輯耗統的方式來使用電腦巾的硬碟， 如ext3、XFS、以及NTFS等槽案系統皆是以此種想定所設計、運 用」然而在使用AoE時，由於A〇E運用乙太網路來取代ata排線， 這就村能破壞此種想定用法，改以AqE方式運作對傳統樓案系 統而言是-件危險的事，有可能導致播案系統的錯亂，或造成作 業系統的核心錯愣（kernei panic)。 對此AoE疋用叢集式的檔案系統來避免這些問題，使硬碟、 ㈣櫃的區塊性儲存、傳輸的裝置設備，能夠同時供多部電腦存 取使用，跳脫過往ΑΤΑ技術只能讓硬碟、硬碟群專供單一部電腦 存取使__。若就實際的功效運作來說，能夠存取AqE硬碟 櫃的飼服ϋ自身就如同是加人叢集系統的—部叢集電腦，是以叢 集的共享機制來共同取用儲存資源，而Α〇Ε硬碟櫃即是一個可供 共享的區塊性儲存設備。GFS、0CFS2等則屬常見的具叢集功效的 檔案系統。 200821852 請參閱第丨圖，第〗圖為習知儲存區域網路環境之示意圖。如第i 圖所示’在習知儲麵酬路環境中，E量硬碟群主要由複數個 硬碟櫃10a〜l〇n所組成，每一硬碟櫃1〇a〜施另由複數個硬碟抽 屜11a、lib...、Ilf、ilg…所組成，而每一硬碟抽屜山、仙…、 nf、ug...t則又包含了複數個硬碟裝置12a、12b...、i2i、i2g...。 若電腦裝置m、17b...、17g〜17n需要存取上述硬碟裝置I2a、 12b...、12f、12g··.時，可採用械與iscsi協定。 在Aoe協定中，城封包是用乙太網路中的聽他來決定封包從 何處發出以及要傳往何處。不過聽位址只能在單-的乙太網路 中使用MAC健是糊域絲.進行廣赋發送。現有的 =太網路硬體多具有流量控制（細咖㈣）的功效機制，能 。使封包飾發送（触㈡如職―）_寬需求降至最 氏。乙太網路細練冗余校驗⑽）演算絲確保封 達。 只是’在習知儲存區域網路環境中，每—台電職置m 其中—個硬碟/光碟連接埠(―般的主機板會 、碟/先碟連接埠），來存取複數個硬碟裝置12a、12b... 般硬:二其!之一’而另一個硬碟/光碟連接埠只能直接裝設- 讓主機极上。、了充分糊這種網路儲存系統的好處，最好能夠 兩個硬碟/光碟連接土阜，都能支援此種網路館存系統。 200821852 【發明内容】 嫩置與方 均得以透過鱗通道存取複數兩個硬碟/光碟連接痒， 〖存管理裝置與方法中， 基於上述目的’在本翻雙通道的網路像 主要利用緩衝記憶單元來暫 置端硬碟介咐軸連接至 =^裝置端硬碟介面、第二裝 後再由控職物轉==料取要求，然 路介面把城私麵要求 料排取縣’而透過網 ψ - , L ^ 專遞至相對於存取要求的硬碟裝 ’猎貫現主機端和網路錯存管理襄置之間的雙通道，而 對主機端提供兩台可供使用的硬碟穿置。 " 【實施方式】 =^第㈣’第2圖為本發明儲存區域網路環境之示意圖。如 ⑥所不’在本發·存區域境巾，巨量硬 由複數個硬碟櫃1〇3〜1〇11所4 &卜 〃要仍 個硬碟抽屜Ua、.·、^、】 是數 n g所、、且成，而母一硬碟抽屜lla、After the AoE hard disk hub is found, it can request a storage service from the hard disk, and then use the storage space and resources in the hard disk cabinet. The vblade software has two types and execution methods, the userspaee executable (which is part of the & (10) Can Tool Package), and the other is the core module. The kernel module works. Although AoE is a simple network protocol, it also increases the complexity of the storage mechanism. Therefore, the details of A〇E are explained below. In the AoE agreement, block storage storage is mainly used. In this case, ΑΤΑ is used as a line protocol for hard disk in AoE. If data is written to the hard disk and then read, it will be returned in one or more block types. Each block has a fixed block size (Block Size). This can be called a block. Then, you can directly and simply send the ATA command to the low I1 white. 'In the road package, the channel is known to be able to effectively replace the wiring of the interface (or ribb〇n cable), and can be used in the end points of the Ethernet 6 200821852 road. (eg 鄕 (hard disk cabinet) or Lai) (4) transport and transmission. In terms of the aligned ata cable, it does not need to pay attention to what kind of bedding block it transmits. The reading is also the same. The standard Ethernet route used by the standard does not need to know the transmission. It is the division of the teacher, the user can read and write any secret block, and the division will be more suitable for the organization and structure of the data than the transmission block. As for the traditional and § 'system is a hard-to-use way to use computer hard disk, such as ext3, XFS, and NTFS are designed and used in this way." However, when using AoE, due to A〇E uses Ethernet to replace the ata cable. This can destroy the intended usage of the village. It is dangerous to operate the AqE system for the traditional building system, which may lead to the broadcast system. Disorder, or cause the core of the operating system (kernei panic). In this regard, AoE uses a cluster file system to avoid these problems, so that the hard disk, (four) cabinet block storage, transmission device and device can be used for multiple computers at the same time, jumping off the past technology can only let Hard disk and hard disk groups are reserved for single computer access to make __. In terms of actual function operation, the feeding service that can access the AqE hard disk cabinet itself is like a cluster computer that adds a human cluster system, and uses a cluster sharing mechanism to jointly access storage resources. The hard disk cabinet is a block storage device that can be shared. GFS, 0CFS2, etc. are common file systems with clustering functions. 