200424865 玖、發明說明： 【發明所屬之技術領域】 本發明涉及一種具有育料匯流排介面之配置，其是與資 料匯流排控制器相連，該配置包含：至少一主大容量記憶 體，其是實際上與邏輯上都是與資料匯流排控制器相連； 至少一預留大容量記憶體，其實際上是與該料匯流排控制 器相連。 對於電腦和大容量記憶體容量之升高之需求會導致外部 大容量記憶體逐漸增多之需求，其另需具有一種高的可支 配性。外部之大容量記憶體（大部份是硬碟）組合成一種複合 物且共同連接至一電腦單元。實際上已達成一種所謂RAID-系統（Redundant Array of Independent Discs)。RAID-系統由 多個各別之硬碟所構成，其上以多重方式（redundant)儲存著 資訊。在一硬碟失效時，仍可由可作用之大容量記憶體來 重建已失去之資訊。 【先前技術】 RAID-系統內部中已設定一種所謂SCSI (Small Computer Standard Interface)-介面作爲各別之硬碟之間和RAID-系統 及所連接之電腦之間之控制介面。但亦存在其它之大容量 記憶體匯流排系統，例如，Fire-Wire。SCSI-系統允許直至 15個具備SCSL·能力之裝置連接至SCSI-匯流排控制器，使 該匯流排控制器可控制全部所連之裝置。 在一般之RAID-系統中，除了匯流排控制器和其上所連接 之由其所控制之硬碟之外，另存在一種控制元件，其將該 200424865 硬碟之狀態信號送回至該匯流排控制器。此種RAID-系統之 二種例子（其具有大量記憶體用之資料匯流排，其中一資料 匯流排特別是由SCSI-介面所形成）如第3，4圖所示。構成 硬碟之大容量記憶體提供一種控制信號至該控制元件。該 控制元件是與一匯流排直接相連（第4圖）或經由一種介面 來告知該匯流排控制器該硬碟之狀態。同時該控制元件可 告知一外部之使用者該硬碟之狀態。多個與該匯流排控制 器相連之硬碟用作預留-硬碟，即，其在進行中之操作時不 使用。 若一包含資料之硬碟失效，則該控制元件將此事告知該 匯流排控制器。該匯流排控制器以該預留-硬碟來取代該有 缺陷之裝置且進行一種更新過程，以便重建已失去之資 料。若未存在任何預留-硬碟，則該RAID-系統保持在臨界 狀態，即，其可在另一硬碟失效時產生資訊損耗。 又，有缺陷之硬碟須由RAID-系統去除且因此實際上亦由 匯流排隔離，使其在更換該有缺陷之硬碟時不會使匯流排 介面上之阻抗改變，這在最壞情況下會造成巨大之匯流排 干擾且使匯流排速率下降。藉由預先保留多個硬碟作爲失 效時之預留-硬碟，則一種具有SCSI-介面之RAID-系統亦可 非最佳化地被使用，此乃因邏輯上與該匯流排控制器相連 之各預留-硬碟不必儲存有效資料。 【發明內容】 本發明之目的是在具有匯流排介面之系統中提供〜種大 容量記憶體更換用之配置和方法，其未具備上述之缺點。 200424865 該目的以申請專利範圍各項特徵來達成。在具有資料匯 流排之配置中，其具有匯流排控制器和至少一主-和一預留_ 大容量記憶體，其經由匯流排介面而與該匯流排控制器實 際上相連接且該主大容量記憶體在邏輯上亦與該匯流排控 制器相連，另設有一種控制元件，藉此亦可在該主大容量 記憶體失效後可使其在邏輯上與該匯流排控制器相隔開且 藉此使至少一預留-大容量記憶體在邏輯上可與該匯流排控 制器相連。 該控制元件在該硬碟失效後測得其所屬之系統辨認碼且 使其在邏輯上與資料匯流排介面相隔開。然後將所測得之 辨認碼分配給該預留-大容量記憶體且使其在邏輯上與該匯 流排控制器相連。然後使該資料匯流排重新起始。 在本發明之配置中，實際上可使較邏輯上空著的可使用 之位置還多之大容量記憶體經由資料匯流排介面而與該匯 流排控制器相連，其中該邏輯上空著的可使用之位置是由 該辨認碼來預設。可任意地選取該預留容量而不會使邏輯 上空著的可使用之已連接之大容量記憶體之數目減少。 一種特別有利之形式是以並聯或串聯之SCSI-匯流排系 統來形成本發明之配置。 在一種形式中，大容量記憶體經由第二匯流排介面而與 第二匯流排控制器相連，第二匯流排控制器實際上經由第 一匯流排介面而與第一匯流排控制器相連。另一方式是第 二匯流排控制器實際上和邏輯上與第一匯流排控制器相 連0 200424865 在本發明之一有利之形式中，大容量記憶體在邏輯上與 資料匯流排控制器相隔開或相連接可藉由該大容量記憶體 之由該控制元件對該供應電流之控制來進行。在需要時該 控制元件可使該供應電流接通或斷開且因此使該大容量記 憶體驅動或去驅動。 另一方式是藉由該控制元件發送一種隔開-或連接信號使 該大容量記憶體可與該資料匯流排控制器在邏輯上相隔開 或相連接。利用此種配置（其中可有利地使用該硬碟記憶體 作爲大容量記憶體），則可簡易地形成RAID-系統。 又，該控制元件是匯流排控制器之組件，這樣可達成一 種高積體化之省空間之配置。此外，藉由該控制元件可使 該匯流排重新起始，則這樣亦是適當的。 另外，若該控制元件可支配一種組態記憶體，則該辨認 資料能以所屬之大容量記憶體來儲存在該組態記憶體中且 因此可快速地被使用。藉由發送一種”重置（reset)指令，，至該 匯流排控制器和全部相連之裝置中，則可包括一種更換過 程且因此可防止資料損耗。 又，本發明將依據第1，2圖所示之以SISC-系統來作成 之具體之實施例來說明。使用SISC-系統來說明對本發明之 構想不會造成限制。 【實施方式】 第1圖顯示SCSI-系統匯流排介面B(其具有連接於其上之 匯流排控制器BC)和數個主大容量記憶體MP1，MP2至 MPn ’其實際上和邏輯上是經由匯流排介面b而與匯流排控 200424865 制器BC相連。所謂，，實際上相連接，，以下是指該大容量記憶 體I至由SCSI-匯流排介面而與匯流排控制器在電性上相 連。若一種SCSI-辨認碼分配給該大容量記憶體且該SCSI-匯流排控制器已知悉該辨認碼，則該二者互相通信，因此 該大容量記憶體亦在邏輯上與匯流排控制器相連。 SCSI-辨認碼或LUN之分配是明確的。可使用之ID，s之數 目是與可使用之匯流排系統有關且在SCSI_匯流排系統中 該數目依據實施結果是7或15。一種SCSI-辨認碼保留給該 匯流排控制器。 該SCSI-匯流排控制器（其通常稱爲”Master”）經由LUN’s 來控制各主大容量記憶體（所謂”Slave”）之間和其與一此處 未顯示之相連之外部之電腦單元之間整個內部之資料交 換。電腦單元大部份經由另一未顯示之介面而與SCSI-匯流 排控制器相連。 該SCSI-匯流排控制器因此是一種資料匯流排控制器，其 經由資料匯流排介面（此處之特殊情況是SCSI-介面）來進行 通信。但資料匯流排控制器和資料匯流排之槪念應限制在 一與內部裝置之資料交換上。就資料交換而言，每一大容 量記憶體和匯流排控制器都配置有一種SCSI-辨認碼（所謂 SCSI-ID)。因此可明確地辨認每一個相連接之裝置。 在SCSI-匯流排之一種簡單之實施形式中，其顯示7個或 15 個空著的 SCSI-碼（LUN，LogicalUnitNumber)，其可分配 給主大容量記憶體。匯流排控制器BC設有SCSI-辨認碼或 LUN。此外，該配置具有一種預留-大容量記憶體RP1，RP2 200424865 至RPn，其經由SCSI-匯流排-介面B而與匯流排控制器BC 實際地相連。 該配置具有一種控制元件EP，其含有由每一主-和每一各 別之預留-大容量記憶體而來之狀態信號IP，其另外指出： 相對應之大容量記憶體是否是活性的（active)且實際上或邏 輯上是否與該匯流排控制器BC相連。該控制元件具有一種 組態記憶體IDS，其中儲存著全部之與SCSI-匯流排-介面B 相連接之大容量記憶體-和其狀態之辨認資料。又，該控制 元件EP經由控制信號DS而分配一種SCSI-辨認碼至每一各 別之大容量記憶體。該控制元件EP藉由該控制信號DS來 驅動或去驅動所屬之大容量記憶體之電流供應器。