1291618 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種資料保護方法，尤其係關於一種儲存系統 之快取記憶體之資料的保護方法。 【先前技術】 在電腦系統中，中央處理單元(CPU)之運算速度極快，但其直 接讀寫區塊設備(如，硬碟、SGSI硬碟、磁碟_等）尤其是對區 • 塊賴之定址時間很長。起初，CPU採用？1〇听聊1〇1^(11/0) 方^買寫區塊設備，即，CPU必需一直等待對區塊設備的讀寫操 作凡成後才可以繼績執行後續任務，極大地浪費了 CPU的時間， 降低了 CPU的貧源。為了解決此問題，採用了中斷的方式，即， 在項寫區塊δ又備日守，可以執行其他任務，當區塊設備讀寫操 雜=，藉由外部設備觸發中斷，讓⑽繼續執行剛才的任務， ^樣提Ν 了 CPU的使用效率。然:而，這種方式沒有充分利用讀寫 • 速度較快的隨機存取記憶體(Random Access Memory，RAM)之特 陡2此’採用一種磁碟快取記憶體(Disk Cache)之方式以進-,3门系、先效月匕’即，利用一部分系統隨機存取記憶體作為區塊 ^讀寫f作之磁魏取記憶體。當對賴進行寫操作時，將需 愔2之讀、設備名勒及在此設備德置記錄在磁碟快取記 1孜二而⑽即認為寫操作結束，然後由系統專門的讀寫設備 快^ =或於系統隨機存取記憶體資源不足時完成儲存在磁碟 …己讀中的寫操作，但磁碟快取記憶體中的資料繼續保留； 1291618 田對。X備進仃物作時’先判斷鱗快取記倾巾衫儲 要讀取之龍’如果有，則直接讀取。否則，CPU發出—個讀二 设備之任務，當前任務會被保留，⑽可以執行其他任務，當將 貧料從設備讀取後，讀取之資料會被添加至磁碟快取記憶體;， 並發送-個中斷信號至CPU，CPU可以繼續執行對設備的讀ς作。 為了提高對區塊設備的讀寫效率’區塊設備之驅綠㈣都採 用了-些提高讀寫效率的方法，如電梯演算法、預先讀 延遲寫演算法等。 ' 作業系統中磁碟快取記憶體描述項（Disk Cache Item)的每— 個描述項代表-個實體記憶體頁(page)或區塊(m〇ck)。對磁碟快 取記憶體的管理-般採用下述方法：爾快取記顏—般儲存在 贓(价w、乾淨(elean)和鎖（lQeked)三姆近最少使时咖 趾恤y Use 鏈結串列中。其中，贓鏈結串列包含所有需 要寫入區塊設備之_快取記鋪描述項，鎖麟㈣包含所有 ，在寫入區塊設備之補快取記㈣贿項，乾淨鏈結串列包含 最新的已儲存在_快取記鐘巾的區塊設備之資料。因此，磁 碟快取記憶體描述項可以在這三個鏈結串列之間遷移。咖執行 對區塊設備之寫操作時，會先從乾淨鏈結串财麵是否有對應 之磁碟快取記鐘描述項，如果有，職對應之剌快取記憶體 描述項從乾淨繼串财刪除，更改磁碟快取記憶體描述項中的 資料並將此磁碟快取記㈣描述項加人贓鏈結㈣；如果沒有， 則從系’洗祕：存取疏體中分配磁碟快取記憶體描述項，修改磁 8 1291618 =快取記憶體贿項中f料絲此磁顧 鍵結串列。接著’在-定時間間隔、系统隨機;喊 足、磁碟快取記憶體所占的隨機存取記憶雜達=二體： 強制清除磁碟快取記憶體時，會二 鎖鏈結串列，以寫入區塊設備，寫入/U體“述項加入 項加入乾、、_壯㈣ ”，、70畢之磁碟快取記憶體描述 、入乾_、、、。串列’且刪除乾淨鏈結串 2自_存取記憶體中供系統使用。在以儲存任務為主的李、 、、先中，大量的隨齡取記㈣彻作磁取記紐。 、 然而，將磁碟快取記憶體描述項加入贓鍵結串列和將寫入區 塊設備之磁碟快取記憶體描述項加入乾淨鏈結串列之過程是非同 步的’且摊是觸取記髓描述項加人贓鏈結串列之過程 較快。因此’磁碟快取記㈣愧存有資料沒有寫人真實之區塊 設備中。如果系統斷電時，磁碟快取記憶財的資料會丢失。因 而在#•對資料兀整性、安全性要求較高之儲存類產品（如， SAN、NAS、磁碟陣列）中，需要保護並處理這部分資料。很多廠商 提供了 Battery Backed(電池供電）RAM的解決方案。 常見之Battery Backed解決方案包含膽Adapter、Smart Array和Battery Backed RAM Adapter。其均為電腦系統之一個 δ又備，提供自己的設備驅動程式，可以加快寫資料之速度和資料 吞吐蓋。其將電腦系統隨機存取記憶體中磁碟快取記憶體之資料 1291618 轉移至自己内部的缓衝區，然後透過設備自身處理資料。其中， RAID Adapter、Smart Array將自身緩衝區之資料寫入和自己相 連之儲存設備，Battery Backed RAM Adapter僅儲存資料。其均 因為具有Battoy Backed功能，可以保護自己内部緩衝區之倾 於系統斷電後的-段時間内不去失，系統恢復正常後即可將^己 緩衝區之資料寫入儲存設備。 然而習知技術的這種做法在系_電時只能保護從儲 之隨機存取記憶體中傳遞到自己内部緩_之#料，不譜雷 腦系統隨機存取記㈣巾磁躲取記題儲存之資料。° ” 【發明内容】 提供知技射的問題與缺陷，本發明之目的在; 腦李、:Γ之快取記憶體之細保護方法，係用以於' 月旬系統断電時，保護一個且右 储存系統之快取記憶趙之f料有電池供電(BatteryBac_功心 =明職供之—種鱗_之快取記憶 :輪她_，顧蝴之隨= 啟動後，監控之㈣11域；讀存系射 ^職鏈結串列和乾淨鍵結宰列 -個磁碟快取記憶體描述 4、、先增加 體描述項在隨機存取_^=依照磁碟快取記憶 (Entir)之序號；館存此磁 〜出保留區域之-個入口 之位置、辦練記軸料、敎 =财摘财取記憶體 、思枝存取3己'丨思體大小、磁碟快 10 1291618 取記憶體描述項官A夕pm 在區塊設備之位置資設備名稱以及磁碟快取記憶體描述項 存取t _，咖獅記働述項在隨機 削保留區域之—個入口之序號，清除此入 隨機❿貝t儲存系統斷電時，透過BatteryBacked功能為 二、、子心-提供電源；當儲存系統重新啟動時，讀取該保留 區域之複數個人π儲存之磁碟快取記憶體描述項之位置和大小資 =依料取之位置和大小資贿㈣驗置之隨機存取記憶體 儲子之貝料，當儲存系統重新啟動後’讀取保籠域之複數個入 口儲存之磁碟快取記憶體描述項之資訊，依照讀取之資訊將對應 位置之隨機存取記憶體儲存之資料寫入對應之區塊設備；以及釋 放保留之隨機存取記憶體之郎，並設置保留區域之每一個入口 之值為零。119. The invention relates to a data protection method, and more particularly to a method for protecting data of a cache memory of a storage system. [Prior Art] In a computer system, the central processing unit (CPU) is extremely fast, but it directly reads and writes block devices (eg, hard disk, SGSI hard disk, disk_etc.), especially for the area/block. Lai's location is very long. At first, the CPU adopted? 