200821852 Please refer to the figure, which is a schematic diagram of the conventional storage area network environment. As shown in Figure i, in the traditional storage environment, the E-hard disk group is mainly composed of a plurality of hard disk cabinets 10a~l〇n, each hard disk cabinet 1〇a~ A hard disk drawer 11a, lib..., Ilf, ilg..., and each hard disk drawer, s..., nf, ug...t further includes a plurality of hard disk devices 12a, 12b.. ., i2i, i2g... If the computer devices m, 17b, ..., 17g to 17n need to access the hard disk devices I2a, 12b, ..., 12f, 12g, ..., the device and the iscsi protocol can be used. In the Aoe agreement, the city packet is used by the listener in the Ethernet to determine where the packet is sent and where it is going. However, the listening address can only be used in the single-ether network. The existing = too network hardware has more traffic control (fine coffee (four)) efficiency mechanism, can. Let the package send (touch (2) as the job -) _ wide demand to the highest. Ethernet fine-tuning redundancy check (10)) The calculation of the wire ensures the sealing. Just "in the traditional storage area network environment, each - Taiwan power position m - a hard disk / CD connection ― ("like motherboard board, disc / first disc connection 埠" to access a plurality of hard drives The devices 12a, 12b... are as hard as: one of them! And the other hard disk/disc port can only be mounted directly - to the host. With the benefits of this network storage system, it is best to have two hard drives/discs connected to the bandit to support such a network storage system. 200821852 [Summary of the Invention] Both the tender and the square are able to access the two hard disk/disc connections through the scale channel, and in the storage management device and method, based on the above purpose, the double-channel network image is mainly used for buffering. The memory unit is connected to the hard disk interface of the temporary hard disk to the =^ device hard disk interface, and after the second device is replaced by the control object == fetching request, and the road interface selects the private material of the city to the county. And through the network -, L ^ is delivered to the dual-channel between the host and the network fault management device with respect to the hard disk of the access request, and two available for the host side. Hard disk wear. < [Embodiment] =^ (4)' Figure 2 is a schematic diagram of the storage area network environment of the present invention. If the 6 are not in the hair of the hair area, a large number of hard disk cabinets 1 〇 3 ~ 1 〇 11 4 & 〃 〃 still still a hard disk drawer Ua, .., ^, 】 It is a number of ng, and it is made, and the mother has a hard disk drawer lla,