相對應 之大容量記憶體因此可保持著與該電流供應器相連接且只 經由該控制信號DP而被驅動或去驅動。 在正常操作時，一種明確之SCSI-辨認碼分配至每一大容 量記憶體且儲存在該控制元件EP之組態記憶體IDS中。該 辨認碼在該系統第一次向上運行時配屬於該主大容量記憶 體。全部所用之主大容量記憶體MP1，MP2至MPn全部獲得 一種控制信號DP以驅動該電流供應器（Drive Power On)。這 是藉由一種內部裝置（大部份是硬碟上之開關）來達成，其可 由該控制信號來切換。 現在若一種硬碟（例如，MP1)失效，則其不再提供一種有 效信號至該匯流排控制器。該匯流排控制器在控制元件EP 上產生該硬碟MP1之一種新的狀態信號IP，其表示該硬碟 失效。相同意義之信號同樣亦可由該匯流排控制器BC經由 10- 200424865 主介面而發送至外部之電腦單元，其應對有缺陷之大容量 記憶體進行存取。該有缺陷之大容量記憶體在該控制裝置 中標記成”DEAD”。 該控制元件測得該有缺陷之大容量記憶體之SCSI-辨認 碼。這可藉由讀取該組態記憶體IDS來達成。該硬碟控制 元件發送使該電流供應器去驅動用之控制信號至該有缺陷 之大容量記憶體。該大容量記憶體於是在邏輯上與該SCSI-匯流排-介面相隔開且因此亦與該匯流排控制器相隔開。 又，該有缺陷之大容量記憶體實際上（即，機械上）是與該匯 流排介面相連接。 已測得之SCSI-辨認碼分配給該預留-大容量記憶體（例 如，RP1)。該控制元件EP然後發送一種驅動用控制信號以 使電流供應至該預留-大容量記憶體RP1。最後，進行該 SCSI-匯流排-介面之重新起始。該SCSI-匯流排控制器於是 重新讀出該已連接之裝置。 該起始過程通常是在進行一種所謂”重置-指令”之後進 行，其藉由該”重置”匯流排信號之驅動來觸發。該信號亦 可有利地用來含蓋該SCSI-辨認碼之更換-或處理過程。該 更換之上述過程中所產生之干擾對其它裝置之作用因此可 被防止。 該控制元件同時可發送RL-信號至外部之LEDs，其表示 各別硬碟之狀態。該硬碟之失效因此可顯示給一外部之使 用者而不會使整個系統之功能受到限制。該SCSI-匯流排之 重新起始用之指令由該控制元件經由一控制介面BI而發送 -11- 200424865 或亦可直接由匯流排控制器發出。 在更換該有缺陷之大容量記憶體之後’起動一種更新過 程，以重建該消失之資料。若多個預留-硬碟存在該Raid-系統中，則正在運行之操作期間不必更換該已失效之硬 碟。這在正常維護期間可在重建結束之後進行。 第2圖顯示另一種形式。相同之組件具有相同之符號， 因此不必再說明。在該實施形式中，全部之大容量記憶體 都已驅動，即，一種使該電流供應器關閉或接通所用之控 制信號已不需要。連續驅動之大容量記憶體之優點是：可 獲得確定之溫度特性，這樣可使所需之冷卻裝置簡化。 在此種情況下全部之預留·大容量記憶體RP1，RP2至RPn 實際上經由匯流排介面B而與匯流排控制器BC相連。一種 邏輯上之連接經由該控制信號DC (<：0111^以/0丨3(：011116以）來進 行。 此處在告知一種失效已發生之後亦測得該有缺陷之大容 量記憶體之SCSI-辨認碼且分配給一預留大容量記憶體。 然後該控制元件EP發送一種DC-控制信號至該有缺陷之大 容量記憶體，使其邏輯上與該匯流排控制器BC相分離。在 分離之後，相對應之”連接信號”DC發送至更換時所用之預 留-大容量記憶體。然後該匯流排又進行一種重新起始過 程。 在有缺陷之硬碟更換之後，該系統由臨界狀態轉換至正 常狀態，其中在有缺陷之大容量記憶體上對已失去之資料 起動一種相對應之重建過程。 -12· 200424865 以連接/去除連接-信號來形成時特別有利。藉由這些信號 使該大容量記憶體保持在驅動-或非驅動狀態，使其可對匯 流排控制器起反應或不起反應。另一方式是此種信號亦可 用來控制一種開關，其安裝在該大容量記憶體和該匯流排 介面之間，且情況需要時可使該大容量記憶體與匯流排介 面相隔開。 該配置特別適合用在RAID-系統中，其中一方面是該資料 存在一種高的備用性（Redundance)，但另一方面亦須確保一 種高的持續性之使用狀態。可將上述二種實施例相組合且 情況需要時該SCSI-匯流排不必重新進行起始過程。在具體 之實施情況中，不必確定該大容量記憶體之形式。該大容 量記憶體可爲硬碟，但亦可爲其它之匯流排裝置，例如， 導帶傳動機構，DVD-傳動機構或DVD-燒錄機。又，該系統 可以並聯之SCSI-系統來構成，其中全部之大容量記憶體都 被驅動或用作SAS(Se:rialAttachedSCSI)系統，其中該預留 -大容量記憶體只有在該主大容量記憶體失效後才被驅動。 在一已擴大之形式中，不只大容量記憶體而且橋式控制 器亦可配屬於各別之SCSI-LUN’s。各橋式控制器又經由該 資料匯流排介面B而與SCSI-匯流排控制器BC相連且被分 配一種明確之SCSI-辨認碼。該橋式控制器又管理該SCSI_ 辨認碼，使大容量記憶體之數目又可進一步提高。因此亦 可更換全部之橋式控制器。 又，亦可使該SCSI·裝置在實際上與SCSI-匯流排-介面相 連接且交替地使用這些裝置。該控制元件因此不只用來使 -13- 200424865 有缺陷之SCSI-裝置可與該SCSI-匯流排相隔開，且亦可在 需要時使邏輯上未相連接之裝置可與該s C SI -匯流排控制 器相連接。 本發明之核心思想當然不限於各實施例中所述之SCSI-介面，而是可擴展至全部之匯流排系統，其同時只可使用 有限數目之可於一裝置中使用之ID’S。 【圖式簡單說明】 第1 圖 本發明 之 實 施 例 0 第2 圖 本發明 另 一 實 施 例 0 第3 圖 習知之 RAID- 系 統 之配置。 第4 圖 另一習 知 之 配 置 〇 主要 元件之符號 表 : B 資 料 匯 流 排 BC 資 料 匯 流 排控制器 MP1， MP2, MPn 主 大 容 量 記憶體 RP1， RP2，RPn 預 留 -大容量記憶體 EP 控 制 元 件 IP 狀 態 信 號 DP 電 流 控 制 信號 DC 連 接 控 制 信號 DS 辨 認 控 制 信號 RL 控 制 發 光 體 IDS 組 態 記 憶 體200424865 发明 Description of the invention: [Technical field to which the invention belongs] The present invention relates to a configuration with a breeding bus interface, which is connected to a data bus controller. The configuration includes: at least one main large-capacity memory, which is In fact, it is logically connected to the data bus controller; at least one reserved large-capacity memory is actually connected to the material bus controller. The increased demand for computers and large-capacity memories will lead to a gradual increase in the demand for external large-capacity memories, which in addition need to be highly disposable. The external large-capacity memory (mostly hard disks) is combined into a complex and connected