JP5891842B2 - Storage system - Google Patents

Description

  The present invention relates to a storage system, and more particularly to a storage system that distributes data and stores it in a plurality of storage devices.

  In recent years, with the development and spread of computers, various types of information have been converted into digital data. As a device for storing such digital data, there are storage devices such as a magnetic tape and a magnetic disk. Since the data to be stored increases day by day and becomes enormous, a large-capacity storage system is required. In addition, reliability is required while reducing the cost of the storage device. In addition to this, it is necessary that data can be easily retrieved later. As a result, there is a demand for a storage system that can automatically increase storage capacity and performance, eliminate duplicate storage, reduce storage costs, and have high redundancy.

  In response to such a situation, in recent years, a content address storage system has been developed as shown in Patent Document 1. In this content address storage system, data is distributed and stored in a plurality of storage devices, and the storage location where the data is stored is specified by a unique content address specified according to the content of the data.

  Specifically, in the content address storage system, block data obtained by dividing predetermined data is further divided into a plurality of fragment data, and fragment data that becomes redundant data (parity data) is added to the plurality of fragment data. Distributed and stored in a plurality of storage devices. Then, by designating the content address later, the data stored in the storage location specified by the content address, that is, fragment data can be read, and the predetermined data before division can be restored from a plurality of fragment data. it can.

  Further, the content address is generated so as to be unique according to the content of data. For this reason, if it is duplicate data, the data of the same content can be acquired by referring to the data at the same storage position. Therefore, it is not necessary to store the duplicate data separately, and duplicate recording can be eliminated and the data capacity can be reduced.

  Here, each piece of fragment data described above is associated with metadata including block data information that is the basis of the fragment data. For example, the metadata includes component configuration information for storing each piece of fragment data and control information such as parity setting, and the pieces of metadata of the pieces of fragment data belonging to the same block data have the same contents. In order to store the storage destination of fragment data and metadata in each storage node without being aware of the state of the disk or node, n logical containers called components are prepared in advance. Components can be arranged over a plurality of disks, and only one piece of fragment data constituting one block data is stored in each component. Arrangement of components with respect to each node and arrangement with respect to disks are autonomously performed by the system. In addition, when block data having the same content as already stored block data is written at the time of data writing, deduplication is eliminated and the same fragment data is not written twice.

  In a storage system such as that described above, when a failure occurs in a storage node that stores data and the storage node is disconnected from the system, the components on that storage node are regenerated on other storage nodes. The In other words, in the storage system described above, the predetermined data is divided into a plurality of fragment data, and the fragment data to be redundant data is further added. Data can be restored. The same applies when a failure occurs in a disk or component in the storage node.

  Here, a process of regenerating data stored in a storage node when a failure occurs in the storage node disclosed in Patent Document 2 will be described with reference to FIGS. 1 and 2.

  First, as shown in the upper side of FIG. 1, in a storage system 300 equipped with a plurality of storage nodes 401 to 404, each component 1 to 12 formed in each storage node 401 or the like has a block that is data to be stored. Fragment data to which data is divided and redundant data is added is distributed and stored. In such a state, if a failure occurs in a predetermined storage node, processing for regenerating lost fragment data from the fragment data stored in the remaining storage nodes is started immediately.

  Specifically, in the regeneration process, first, as shown in the lower side of FIG. 1, the components 10, 11, and 12 that store the data formed in the storage node 404 in which the failure has occurred are operated in storage. Regenerate to nodes 401-403. Then, as shown in the upper side of FIG. 2, the fragment data 1 to 9 stored in the operating storage nodes 401 to 403 are read, and the block data D that is the basis of the fragment data is read from the fragment data 1 to 9. Is regenerated, and the fragment data lost by re-dividing the data D is regenerated. After that, as shown in the lower side of FIG. 2, the regenerated fragment data is distributed and stored in the newly generated components 10, 11, and 12, that is, in each of the operating storage nodes 401 to 403. Note that some data cannot be accessed until this series of processing is completed.

JP 2005-235171 A JP 2011-154428 A

  On the other hand, in the above-described storage system, the number of parity that is redundant data can be freely set for each file system. Then, since data having different numbers of parities are mixed in the same component, data of different parities are stored on the same physical disk. Then, in a storage system composed of multiple storage nodes, if a failure occurs in a certain node / disk, part of the fragment data will be lost, but the number of failed nodes / disks and the settings Depending on the number of parity bits, there is a problem that data cannot be regenerated from the remaining fragment data. For example, when data having x number of parities is stored, even if a failure occurs in x number of nodes, data can be regenerated in such a case. However, if y fragments cannot be read due to multiple node / disk failures in a state where the regeneration of such data is not completed, the number of parity data from x to (x + y-1) Will be unable to be regenerated (data loss).