1〇Listen 1〇1^(11/0) Square^Buy the block device, that is, the CPU must wait for the read and write operations of the block device before it can perform subsequent tasks, which is a great waste. The CPU time reduces the poor source of the CPU. In order to solve this problem, the interrupt method is adopted, that is, in the item writing block δ, the other tasks can be performed, and when the block device reads and writes the operation, the interrupt is triggered by the external device, and (10) continues to be executed. The task just now, ^ improved the efficiency of the CPU. However: this method does not make full use of the read and write speeds of the faster random access memory (RAM). This uses a disk cache (Disk Cache) In-, 3-door system, first-effect moonlight', that is, using a part of the system random access memory as a block ^ read and write f for the magnetic Wei memory. When writing to Lai, it will take 愔2 to read, the device name and the device to record on the disk cache 1 (2), that is, the write operation ends, and then the system special read and write device Fast ^ = or when the system random access memory resources are insufficient to complete the write operation stored in the disk ... read, but the data in the disk cache memory remains; 1291618 Tian right. When X is ready for the sputum, the first thing to judge is that the scalp is taken to remember the sloping towel. If you have it, read it directly. Otherwise, the CPU issues a task of reading the second device, the current task will be retained, and (10) other tasks can be performed. When the poor material is read from the device, the read data is added to the disk cache memory; And send an interrupt signal to the CPU, the CPU can continue to perform reading operations on the device. In order to improve the efficiency of reading and writing to block devices, the greening of block devices (4) has adopted methods to improve read and write efficiency, such as elevator algorithms and pre-read delay write algorithms. ' Each description item of the Disk Cache Item in the operating system represents - an entity memory page or block (m〇ck). The management of the disk cache memory generally uses the following method: 快快取颜颜-like storage in 赃 (price w, clean (elean) and lock (lQeked) three m near minimum time toe y use y Use In the chain of links, the 赃 chain string contains all the _ cache description items that need to be written to the block device, the lock lin (4) contains all, and the write block device (4) bribe items The clean link list contains the latest information about the block devices that have been stored in the _ cache. Therefore, the disk cache memory description can be migrated between the three links. When writing to a block device, it will first delete the memory description item from the clean link string. If there is, the corresponding cache memory description item is deleted from the clean line. , change the data in the disk cache memory description item and add the disk cache (4) description item to the human link (4); if not, then the system 'wash secret: access disk volume in the access volume Take the memory description item, modify the magnetic 8 1291618 = cache the memory bribe in the f wire this magnetic connection key string. 