to a computer unit. In fact, a so-called RAID-system (Redundant Array of Independent Discs) has been reached. A RAID-system consists of a number of individual hard disks, on which information is stored in multiple ways (redundant). In the event of a hard disk failure, the available large-capacity memory can still be used to reconstruct lost information. [Previous Technology] A so-called SCSI (Small Computer Standard Interface) -interface has been set in the RAID-system as a control interface between the individual hard disks and between the RAID-system and the connected computer. But there are other high-capacity memory bus systems, such as Fire-Wire. The SCSI-system allows up to 15 SCSL-capable devices to be connected to the SCSI-bus controller, allowing the bus controller to control all connected devices. In a general RAID-system, in addition to the bus controller and the hard disk controlled by it, there is another control element that returns the status signal of the 200424865 hard disk to the bus. Controller. Two examples of such RAID-systems (which have a large amount of memory data buses, one of which is especially formed by a SCSI-interface) are shown in Figures 3 and 4. The large-capacity memory constituting the hard disk provides a control signal to the control element. The control element is directly connected to a bus (Figure 4) or via an interface to inform the bus controller of the status of the hard disk. At the same time, the control element can inform an external user of the status of the hard disk. Multiple hard disks connected to the bus controller are used as reserve-hard disks, i.e. they are not used during ongoing operations. If a hard disk containing data fails, the control element informs the bus controller of this fact. The bus controller replaces the defective device with the reserved-hard disk and performs an update process in order to rebuild the lost data. If no reserved hard disks are present, the RAID-system remains in a critical state, that is, it can cause information loss when another hard disk fails. In addition, the defective hard disk must be removed by the RAID-system and therefore actually isolated by the bus, so that the replacement of the defective hard disk will not change the impedance of the bus interface, which is the worst case. This will cause huge bus interference and reduce the bus rate. By pre-reserving multiple hard disks as reserve-hard disks in the event of failure, a RAID-system with a SCSI-interface can also be used non-optimally because it is logically connected to the bus controller Each reserve-the hard disk does not have to store valid data. SUMMARY OF THE INVENTION The object of the present invention is to provide a configuration and method for replacing large-capacity memory in a system having a bus interface, which does not have the above-mentioned disadvantages. 