  As described above, in the above-described storage system, there is a possibility that stored data may be lost due to a node failure, and reliability is lowered.

  For this reason, an object of the present invention is to solve the above-mentioned problem, that is, a problem of a decrease in the reliability of recorded data in a storage system having a duplicate recording elimination function.

A storage system according to an aspect of the present invention
A plurality of storage means;
Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and distributing and storing the plurality of fragment data in the plurality of storage means Distributed storage processing means;
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Data regenerating means for regenerating based on fragment data,
The data reproducing means is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. The fragment data constituting the storage target data is regenerated.
The configuration is as follows.

An information processing apparatus according to another aspect of the present invention
Information processing apparatus comprising at least one of a plurality of storage means for distributing and storing a plurality of fragment data composed of divided data obtained by dividing storage target data into a plurality of pieces and redundant data for restoring the storage target data Because
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Data regenerating means for regenerating based on fragment data is provided,
The data reproducing means is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. The fragment data constituting the storage target data is regenerated.
The configuration is as follows.

Moreover, the program which is the other form of this invention is:
Information processing apparatus comprising at least one of a plurality of storage means for distributing and storing a plurality of fragment data composed of divided data obtained by dividing storage target data into a plurality of pieces and redundant data for restoring the storage target data In addition,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Realize data regeneration means to regenerate based on fragment data,
The data reproducing means is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. The fragment data constituting the storage target data is regenerated.
It is a program for realizing this.

In addition, an information processing method according to another aspect of the present invention includes:
Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and storing the plurality of fragment data in a plurality of storage means in a distributed manner ,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred The storage in which the failure has occurred based on the other fragment data in the priority order based on the number of the redundant data among the fragment data constituting the storage target data when regenerating based on the fragment data Regenerating the fragment data constituting the storage target data stored in the means;
The configuration is as follows.

  Since the present invention is configured as described above, it is possible to improve the reliability of recorded data in a storage system having a duplicate recording exclusion function.

It is a figure which shows operation | movement of the storage system relevant to this invention. It is a figure which shows operation | movement of the storage system relevant to this invention. It is a block diagram which shows the structure of the whole system in Embodiment 1 of this invention. FIG. 4 is a block diagram showing an outline of a configuration of a storage system disclosed in FIG. 3. FIG. 5 is a block diagram showing a configuration of a storage system disclosed in FIG. 4. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. FIG. 6 is an explanatory diagram for explaining the operation of the storage system disclosed in FIG. 5. 7 is a flowchart showing an operation of the storage system disclosed in FIG. 6. 7 is a flowchart showing an operation of the storage system disclosed in FIG. 6. 7 is a flowchart showing an operation of the storage system disclosed in FIG. 6. It is a block diagram which shows the structure of the storage system in attachment 1 of this invention.

<Embodiment 1>
A first embodiment of the present invention will be described with reference to FIGS. FIG. 3 is a block diagram showing the configuration of the entire system. FIG. 4 is a block diagram showing an outline of the storage system, and FIG. 5 is a block diagram showing a detailed configuration of the storage system. 6 to 12 are explanatory diagrams for explaining the operation of the storage system, and FIGS. 13 to 15 are flowcharts showing the operation of the storage system.

  Here, in the present embodiment, a case will be described in which the storage system is configured by connecting a plurality of server computers. However, the storage system according to the present invention is not limited to being configured by a plurality of computers, and may be configured by a single computer (information processing apparatus).

  As shown in FIG. 3, the storage system 10 according to the present invention is connected to a backup system 11 that controls backup processing via a network N. Then, the backup system 11 acquires the backup target data stored in the backup target device 12 connected via the network N and requests the storage system 10 to store it. Thereby, the storage system 10 stores the backup target data requested to be stored for backup. The storage system 10 according to the present embodiment is described by way of example in which backup target data is stored. However, this is an example, and any data may be stored.

  As shown in FIG. 4, the storage system 10 according to this embodiment is configured by connecting a plurality of server computers. Specifically, the storage system 10 includes an accelerator node 20 that is a server computer that controls storage and reproduction operations in the storage system 10 itself, and a storage node 30 that is a server computer including a storage device (storage means) that stores data. It is equipped with. Note that the number of accelerator nodes 20 and the number of storage nodes 30 are not limited to those shown in FIG. 3, and more nodes 20 and 30 may be connected.