Ilf llg···中則又包含了複數個硬碟裝置12a、i2b···、 2卜糊裝置17心咖..，，需要存取上述硬碟裳置 9 200821852 m、12f、12g.,，主要是透過各自所屬的雙 路儲存管理裝置16a〜16η來進行。 、、的網 舉例來說，若電《置⑽卩羯端）要透過雙通道 3置^，經由網路通道來存取硬碟裝置， =管理裝置版就會需要特殊的元件來達成。需的^館 細裝置Ha、網路儲存管理裝置16a有著相同的運作 2參閱第㈣’第3 _本發明雙通道的網路儲存管理裝置之_ ::。如第3圖所示，本發明雙通道的網路儲存管理襄置心 硬f b)、網路介面28、緩衝記鮮元％ ^ 成。電腦裝置na(尤其是其·主機細至少 域戶^且 碟介面20a、第二主播媸絲说入 弟主機^硬 衝谏單1二 面2〇b、CPU(未描纷)等元件。緩 端硬猶記憶體組成(共崎存來自第一褒置 缓衝a 裝置端硬碟介面22b的存取要求）或由兩個 =雜成(分別暫存來自第一裝置端硬 ： 置端硬碟介面22b的存取要求）。 弟一裝 蹄管理裝置料，其所屬的裝置端硬碟介面組 弟裝置端硬碟介面22a、第二裝置端硬人 $ 接糊峨 機端硬碟介面咖，並可藉此接收來自第-主機端硬碟介面 200821852 20a、第二主機端硬碟介面的主機端之存取要求。這個存取要 求係針對該些台硬碟裳置12a、12b...、⑵、收…中之—（例如. 已經配置給電置17a顧的硬縣置⑵、邮要求提供所需 資料。 簡單來說，本發明雙通道的網路儲存管理裝置16a，主要利用緩衝 記憶單元26來暫存來自第一裝置端硬碟介面22a、第二褒置端硬 碟介面22b(分別電性連接至主機端)的主機端之存取要求 由控制器24從緩衝記憶單元26依序取得這些存取要求爾再 網路介面28把主機端之存取晷卡指、备 冉透過 “ 縣’傳拉姆於魏要求的硬碟 :門的"、g，错此實現電腦裝置17场網路儲存管理裳置恤 裝：二::而如_職置17a提供兩台可供使用的硬碟 在這種架構下，使用者不會察營到 # θ ^ “ 見财所使_硬碟裝置m、i2g， ，、只疋在w的遠端處，而非裝置在電腦機殼中。芙於 路儲存媒體架構，萬—其所使用的硬縣。二，、冯 外配置，供使用的如心;=置 了達到使財嫩蝴物使_物㈣已經損壞 舉例來說，若電職置i7a的使用者將第— 第二主機端硬碟介面施所連接的硬碟裝 ^面20a、 C槽、D槽磁碟機，這其中主要是透過雙2分別定義成 逍道的網路儲存管理裝置 11 200821852 =來達_，並邱定G師靖置 〇槽的硬碟裝置12_分硬碟裝㈣中的:二心 =要將=裝置12g所備份的資料還原回硬碟裝置/ 先攸弟-主機端硬碟介面2〇b發出讀取硬碟裝 : 儲存管理裝置碟介面22b傳入網路 理其他事情，為避免絲，控制可能同時在處 凡26中’縫概__心鱗娜域敵存7 再來透過網路介面28把駐機 、 要求的硬碟裝置取。 存取要求，傳遞至相對於存取 112gM_#取要Μ ’便會開始把所需 道回傳至網路儲存管理裝置恤，同樣地控 元=4妙也會將回應存取要求所提供的資料，先暫存至緩衝記憶單 =後再缺衝記憶單元26依序取得回應存取要求所提供的 貝抖’透過網路介面28傳遞至屬於c槽的硬碟裝置12 «裝置17a倾整個㈣傳輸情況，還是需要透過裝置端硬碟 二面組(特別是相對於C槽的第一裝置端硬碟介面22a、第—主機 &硬碟介面20a) ’向電腦裝置17a回報資料傳輸情況，以便作到 如同使用-般硬碟裳置。不過，在上述資料傳遞手段中，硬碟枣 =⑵、12g彼此也能直接互傳資料’不需經由網路儲存管理農置 12 200821852 請參=4圖，第4圖為本發明雙通道的網路館存管理裝置之另 恤二口。如弟4圖所示，本發明雙通道的網路儲存管理裝置 a * 了包含裝置端硬碟介面組(第—裝置端 、一 裝置端硬磾介而99M λ 蜂"面22a、弟二 24外，、= 網路介面28、緩衝記憶單元26、控制哭 确外具村魏連敍硬碟技2 面20c。 罘一主機蝠硬碟介 在本發明雙通道的網路儲存朽里裝置16a中，額外 端硬碟介面2〇c之後，控制器24可依據主機端之^ 機 述所傳輸的資料也經由第三主機端硬碟介面子上 烈，或是從硬碟裝置23讀出資料，再傳入其他靖^裝置 此外，控制㈣可透過網路通道，向如第2 _示之管縣龍 取得該些台硬縣置的舰存取控難_ Aceess㈤㈣，_ 立址’而可猎著網路介面28透過網路管道使網路儲存管理裝置 16a和該些台硬碟裝置互相傳遞資料封包。 為了讓電月®裝置17a能夠支援非標準型態下使用的硬碟裝置 12t> i2f、丨竑…’控制器24必須有特殊的設計。 請參閱第5圖’第5圖為本發明控制器之示意圖。如第5圖所示， 本發明控制器24進-步包含控制權子單元施、啟動(b〇〇t)子單 το 24b。控制權子單元24a會在使用者啟動主機端時，自動地對主 機端提供在邏輯區塊定址(L〇gical B1〇ck Addressing，_模式 下編號為G的第零個邏輯區塊⑽Q)，而暫不提供含在硬碟裝置 13 200821852 12f(安裝有作業系統者)的第零個邏輯區塊( G)，然後在使用 者選擇啟動(boot)儲存在特定硬碟裝置幅特定作㈣統之後， 由啟動(boot)子單元24b經由網路管道取得含有被選定的作業系 統的相對的第零個邏輯區塊(LBA 0)，再經由控制權子單元24a回 傳給主機端的中央處理器（CPU)，而進行啟動(b〇〇t)被選定的作業 系統。 藉由以上較挂具Μ實施例之詳述，係希望能更加清楚描述本發明. 之特徵與精神，而並非以上述所揭露的較佳具體實施例來對本發 明之範彆加以限制。相反地，其目的是希望能涵蓋各種改變及具 相等性的安排於本發明所欲申請之專利範圍的範膏内。 【圖式簡單說明】 第1圖為習知儲存區域網路環境之示意圖。 第2圖為本發明儲存區域網路環境之示意圖。 苐3圖為本發明雙通道的網路儲存管理裝置之示 意圖。 第4圖為本發明雙通道的網路儲存管理裝置之另 一示意圖。 第5圖為本發明控制器之示意圖。 14 200821852 【主要元件符號說明】 10a〜10η硬碟櫃 11a、lib…、Ilf、llg···硬碟抽屜 12a、12b···、12f、12g···硬碟裝置 17a、17b···、17g〜17n 電腦裝置 16a、16b···、16g〜16η網路儲存管理裝置 14管理裝置 20a第一主機端硬碟介面 20b第二主機端硬碟介面 20c第三主機端硬碟介面 22a第一裝置端硬碟介面 22b第一裝置端硬碟介面 24控制器 24a控制權子單元 24b啟動子單元 26緩衝記憶體單元 28網路介面 15Ilf llg··· contains a plurality of hard disk devices 12a, i2b, ..., 2 paste device 17 heart coffee.., need to access the above-mentioned hard disk skirt 9 200821852 m, 12f, 12g. It is mainly performed by the respective two-way storage management devices 16a to 16n to which they belong. For example, if the "Set (10) terminal) is to access the hard disk device via the network channel through the dual channel 3, the management device version will require special components to achieve. The required device, the device Ha, and the network storage management device 16a have the same operation. 2 Refer to (4) '3rd _ the two-channel network storage management device of the present invention _ ::. As shown in Fig. 3, the dual-channel network storage management of the present invention is hard-wired, network interface 28, and buffered. The computer device na (especially its host is at least the domain of the domain ^ and the disk interface 20a, the second anchor 媸 silk said to enter the brother host ^ hard rushing single 1 two sides 2 〇 b, CPU (not drawn) and other components. The end hard memory is composed of memory (the total resilience from the first device buffer a device end hard disk interface 22b access request) or by two = miscellaneous (scratchly from the first device side hard: fixed end hard The access requirement of the disc interface 22b). The device is