200424865 This objective is achieved by various features of the scope of patent application. In a configuration with a data bus, it has a bus controller and at least one main- and one reserved_ large-capacity memory, which is actually connected to the bus controller via the bus interface and the main The capacity memory is also logically connected to the bus controller, and a control element is provided, so that the main mass memory can be logically separated from the bus controller after the main memory fails. As a result, at least one reserved-large-capacity memory can be logically connected to the bus controller. The control element measures the system identification code to which it belongs after the hard disk fails and logically isolates it from the data bus interface. The measured identification code is then assigned to the reserved-large-capacity memory and logically connected to the bus controller. Then restart the data bus. In the configuration of the present invention, in fact, a larger amount of memory that can be used than the logically empty position can be connected to the bus controller through the data bus interface, where the logically empty one can be used. The position is preset by the identification code. The reserved capacity can be arbitrarily selected without reducing the number of logically available connected large-capacity memories. A particularly advantageous form is to form the configuration of the invention as a parallel or serial SCSI-bus system. In one form, the large-capacity memory is connected to the second bus controller via the second bus interface, and the second bus controller is actually connected to the first bus controller via the first bus interface. The other way is that the second bus controller is actually logically connected to the first bus controller. 0 200424865 In an advantageous form of the invention, the large-capacity memory is logically separated from the data bus controller. Or the phase connection can be performed by controlling the supply current of the large-capacity memory by the control element. The control element can switch the supply current on and off and thus drive or de-energize the large-capacity memory when needed. Another way is to send a partition-or connection signal by the control element so that the large-capacity memory can be logically separated or connected to the data bus controller. With this configuration, in which the hard disk memory can be advantageously used as a large-capacity memory, a RAID-system can be easily formed. In addition, the control element is a component of the bus controller, so that a space-saving configuration with high integration can be achieved. It is also appropriate that the busbar can be restarted by the control element. In addition, if the control element can dominate a configuration memory, the identification data can be stored in the configuration memory with the corresponding large-capacity memory and can therefore be used quickly. By sending a "reset" command, the bus controller and all connected devices can include a replacement process and thus prevent data loss. In addition, the present invention will be based on Figures 1 and 2 The illustrated specific embodiment is made by using the SISC-system. The SISC-system is used to illustrate the concept of the present invention. [Embodiment] FIG. 1 shows the SCSI-system bus interface B (which has The bus controller (BC) connected to it and several main large-capacity memories MP1, MP2 to MPn 'are actually and logically connected to the bus control 200424865 controller BC via the bus interface b. So-called, , Actually connected, the following means that the large-capacity memory I is electrically connected to the bus controller through the SCSI-bus interface. If a SCSI-identification code is assigned to