  Furthermore, the storage system 10 according to the present embodiment divides and makes the data redundant, stores the data in a plurality of storage devices, and stores the data by a unique content address set according to the content of the stored data. It is a content address storage system for specifying a stored location. This content address storage system will be described in detail later.

  FIG. 5 shows the configuration of the storage system 10. As shown in this figure, first, the accelerator node 20 constituting the storage system 10 includes a file system service 21 configured by incorporating a program in an arithmetic device such as a CPU (Central Processing Unit), A block division processing unit 22, a deduplication processing unit 23, and a distributed processing unit 24 are provided. Note that all or part of the functions described above may be provided in the storage node 30 constituting the storage system 10. Hereinafter, each configuration of the storage system 10 will be described in detail, and its operation will be described with reference to the flowcharts of FIGS.

  The file system service 21 functions as a file system that receives an input of backup target data transmitted from the backup target device 12 and controls an operation of storing the data in the storage node 30. At this time, as will be described later with reference to FIG. 8, for example, a plurality of file system services 21 exist according to the type of the backup target device 12 (file systems 1, 2, 3, etc.), and data (block data). A different number of redundant data (parity data) is added to (storage target data)).

  Specifically, the block division processing unit 22, the deduplication processing unit 23, and the distributed processing unit 24 distribute and store data to be backed up in the storage node 30 or stored in the storage node 30. Process to read data. Here, an example of the distributed storage processing by each of the processing units 22, 23, 24 is shown in FIGS.

  First, when the storage system 10 receives an input of data A, which is backup target data (arrow Y1 in FIGS. 6 and 7, step S1 in FIG. 13), the block division processing unit 22 changes to the arrow Y2 in FIG. As shown, the data A is divided into block data D (data to be stored) having a predetermined capacity (for example, 64 KB) (step S2 in FIG. 13). Based on the data content of the block data D, a unique hash value H representing the data content is calculated (arrow Y3 in FIG. 7). For example, the hash value H is calculated from the data content of the block data D using a preset hash function.

  Subsequently, in the deduplication processing unit 23, the storage system 10 performs deduplication processing using the hash value H of the block data D in order to eliminate duplicate recording of the block data D having the same contents (FIG. 13). Step S3). Specifically, first, the block data D that has already been stored is registered in association with the hash value H of the block data D and the content address CA representing the storage position, as will be described later. If the hash value H of the block data D already exists, it can be determined that the block data D having the same content has already been stored (Yes in step S4 in FIG. 13). In this case, the content address CA associated with the registered hash value H that matches the hash value H of the block data D before storage is acquired. Then, the content address CA is referred to as the content address CA of the block data D requested to be written. As a result, the already stored data referred to by the content address CA is used as the block data D requested to be written, and there is no need to store the block data D related to the write request.

  As described above, the writing process performed by referring to the content address CA of the block data D that is already stored is performed by the data storage processing unit 34 of the storage node 30. When the writing process by referring to the content address CA is completed, an “ACK” signal indicating that the writing process has been completed is returned to the host device such as the backup system 11 or the backup target device 12 (step in FIG. 13). S10). The subsequent processing by the data storage processing unit 34 of the storage node 30 will be described later.

  If it is determined that the block data D related to the write request is not yet stored, the storage system 10 performs processing for storing the block data D in the distributed processing unit 24 (step of FIG. 13). S5, S6, S7, S8). Specifically, first, the block data D is divided into a plurality of pieces of fragment data (divided data) having a predetermined capacity. For example, as shown by reference numerals D1 to D9 in FIG. 6, the data is divided into nine fragment data (divided data 41). Further, the storage system 10 generates redundant data so that the original block data D can be restored even if some of the divided fragment data is missing, and adds it to the divided fragment data 41. . For example, three pieces of fragment data (redundant data 42) are added as indicated by reference numerals D10 to D12 in FIG. As a result, a data set 40 composed of 12 pieces of fragment data composed of nine divided data 41 and three redundant data 42 is generated (arrow Y4 in FIG. 7).

  Each fragment data generated as described above is distributed and stored in each component C formed in each storage node 30 described later by the distributed processing unit 24 and the data storage processing unit 34 of the storage node 30 described later. The For example, as shown in FIG. 6, when 12 pieces of fragment data D1 to D12 are generated, one piece of fragment data D1 to D12 is added to each component C that is a data storage area formed in the storage node 30. Each one is stored (see arrow Y5 in FIG. 7). As described above, the processing units 22, 23, and 24 equipped in the accelerator node 20 and the data storage processing unit 34 equipped in the storage node cooperate with each other from the block data obtained by dividing the backup target data. It functions as a distributed storage processing means for generating a plurality of fragment data (including redundant data) obtained by further dividing the data and distributing and storing them in a plurality of storage devices.