'At-time interval, system random; shouting, disk cache memory occupied by random access memory hybrid = two body: forced to clear the disk cache memory, the second chain of links, to Write to the block device, write /U body "Additional item added to dry, _strong (four)", 70-bit disk cache memory description, into the dry _,,,. The link string 2 is used by the system in the _ access memory. In the storage task, Li, 、, and 、, a large number of age-keeping notes (4) are used for magnetic retrieval. However, the disk is fast. The process of adding a memory description item to the 赃 bond sequence and adding the disk cache memory description item written to the block device to the clean link list is asynchronous, and the method is to touch the description item. The process of splicing the links is faster. Therefore, the disk cache (4) stores the data in the real block device. If the system is powered off, the data of the disk cache will be lost. #•Storage products with high data integrity and security requirements (eg, SAN, NAS, disk array) This part of the data needs to be protected and processed. Many vendors offer Battery Backed RAM solutions. The common Battery Backed solution includes the biliary Adapter, Smart Array and Battery Backed RAM Adapter, all of which are a δ of the computer system. Also, provide your own device driver, which can speed up the writing speed and data throughput cover. It transfers the data 1292118 of the disk cache memory in the random access memory of the computer system to its internal buffer, and then The data is processed by the device itself, wherein the RAID Adapter and the Smart Array write the data of the buffer itself to the storage device connected to itself, and the Battery Backed RAM Adapter stores only the data. Because of its Battoy Backed function, it can protect its internal buffer from being lost after the system is powered off. After the system returns to normal, it can write the data of the buffer to the storage device. However, this kind of practice of the prior art can only protect the material from the random access memory of the storage to the internal slow-moving _, the non-spectral Ray-Brain system random access record (four) towel magnetic stealing Information on the storage of the problem. °" [Abstract] The problems and disadvantages of the invention are provided. The purpose of the present invention is: a method for protecting the memory of the brain and the memory of the memory, which is used to protect one when the system is powered off in the month of the month. And the right storage system of the memory of the memory of the Zhao f material has battery power (BatteryBac_ power = Ming job for the sake of - scales _ fast memory: round her _, Gu Butterfly with = after the start, monitor the (four) 11 domain ; read the memory system and the chain link and the clean key node - a disk cache memory description 4, first increase the body description item in random access _ ^ = according to the disk cache memory (Entir) The serial number; the library stores this magnetic ~ out of the reserved area - the location of an entrance, the training of the axis, 敎 = financial resources, memory, Sizhi access 3 own '丨思体 size, disk fast 10 1291618 Take the memory description item A pm pm in the location of the block device device name and the disk cache memory description item access t _, the number of the entry in the random cut reserved area Clear this random mussels t storage system when power is off, through the BatteryBacked function for the second, the sub-heart - provide power; When the storage system is restarted, the location and size of the disk description of the disk cache of the plurality of individuals stored in the reserved area are read. Depending on the location and size of the reservation, the random access memory is stored. After the storage system is restarted, the information of the disk cache memory description item stored in the plurality of entries stored in the cage field is read, and the random access memory of the corresponding location is stored according to the read information. The data is written to the corresponding block device; and the reserved random access memory is used, and the value of each entry of the reserved area is set to zero.
其中’保留區域之適當開始位置係為儲存系统之作業系統啟 始化所而之隨機存輸憶體位i以及硬體在隨機存取記憶體映射 之位置以外之位置，而適當長度係透過如下公式計算出來：（（RAM size)/(cache item size)+l)*(保留區域之 entry size)+(uuiD size) ’ UUID 為通用唯一識別碼（Universal Unique IDentifier)。 保邊區域之入口序號係透過如下公式計算出來··（cache Hem在 RAM 之位置）/(cache item size)。 11 1291618 ，’不上所述，本發明之優點在於利用電 現了物細働取記謝磁碟快=1 體儲存之貧料的斷電保護功能。 【實施方式】 =下’將龄目切份對本發狀較佳f财式作詳細說明。 »月參考第1圖」，此圖為本發明一種儲存系統Wherein the appropriate starting position of the reserved area is the random memory location i of the storage system's operating system initialization and the location of the hardware outside the location of the random access memory mapping, and the appropriate length is transmitted through the following formula Calculated: ((RAM size) / (cache item size) + l) * (entry area entry size) + (uuiD size) ' UUID is the Universal Unique IDentifier (Universal Unique IDentifier). The entry number of the marginal area is calculated by the following formula: (cache Hem is in the RAM position) / (cache item size). 11 1291618, 'Unexpectedly, the present invention has the advantage of utilizing the power-off protection function of the poor material of the memory disk. [Embodiment] = The lower part will be described in detail for the hair style of the present invention. »Monthly reference to Figure 1", this figure is a storage system of the present invention
之資料的賴方法之方法流糊，如騎示，此綠包含以3 驟· 當儲存系献_，在儲存纽之雜存取記賴之適當開 始位置保留適當長度之—娜留區域⑽G)，其中，保留區域之 適田開始位置為儲存系統之作業系統啟始化所需之隨機存取記憶 ，位置以及硬體在隨機存取記㈣映射之位置财卜之位置，而適 當長度#'透過如下公式計算出來：（(腦size)/(cache —The method of the method of data is flowing, such as riding, this green contains 3 steps · When the storage system is _, the appropriate length is reserved at the appropriate starting position of the storage access record - Naliu area (10) G) Wherein, the starting position of the reserved area is the random access memory required for the initialization of the operating system of the storage system, the location and the position of the hardware in the position of the random access record (4) mapping, and the appropriate length #' Calculated by the following formula: ((brain size) / (cache -