the large-capacity memory and the The SCSI-bus controller is aware of the identification code, then the two communicate with each other, so the large-capacity memory is also logically connected to the bus controller. The SCSI-identification code or LUN allocation is clear. Can be used Of I The number of D, s is related to the available bus system and the number in the SCSI_bus system is 7 or 15. Depending on the implementation result, a SCSI-identification code is reserved for the bus controller. The SCSI-bus The controller (which is often referred to as the "Master") controls the entire internal data between the main mass memory (so-called "Slave") and its connection to an external computer unit not shown here via LUN's Exchange. Most of the computer units are connected to the SCSI-bus controller via another unshown interface. The SCSI-bus controller is therefore a data bus controller, which is connected via the data bus interface (here Special case is SCSI-interface) for communication. But the idea of data bus controller and data bus should be limited to data exchange with internal devices. As far as data exchange is concerned, each large-capacity memory and bus The bus controllers are all equipped with a SCSI-identification code (so-called SCSI-ID). Therefore, each connected device can be clearly identified. In a simple implementation form of the SCSI-bus, Shows 7 or 15 empty SCSI-codes (LUNs, LogicalUnitNumber), which can be assigned to the main mass memory. The bus controller BC is equipped with SCSI-identification codes or LUNs. In addition, this configuration has a reservation -Large-capacity memories RP1, RP2 200424865 to RPn, which are physically connected to the bus controller BC via the SCSI-bus-interface B. This configuration has a control element EP, which contains each master- and each Each reserved-state signal IP from the large-capacity memory, which additionally indicates whether the corresponding large-capacity memory is active and is actually or logically connected to the bus controller BC . The control element has a configuration memory IDS, which stores all the large-capacity memory connected to the SCSI-bus-interface B and identification data of its status. In addition, the control element EP allocates a SCSI-identification code to each of the respective large-capacity memories via the control signal DS. The control element EP drives or drives the current supplier of the large-capacity memory to which it belongs by the control signal DS. The corresponding large-capacity memory can therefore remain connected to the current supply and be driven or de-driven only by the control signal DP. In normal operation, an explicit SCSI-identification code is allocated to each large-capacity memory and stored in the configuration memory IDS of the control element EP. The identification code is assigned to the main mass memory when the system is first run up. All the main large-capacity memories MP1, MP2 to MPn are all given a control signal DP to drive the power supply (Drive Power On). This is achieved by an internal device (mostly a switch on the hard disk), which can be switched by the control signal. If a hard disk (eg, MP1) fails, it no longer provides a valid signal to the bus controller. The bus controller generates a new status signal IP of the hard disk MP1 on the control