  In the distributed storage process described above, in addition to the fragment data D1 to D12, metadata including information related to each fragment data is also stored in each component C. Such processing will be described when the function of the data storage processing unit 34 of the storage node 30 is described.

  When the fragment data is stored as described above, the storage position of the fragment data D1 to D12, that is, the storage position of the block data D restored by the fragment data D1 to D12 is stored in the storage node 30. Is generated. At this time, the content address CA includes, for example, a part of the hash value H (short hash) calculated based on the contents of the stored block data D (for example, the top 8B (bytes) of the hash value H) and the logical storage position. And information representing the information is generated. The content address CA is managed in the file system in association with identification information such as the file name of the backup target data and the content address CA in the accelerator node 20 that manages the file system in the storage system 10. The

  When the storage system 10 receives a file read request, the storage system 10 specifies the storage location specified by the content address CA corresponding to the requested file, and each fragment data stored in the specified storage location. Can be read as the data requested to be read. As described above, the storage system 10 has a function of reading and writing data.

  Next, the configuration of the storage node 30 will be described. In the present embodiment, a plurality of storage nodes 30 are provided. Since the configuration of the storage nodes 30 is the same, the configuration of one of the storage nodes 30 will be described.

  As shown in FIG. 5, the storage node 30 includes a node / disk failure detector 31 and a data regeneration controller that are configured by incorporating a program into an arithmetic device such as a CPU (Central Processing Unit). 32, a regeneration processor 33, and a data storage processor 34. The storage node 30 also includes a plurality of disks 35 that are storage devices (storage means). Hereinafter, each configuration will be described in detail.

  As described above, the data storage processing unit 34 (distributed storage processing means) further divides block data (storage target data) obtained by dividing the backup target data into a plurality of data at the accelerator node 20 and adds redundant data. The generated fragment data is distributed and stored in each component C configured on the plurality of disks 35 equipped in the plurality of storage nodes 30. At this time, the data storage processing unit 34 associates each piece of fragment data with metadata including information related to the fragment data, and stores it in the same component C. As described above, when the writing process by referring to the content address CA is completed, an “ACK” signal indicating that the writing process has been completed is returned to a higher-level device such as the backup system 11 or the backup target device 12. (Step S9 in FIG. 13).

  The metadata first includes component configuration information representing the configuration of the component to which the block data that is the source of fragment data belongs. For example, each piece of metadata of each piece of fragment data that constitutes one data set 40 includes component configuration information that identifies the component C in which the fragment data that constitutes the same data set 40 is stored. Further, the metadata includes a parity number indicating the number of redundant data added when fragment data is generated from block data. For example, one piece of fragment data indicated by hatching is added to the data set of fragment data generated by the file system 1 shown in FIG. 8, and in this case, the metadata of each fragment data has a parity number of “1”. Is included. Similarly, the metadata of the fragment data generated by the file system 2 shown in FIG. 8 includes the parity number “3”, and the metadata of the fragment data generated by the file system 3 shown in FIG. A parity number of “11” is included.

  Furthermore, the metadata includes a referenced number that represents the number of block data that is the source of fragment data that is referred to as other block data determined to have the same data content. For example, in the example of FIG. 9, “file 1” is composed of block data “block 1” and “block 2”, and “file 2” is composed of block data “block 1” and “block 3”. It is supposed to represent that. In this case, “file 1” is stored with reference to the content addresses “CA1” and “CA2” referring to the storage positions of “block 1” and “block 2”, and “file 2” is stored in “block 1”. ”And“ Block 3 ”are stored with reference to the content addresses“ CA1 ”and“ CA3 ”referring to the storage locations. Then, “block 1”, which is actually stored block data, is referred to by “file 1” and “file 2”, so the number of referenced is “2”. For this reason, each piece of metadata of each piece of fragment data constituting the block data of “block 1” includes the number of referenced data of “2”. Similarly, “block 2” and “block 3” have a reference count of “1”, and each piece of metadata included in each piece of block data includes a reference count of “1”. Become.

  As described above, each piece of metadata associated with each piece of fragment data includes the component configuration information, the number of parity, and the number of referenced data of the block data constituted by the fragment data. And all the metadata linked | related with each fragment data produced | generated from one block data become the same content.