a device for managing the device, and the device-side hard disk interface of the device is connected to the hard disk interface 22a of the device device, and the hard disk device of the second device is connected to the hard disk interface. And can receive the access request from the host side of the first host hard disk interface 200821852 20a and the second host hard disk interface. The access request is for the hard disks 12a, 12b.. (2), receiving... (for example, the hard county (2) that has been allocated to the electrical 17a, the postal request provides the required information. Briefly, the dual-channel network storage management device 16a of the present invention mainly utilizes buffering. The memory unit 26 temporarily stores the hard disk interface from the first device side hard disk interface 22a and the second device side The access requirements of the host end of the 22b (electrically connected to the host end respectively) are obtained by the controller 24 from the buffer memory unit 26 in sequence. The re-network interface 28 is used to access the host card.冉 “ “ “ “ “ “ “ 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏The available hard disk in this architecture, the user will not see the camp to # θ ^ " see the financial office _ hard disk device m, i2g, ,, only at the far end of w, not the device In the computer case. Fu Yu Road storage media architecture, 10,000 - the hard county used. Second, Feng outside configuration, for the use of the heart; = set up to make the financial tenders make _ things (four) has been damaged For example, if the user of the i7a is connected to the hard disk of the first host-side hard disk interface, the 20-, C-, and D-slot drives are mainly defined by the double 2 Chengyudao's network storage management device 11 200821852 = Laida _, and Qiu Ding G Shijing's hard disk device 12_ hard disk mounted (four): two Heart = to restore the data backed up by device 12g back to the hard disk device / first brother - host hard disk interface 2 〇 b issued read hard disk: storage management device disk interface 22b incoming network for other things, In order to avoid the wire, the control may be taken at the same time in the 26th 'Sleeve __ Heart Scalena domain enemy 7 and then through the network interface 28 to take the host, the required hard disk device. Access requirements, passed to the relative Access 112gM_#取ΜΜ will start to transfer the required route back to the network storage management device, and the same control element = 4 will also respond to the information provided by the access request, temporarily stored in the buffer memory. Single = after the memory unit 26 sequentially obtains the echo request provided by the access request. The transmission is transmitted through the network interface 28 to the hard disk device 12 belonging to the c slot. The device 17a dumps the entire (four) transmission condition, or needs to pass through the device. The hard disk two-sided group (especially the first device-side hard disk interface 22a, the first host & the hard disk interface 20a with respect to the C-slot) reports the data transmission to the computer device 17a so as to be used as if The hard disk is on the shelf. However, in the above data transmission means, the hard disk jujube = (2), 12g can also directly transfer data to each other 'do not need to manage the farm through the network storage 12 200821852 Please refer to the 4 map, the fourth figure is the dual channel of the present invention A second edition of the online library management device. As shown in Figure 4, the dual-channel network storage