element EP, which indicates that the hard disk has failed. Signals with the same meaning can also be sent from the bus controller BC to the external computer unit through the main interface of 10-200424865, which should access the defective large-capacity memory. The defective large-capacity memory is marked as "DEAD" in the control device. The control unit measures the SCSI-identification code of the defective mass memory. This can be achieved by reading the configuration memory IDS. The hard disk control element sends a control signal for driving the current supply to the defective large-capacity memory. The large-capacity memory is then logically separated from the SCSI-bus-interface and therefore also from the bus controller. Also, the defective large-capacity memory is actually (ie, mechanically) connected to the bus interface. The measured SCSI-identification code is assigned to the reserved-large-capacity memory (for example, RP1). The control element EP then sends a drive control signal to supply a current to the reserved-large-capacity memory RP1. Finally, restart the SCSI-bus-interface. The SCSI-bus controller then rereads the connected device. The initial process is usually performed after performing a so-called "reset-command", which is triggered by the "reset" bus signal. The signal can also be advantageously used to cover the replacement- or processing of the SCSI-identification code. The interference on other devices caused by the above-mentioned replacement process can be prevented. The control element can also send RL-signal to external LEDs at the same time, which indicates the status of each hard disk. The failure of the hard disk can therefore be displayed to an external user without limiting the functionality of the entire system. The command for restarting the SCSI-bus is sent by the control element through a control interface BI -11- 200424865 or may be directly sent by the bus controller. After replacing the defective large-capacity memory, an update process is initiated to reconstruct the lost data. If multiple reserved-hard disks are stored in the Raid-system, it is not necessary to replace the failed hard disk during a running operation. This can be done during normal maintenance after the end of the reconstruction. Figure 2 shows another form. Identical components have the same symbols and need not be explained again. In this embodiment, all the large-capacity memories have been driven, that is, a control signal for turning the current supply off or on is no longer needed. The advantage of the continuously driven large-capacity memory is that a certain temperature characteristic can be obtained, which can simplify the required cooling device. In this case, all of the reserved and large-capacity memories RP1, RP2 to RPn are actually connected to the bus controller BC via the bus interface B. A logical connection is made via the control signal DC (<: 0111 ^ to / 0 丨 3 (: 011116)). Here, even after a failure has been notified, the defective large-capacity memory is also measured. The SCSI-identification code is allocated to a reserved large-capacity memory. The control element EP then sends a DC-control signal to the defective large-capacity memory to logically separate it from the bus controller BC. After the separation, the corresponding "connection signal" DC is sent