  Here, the above-described metadata is updated by the data storage processing unit 34 at an arbitrary timing. In other words, since the number of referenceds included in the metadata may be changed every time the block data is stored, the number of referenceds in the metadata stored in the component C is updated. Specifically, as described above, when the block data to be newly stored is already stored in the storage node 30, the data storage processing unit 34 newly sets the content address CA of the already stored block data. The write process is performed by referring to the block data to be stored. After the write process is completed (step S10 in FIG. 13), the number of references in the metadata of the referenced block data is asynchronous with the write process. Is updated (step S11 in FIG. 13). Note that the update processing of the number of referenceds in the metadata is not limited to being performed asynchronously after completion of the write processing as described above, and may be executed at any timing.

  The node / disk failure detector 31 (data reproducing means) constantly monitors the storage system 10 and detects a failure in the storage node 30 or the disk 35 (step S21 in FIG. 14). When the node / disk failure detector 31 detects a failure in the storage node 30 or the disk 35, the node / disk failure detector 31 identifies the storage node 30 or the disk 35 in which the failure has occurred, and generates a list of components that may have been lost due to the failure. Then, the data is transferred to the data regeneration controller 32. For example, as shown in FIG. 10, it is assumed that a failure has occurred in the disk 2 equipped in the storage node 30 in which the components C2, C8, and C11 are configured among the storage nodes 30 that constitute the storage system 10. In this case, it is determined that the component C11 formed on the disk 2 has been lost, and information on the component C11 is passed to the data regeneration controller 32. For example, in the storage node 30 or the storage system 10, information indicating in which storage node the disk is formed is stored in advance in the storage node 30 or the storage system 10. Do it. Alternatively, the missing component may be identified with reference to component configuration information included in metadata stored in another storage node 30 in which no failure has occurred, or may be performed by other methods. .

  The data regeneration controller 32 (data reproducing means) includes a list of components to be recovered from the information passed from the node / disk failure detector 31 and the current component arrangement configuration, and the arrangement of the component (disk ) Temporary candidates (step S22 in FIG. 14). For example, calculation is performed so that a component lost due to the failure is recovered to another disk 35 in the same storage node 30 including the failed disk 35. In addition, the data regeneration controller 32 communicates with the data regeneration controller 32 provided in the other storage node 30 to determine the final configuration of the component to be recovered from the temporary candidates described above. 30 is notified to the data regeneration controller 32 (step S23 in FIG. 14). For example, the final configuration is determined by majority from the temporary candidates calculated by the data regeneration controller 32 of all the storage nodes 30. Here, in the example of FIG. 12, it is assumed that it is determined that a new component C2 'is recovered in the same storage node 30 when a failure occurs in the component C2.

  Further, the data regeneration controller 32 determines the priority order of the fragment data to be recovered when there are a plurality of fragment data stored in the failed component (step S24 in FIG. 14). At this time, it is determined to perform recovery in ascending order of the number of parity of block data composed of fragment data stored in the failed component, that is, the number of redundant data. For example, the fragment data that makes up the block data stored in the failed component is also stored in other non-failed components, so it is stored in the non-failed component. The parity number of the block data stored in the failed component is identified with reference to the parity number in the metadata of the fragment data. Then, the data regeneration controller 32 notifies the regeneration processor 33 of the configuration based on the determined component recovery and the priority order based on the number of parity (step S25 in FIG. 14).

  Specifically, the data regeneration controller 32 determines the priority of recovery in accordance with the number of fragment data lost among the number of fragment data constituting the block data due to a component failure. For example, when x pieces of fragment data are lost due to a component failure, the priority order is determined so that the number of parity is recovered preferentially in the order of x, x + 1, X + 2,. .

  Here, with reference to FIG. 11, it will be described in detail that the priority order of the block data to be recovered is determined in ascending order of the number of parity. For example, as shown in FIG. 11A, it is assumed that fragment data F (and corresponding metadata M) generated from block data having different numbers of parities is stored in one component C. . At this time, it is assumed that each piece of fragment data F constituting one block data is distributed so that only one is stored per component C. Further, the fragment data F indicated by the reference symbol P1 has a parity number “1”, the fragment data F indicated by the reference symbol P2 has a parity number “2”, and the fragment data F indicated by the reference symbol P3 has a parity number “3”. Suppose that

  In the above case, when a failure occurs in one component C, the fragment F in the component C is lost, but the block data composed of the fragment data F having the parity number “1” is If no more fragment F is lost, it cannot be recovered. In other words, if a failure occurs in another component C while the fragment F in the component C is being recovered, data cannot be regenerated. For this reason, it is desirable that the smaller the number of parities be, the faster the recovery is prioritized. For this reason, the fragment data F is recovered from the smallest number of parities. FIG. 11B shows a case where fragment data F (and corresponding metadata) having different numbers of parities are stored in one disk 35. In this case as well, When a failure occurs in the disk 35 and the fragment data F is lost, recovery is performed in ascending order of the parity number of the block data constituted by the fragment data F.