management device a* of the present invention includes a device-side hard disk interface group (the first device side, the device device hard interface, and the 99M λ bee " face 22a, the second brother 24, , = network interface 28, buffer memory unit 26, control crying outside the village Wei Lianxu hard disk technology 2 surface 20c. A host bat hard disk is introduced in the dual channel network storage device 16a of the present invention, After the extra hard disk interface 2〇c, the controller 24 can also read the data according to the data transmitted by the host computer through the third host hard disk interface, or read data from the hard disk device 23, and then In addition to the other Jing device, in addition, the control (4) can access the hardships of the hardships of the Taiwanese county, such as the 2nd _ Zhizhi County, _ Aceess (5) (4), _ The network interface 28 causes the network storage management device 16a and the hard disk devices to transmit data packets to each other through the network pipe. In order to enable the power system device 17a to support the hard disk device 12t> i2f used in a non-standard type,丨竑...'The controller 24 must have a special design. See Figure 5 Figure 5 is a schematic diagram of the controller of the present invention. As shown in Figure 5, the controller 24 of the present invention further includes a control sub-unit, a start (b〇〇t) sub-single το 24b. The control sub-unit 24a will automatically provide the host side with the logical block addressing (L〇gical B1〇ck Addressing, the zeroth logical block (10) Q numbered G in the _ mode) when the user starts the host, but does not provide The zeroth logical block (G) contained in the hard disk device 13 200821852 12f (with the operating system installed), and then activated after the user selects to boot the specific hard disk device specific (four) system. The (boot) subunit 24b retrieves the relative zeroth logical block (LBA 0) containing the selected operating system via the network pipe, and transmits it back to the central processing unit (CPU) of the host terminal via the control right subunit 24a. The operating system is selected to be activated (b〇〇t). It is intended to more clearly describe the features and spirit of the present invention, rather than the above-disclosed embodiments. The specific embodiments are intended to limit the scope of the invention. On the contrary, the purpose is to cover various changes and equivalent arrangements within the scope of the patent application scope of the present invention. [Simplified Schematic] FIG. 1 is a schematic diagram of a conventional storage area network environment. 2 is a schematic diagram of a network environment of a storage area according to the present invention. FIG. 3 is a schematic diagram of a two-channel network storage management apparatus according to the present invention. FIG. 4 is another schematic diagram of a two-channel network storage management apparatus according to the present invention. Fig. 5 is a schematic diagram of the controller of the present invention. 14 200821852 [Description of main component symbols] 10a~10η hard disk cabinets 11a, lib..., Ilf, llg···hard disk drawers 12a, 12b···, 12f, 12g ··· hard disk device 17a, 17b···, 17g~17n computer device 16a, 16b···, 16g~16n network storage management device 14 management device 20a first host hard disk interface 20b second host hard Disc interface 20c third host hard disk interface 22a first device hard disk interface 22b first device hard disk interface 24 controller 24a control subunit 24b boot subunit 26 buffer memory unit 28 network interface 15