to the reserved large-capacity memory used in the replacement. Then the bus undergoes a restart process. After the defective hard disk is replaced, the system is replaced by The critical state transitions to the normal state, in which a corresponding reconstruction process is started on the defective large-capacity memory for the lost data. -12 · 200424865 It is particularly advantageous when formed by connecting / removing the connection-signal. With these The signal keeps the large-capacity memory in a driving- or non-driving state, so that it can react or not respond to the bus controller. Another way is that this signal can also be used to control a A switch that is installed between the large-capacity memory and the bus interface, and can separate the large-capacity memory from the bus interface when the situation requires. This configuration is particularly suitable for use in a RAID-system, one of which is On the one hand, there is a high degree of redundancy in the data, but on the other hand, it must also ensure a high continuous use state. The above two embodiments can be combined and the SCSI-bus does not have to be re-used when the situation requires Perform the initial process. In the specific implementation, it is not necessary to determine the form of the large-capacity memory. The large-capacity memory can be a hard disk, but it can also be other bus devices, such as a belt drive mechanism, DVD -Transmission mechanism or DVD-recorder. In addition, the system can be configured in parallel with a SCSI-system, in which all large-capacity memories are driven or used as a SAS (Se: rialAttachedSCSI) system, in which the reserved-large The capacity memory is only driven when the main mass memory fails. In an expanded form, not only the mass memory but also the bridge controller can be equipped with a separate SCSI-L UN's. Each bridge controller is connected to the SCSI-bus controller BC via the data bus interface B and assigned a clear SCSI-identification code. The bridge controller also manages the SCSI_ identification code to make the The number of capacity memories can be further increased. Therefore, all bridge controllers can be replaced. Also, the SCSI · device can be actually connected to the SCSI-bus-interface and use these devices alternately. The The control element is therefore not only used to separate the 13-200424865 defective SCSI-device from the SCSI-bus, but also to enable logically unconnected devices to be connected to the s C SI-bus when needed The controllers are connected. The core idea of the present invention is of course not limited to the SCSI-interface described in the embodiments, but can be extended to all bus systems, which can only use a limited number of ID'S that can be used in one device at the same time. [Brief description of the drawings] FIG. 1 is an embodiment of the present invention 0 FIG. 2 is another embodiment of the present invention 0 FIG. 3 is a conventional configuration of a RAID-system. Figure 4 Another conventional configuration. Symbol table of the main components: B Data bus BC Data bus controller MP1, MP2, MPn Main mass memory RP1, RP2, RPn Reserved-Mass memory EP control element IP status signal DP current control signal DC connection control signal DS identification control signal RL control light IDS configuration memory
-14--14-