  The data regenerator 33 (data reproducing means) receives the configuration based on the component recovery and the priority order based on the number of parity from the data regeneration control unit 32 (step S31 in FIG. 15). Then, the data regenerator 33 regenerates the component C according to the received configuration, and first recovers the metadata stored in the component C (step S32 in FIG. 15). For example, the metadata is regenerated in the regenerated component C by copying the metadata stored in the other component C in which no failure has occurred. In the example of FIG. 10, when a failure occurs in the disk 2 of the storage node 30 and the data in the component C11 is lost, first, as indicated by the arrow Y11, another disk 3 in the same storage node 30. The component C11 is regenerated. Thereafter, as indicated by an arrow Y12, the metadata in the other storage node is copied and regenerated in the regenerated component C11.

  Subsequently, the data regenerator 33 refers to the metadata in the regenerated component C in the priority order of the number of parity notified from the data regenerating control unit 32, and other components in which no failure has occurred. From the fragment data in C, the fragment data lost due to the failure is regenerated. At this time, if there are a plurality of fragment data having the same parity number, recovery is performed in descending order of the number of referenced data of the block data constituted by the fragment data (steps S33 to S37 in FIG. 15). For example, when there are a plurality of fragment data to be recovered with the parity number “1”, the number of referenced is extracted from the metadata of the fragment data. Then, of the fragment data to be recovered with the parity number “1”, the fragment data to be recovered is selected in the descending order of the number of referenced, and the recovery is performed (see arrow Y21 in FIG. 12). Thereafter, when all the fragment data to be recovered with the parity number “1” is recovered, the fragment data with the parity number “2” is recovered in descending order of the number of referenced (see arrow Y22 in FIG. 12).

  As described above, the reason why the fragment data is recovered in descending order of the number of referenced is the data constituting a larger number of files as the number of referenced data is larger. Because it is big. That is, if recovery is impossible due to the loss of other fragment data while recovering the fragment F in the component C, fragment data with a larger number of references has a greater effect on the reliability of the storage system. . For this reason, the larger the number of references, the faster the recovery with priority.

  Here, as described above, the parity value and the number of referenceds that determine the priority for performing fragment data recovery are information held on the storage node 30 side as metadata to which the fragment data belongs. Therefore, the recovery process can be performed independently on the storage node 30 side, and the influence on the side accessing the storage system 10 can be suppressed.

  In the above description, it has been described that the metadata is updated asynchronously after the data writing process is completed. However, since the number of referenced block data is always changed for each writing process, the more important fragment data is This is because the setting is made quickly so that the recovery is performed with priority. For example, the area release processing for checking the block data reference count and releasing the block data storage area that is not referenced may be executed at any timing, but such area release processing is not executed for a long period of time. In some cases, important fragment data may not be recovered preferentially. Compared to such a case, the present invention is effective. However, if the metadata update is performed in synchronization with the end of the data writing process, a problem of performance degradation of the entire storage system also occurs. Therefore, in the present invention, the metadata is asynchronous with the writing process after the end of the writing process. To update. For example, after completion of the write process, the metadata may be updated when there is a preset margin in the load of the entire storage system or the storage node, and the condition setting for updating asynchronously is arbitrary.

<Appendix>
Part or all of the above-described embodiment can be described as in the following supplementary notes. The outline of the configuration of the storage system (see FIG. 16), information processing apparatus, program, and information processing method in the present invention will be described below. However, the present invention is not limited to the following configuration.

(Appendix 1)
A plurality of storage means 110;
Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and distributing and storing the plurality of fragment data in the plurality of storage means Distributed storage processing means 101;
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Data regenerating means 102 for regenerating based on the fragment data,
The data reproduction means 102 is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. Regenerating the fragment data constituting the storage target data that has been stored,
Storage system 100.

(Appendix 2)
The storage system according to attachment 1, wherein
The distributed storage processing means stores a parity number representing the number of redundant data of the fragment data constituting the storage target data in association with the fragment data,
The data regeneration unit is configured to store the storage target based on the parity number associated with the other fragment data constituting the storage target data stored in the other storage unit in which no failure has occurred. The number of the redundant data of the fragment data constituting the data is specified and stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data. The fragment data constituting the storage target data is regenerated.
Storage system.