S1Ze)+1)*(保留區域之 _ry Size)+(UUID size)，UUID 為通用 唯一则碼。當儲存祕之健⑽鍵立啟動健祕自身需要 之兄憶體訪問H域的管理機辦，將㈣區域之開始位置和長度 資料添加進去，以_保留此保留區域之目的。 虽儲存系統啟動後，監控系統對贓鏈結串列和乾淨鏈結串列 之操作（S110)。當偵測到系統增加一個磁碟快取記憶體描述項至 贓鏈、、，。串列日守，依照公式：（cache item在RAM之位置）/(cache item size)利用磁碟快取記憶體描述項在隨機存取記憶體之位置 计异出保留區域之一個入口之序號(sl2〇)，並依照計算出來之入 12 !291618 、"虎儲存磁碟&取€憶體描述項在隨機存取記憶體之位置以及 =碟快取記憶體描述項之隨機存取記憶體大小資訊至此入口 S130)。當細到系統增加一個磁碟快取記憶體描述項至乾淨鏈 結串列時’依照公式：（cache i⑽在讀之位置）/(cache㈣ 咖)利胸躲取記龍描述項在隨齡取記紐之位置計算 =保留區域之-個人σ之序號，並依照計算出來之人口序號清除 此入口儲存之資訊(S140)。 :儲存系統斷電時，透過臉吻此〔_功能為系統之隨機 子取心Μ提供電源，以保護系統隨機存取記憶體儲存之資料。 且為儲存系統之CPU提供較短時間的電源，以讓CPU完成將cpu 之快=記憶體資料同步至系統之隨機存取記憶體的任務咖0)。 /當儲存系統重新啟動時，透過修改儲存系統之基本輸入/輸 機存取繼自檢之參數設定來刪統隨 此體儲存之所有f料。重複_之保留過程。並讀取 ===S1GG細__㈣存之磁碟快 4體Μ奴位置和別、_，韻讀取讀置和大小 呆=應位置之隨機存取記憶體儲存之資料⑽0)。 田儲存系統重新啟動後’讀取保留區域之每—個入 _快取記描朗之f訊，颇之倾 f ,賴細_，彳_錄之隨機存取記_取磁 心體=述要寫人之資料、磁碟快取記憶體描述項需要 之區塊名細及磁碟快取記憶體描制在區塊設備之位置。 13 I291618 並依照讀取之資訊將需要寫入之資料寫入對應… 現對系統之隨機存取記憶體中磁碟快 二二，以貫 (S170)。 〒“取4體儲存之資料的保護 當系狀隨機存取記㈣巾财磁碟錄記憶體儲存之資 之區塊設備後，釋放儲衫紐動時保留之隨機存取 磁魏取記龍之郎，並設置隨機存取記憶體之 &域中母一個入口之值為零(S180)。 雖然本發咖前述之難實财式聽如上，财並非 之麟人祕#_到林麟本發明所附 明專仏_揭较本發明之範圍和精狀情 ΐ動简，均屬本發明之專利保護範圍™本發8= 疋之保護範圍請參考所附之申請專利範圍。 ， 【圖式簡單說明】 第1圖為本發明之方法流程圖。 【主要元件符號說明】 適#儲存魏啟動時，麵㈣統之隨機存取記憶體之 適虽開始位置保留適當長度之一個保留區域 體之 列之=當儲存系統啟動後，監控對贓鏈結串列和乾淨鏈結串 S12G切加—伽碟練錢翻述項至贓鏈結串 、碟蛛雜體描述項在隨機存取記'_之位置計算 區域之一個入σ之序號 1乐召 14 !291618 S130館存磁碟快取記憶體描述項在隨機存取記憶體之 以及磁碟快取記憶體描述項之大小資訊至此入口 ， ” S140 f增加-個快取記憶體描述項至乾淨鍵結串列時，依 二夬取記憶體描述項在_存取記龍之位置計算出保留區域之 固入口之序號’清除此入口儲存之資訊 " 存：體it，電時，透過― 大小資訊保_位置之Z大小纽’紐讀取之位置和 S170 U喊存取記憶體儲存之資料 儲存之快取，_留_之複數個入口 隨機存取記憶體儲存之龍f ’錢讀取之資訊將對應位置之 咖釋放保留=入對應之區塊設備 之每—個心之值= 記憶體之空間，並設置保留區域S1Ze) +1)* (_ry Size of reserved area) + (UUID size), UUID is the universal unique code. When storing the secret health (10) key to start the secret of the secret need to access the H domain management machine, add the start position and length data of the (4) area, to retain the reserved area. After the storage system is started, the monitoring system operates on the tandem link and the clean link (S110). When it detects that the system adds a disk cache memory description item to the chain, . Serial array, according to the formula: (cache item in the RAM location) / (cache item size) using the disk cache memory description item in the random access memory location to calculate the number of an entry in the reserved area ( Sl2〇), and according to the calculation into the 12!291618, " tiger storage disk & memory record description in the random access memory location and = disk cache memory description of the random access memory The body size information is here to the entrance S130). When the system adds a disk cache memory description item to the clean link series, 'according to the formula: (cache i (10) in the reading position) / (cache (four) coffee), the chest to hide the description of the dragon in the age New position calculation = reserved area - the number of the individual σ, and the information stored in this entry is cleared according to the calculated population number (S140). : When the storage system is powered off, the power is supplied to the system to protect the system's random access memory. Moreover, the CPU of the storage system is provided with a short-time power supply, so that the CPU can complete the task of synchronizing the CPU = the memory data to the random access memory of the system. / When the storage system is restarted, all the materials stored with this body are deleted by modifying the basic input/transport access of the storage system to follow the parameter settings of the self-test. Repeat the retention process. And read ===S1GG fine __ (four) save the disk fast 4 body Μ slave position and other, _, rhyme read read and size Stay = location of the random access memory storage information (10) 0). After the field storage system is restarted, 'read each area of the reserved area--------------------------------------------------------------------------------------------------------------------------------------------------------------- Write the data of the person, the block name required for the disk cache memory description item and the location of the disk cache memory in the block device. 13 I291618 and according to the information read, write the data to be written to the corresponding... The disk in the random access memory of the system is faster than the second (S170). 〒 “The protection of the data stored in the 4-body storage system is the random access memory that is reserved when the storage device is stored in the storage device. Lang, and set the value of a parent in the & domain of the random access memory is zero (S180). Although the above-mentioned hard-to-find financial style of this hair coffee is as above, the money is not the lindaughter #_到林麟The scope of the present invention and the simplification of the invention are all within the scope of patent protection of the present invention. The scope of protection of the present invention is as follows: BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a flow chart of the method of the present invention. [Description of main component symbols] When the storage is started, the face of the random access memory of the face (four) system retains a reserved area of an appropriate length. Body column = When the storage system is started, monitor the pair of chain links and the clean chain string S12G cut-and-sweep the money to describe the item to the chain link, the dish spider description item in random access Record the position of the '_ position calculation area into the σ sequence number 1 Le Zhao 14 !291618 S13 0 The library saves the memory description item in the random access memory and the size information of the disk cache memory description item to the entry," S140 f adds - a cache memory description item to the clean key string When the column is selected, the memory description item is calculated at the position of the _ access record dragon to calculate the serial number of the reserved area of the reserved area. 'Clear the information stored in this entry" Save: body it, when electricity, through the size information _Location Z size New Zealand's location and S170 U call access memory storage data storage cache, _ stay _ multiple entry random access memory storage dragon f 'money reading information Release the coffee at the corresponding location = enter the value of each heart of the corresponding block device = the space of the memory, and set the reserved area