(Appendix 3)
The storage system according to appendix 2,
The data regenerating unit is stored in the storage unit in which the failure has occurred based on the other fragment data in order of decreasing number of the redundant data among the fragment data constituting the storage target data. Regenerating the fragment data constituting the data to be stored;
Storage system.

(Appendix 4)
The storage system according to attachment 3, wherein
The distributed storage processing unit is configured to store the storage target data that is already stored in the plurality of storage units and other storage target data that is determined to be the same as the data content based on a preset criterion. To store the data to be stored that is already stored as the other data to be stored,
When the number of the redundant data among the fragment data constituting the storage target data to be regenerated is the same, the data regeneration unit is configured to use the storage target data as the other storage target data. Regenerating the fragment data constituting the storage target data stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number referred to;
Storage system.

(Appendix 5)
The storage system according to appendix 4, wherein
The distributed storage processing means stores a number of references representing the number of reference to the storage target data as the other storage target data in association with the fragment data constituting the storage target data,
When the number of the redundant data among the fragment data constituting the storage target data to be regenerated is the same, the data regenerating unit stores the data in the other storage unit in which no failure has occurred. Based on the referenced number associated with the other fragment data constituting the stored storage target data, the number of the storage target data is referred to, and the referenced number is Regenerating the fragment data constituting the storage target data stored in the storage means in which the failure has occurred based on the other fragment data in descending order;
Storage system.

(Appendix 6)
The storage system according to appendix 5,
The distributed storage processing means refers to the storage target already stored after the write processing of the other storage target data is completed by referring to the storage target data already stored in the other storage target data. Updating the number of referenceds associated with the data;
Storage system.

(Appendix 7)
Information processing apparatus comprising at least one of a plurality of storage means for distributing and storing a plurality of fragment data composed of divided data obtained by dividing storage target data into a plurality of pieces and redundant data for restoring the storage target data Because
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Data regenerating means for regenerating based on fragment data is provided,
The data reproducing means is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. The fragment data constituting the storage target data is regenerated.
Information processing device.

(Appendix 8)
An information processing apparatus according to appendix 7,
The storage means stores a parity number indicating the number of redundant data of the fragment data constituting the storage target data in association with the fragment data,
The data regeneration unit is configured to store the storage target based on the parity number associated with the other fragment data constituting the storage target data stored in the other storage unit in which no failure has occurred. The number of the redundant data of the fragment data constituting the data is specified and stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data. The fragment data constituting the storage target data is regenerated.
Information processing device.

(Appendix 9)
Information processing apparatus comprising at least one of a plurality of storage means for distributing and storing a plurality of fragment data composed of divided data obtained by dividing storage target data into a plurality of pieces and redundant data for restoring the storage target data In addition,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Realize data regeneration means to regenerate based on fragment data,
The data reproducing means is stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data of the fragment data constituting the storage target data. The fragment data constituting the storage target data is regenerated.
A program to make things happen.

(Appendix 10)
The program according to appendix 9, wherein
The storage means stores a parity number indicating the number of redundant data of the fragment data constituting the storage target data in association with the fragment data,
The data regeneration unit is configured to store the storage target based on the parity number associated with the other fragment data constituting the storage target data stored in the other storage unit in which no failure has occurred. The number of the redundant data of the fragment data constituting the data is specified and stored in the storage means in which the failure has occurred based on the other fragment data in a priority order based on the number of redundant data. The fragment data constituting the storage target data is regenerated.
program.

(Appendix 11)
Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and storing the plurality of fragment data in a plurality of storage means in a distributed manner ,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred The storage in which the failure has occurred based on the other fragment data in the priority order based on the number of the redundant data among the fragment data constituting the storage target data when regenerating based on the fragment data Regenerating the fragment data constituting the storage target data stored in the means;
Information processing method.

(Appendix 12)
An information processing method according to attachment 11, wherein
A parity number representing the number of redundant data of the fragment data constituting the storage target data is distributed and stored in the plurality of storage means in association with the fragment data;
The fragment data constituting the storage target data based on the number of parities associated with the other fragment data constituting the storage target data stored in the other storage means in which no failure has occurred The number of the redundant data is specified, and the storage target data stored in the storage unit in which the failure has occurred is configured based on the other fragment data in the priority order based on the number of the redundant data. The fragment data is regenerated.
Information processing method.

  Note that the above-described program is stored in a storage device or recorded on a computer-readable recording medium. For example, the recording medium is a portable medium such as a flexible disk, an optical disk, a magneto-optical disk, and a semiconductor memory.

  Although the present invention has been described with reference to the above-described embodiment and the like, the present invention is not limited to the above-described embodiment. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

10 Storage System 11 Backup System 12 Backup Target Device 20 Accelerator Node 21 File System Service 22 Block Division Processing Unit 23 Deduplication Processing Unit 24 Distributed Processing Unit 30 Storage Node 31 Node / Disk Failure Detection Unit 32 Data Regeneration Controller 33 Regeneration Processor 34 Data storage processor 35 Disk

Claims (4)

A plurality of storage means;
Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and distributing and storing the plurality of fragment data in the plurality of storage means Distributed storage processing means;
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Data regenerating means for regenerating based on fragment data,
The distributed storage processing means stores a parity number indicating the number of redundant data among the fragment data constituting the storage target data in association with the fragment data, and has already been stored in the plurality of storage means. When the other storage target data determined to be the same as the storage target data and the data content are stored in the plurality of storage means, the data is already stored as the other storage target data. Referring to the storage target data, and storing a reference number representing the number of the storage target data referred to as the other storage target data in association with the fragment data constituting the storage target data,
The data regeneration unit is configured to store the storage target based on the parity number associated with the other fragment data constituting the storage target data stored in the other storage unit in which no failure has occurred. The number of the redundant data of the fragment data constituting the data is specified, and the storage stored in the storage unit in which the failure has occurred based on the other fragment data in the order of decreasing the number of the redundant data There line regeneration of the fragment data constituting the target data, further, when the number of the redundant data of said fragment data configuring the storage target data to be regenerated is the same, a failure The other fragment data constituting the storage target data stored in the other storage means not generated Based on the number of referenced, the number of the storage target data is referred to as the other storage target data, and based on the other fragment data in descending order of the referenced number Regenerating the fragment data constituting the storage target data stored in the storage means in which the failure has occurred,
Storage system. The storage system according to claim 1 ,
The distributed storage processing unit, after the writing process of the other storage target data has been completed by referring to the storage target data already stored as the other storage target data, the storage object already stored Updating the number of referenceds associated with the data;
Storage system. Information processing apparatus comprising at least one of a plurality of storage means for distributing and storing a plurality of fragment data composed of divided data obtained by dividing storage target data into a plurality of pieces and redundant data for restoring the storage target data In addition,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred Realize data regeneration means to regenerate based on fragment data,
The storage means stores a parity number representing the number of redundant data of the fragment data constituting the data to be stored in association with the fragment data, and further stores the parity number in the plurality of storage means. When the other storage target data determined to be the same as the data to be stored and the data content based on a preset criterion are stored in the plurality of storage means, the other storage target data has already been stored. The stored data to be stored is referred to, and the number of references that represents the number of the stored data to be referred to as the other data to be stored is stored in association with the fragment data constituting the data to be stored. And
The data regeneration unit is configured to store the storage target based on the parity number associated with the other fragment data constituting the storage target data stored in the other storage unit in which no failure has occurred. The number of the redundant data of the fragment data constituting the data is specified, and the storage stored in the storage unit in which the failure has occurred based on the other fragment data in the order of decreasing the number of the redundant data There line regeneration of the fragment data constituting the target data, further, when the number of the redundant data of said fragment data configuring the storage target data to be regenerated is the same, a failure The other fragment data constituting the storage target data stored in the other storage means not generated Based on the number of referenced, the number of the storage target data is referred to as the other storage target data, and based on the other fragment data in descending order of the referenced number Regenerating the fragment data constituting the storage target data stored in the storage means in which the failure has occurred,
A program to make things happen. Generating a plurality of fragment data composed of divided data obtained by dividing the storage target data into a plurality of data and redundant data for restoring the storage target data, and storing the plurality of fragment data in a plurality of storage means in a distributed manner ,
The parity data representing the number of redundant data of the fragment data constituting the storage target data is stored in association with the fragment data, and the storage target data and data already stored in the plurality of storage means When storing other storage target data whose contents are determined to be the same based on a preset criterion in the plurality of storage means, refer to the storage target data already stored as the other storage target data And storing the number of references representing the number of the storage target data referred to as the other storage target data in association with the fragment data constituting the storage target data,
The fragment data constituting the storage target data stored in the storage means in which a failure has occurred, the other fragment constituting the storage target data stored in another storage means in which no failure has occurred When regenerating based on fragment data, based on the number of parities associated with the other fragment data constituting the storage target data stored in the other storage means in which no failure has occurred identifies the number of the redundant data of said fragment data configuring the storage target data, in the order number of the redundant data is small, is stored in the storage means that the problem has occurred on the basis of the other fragment data There line regeneration of the fragment data composing the which was the storage target data, further, it the object of regeneration When the number of the redundant data among the fragment data constituting the storage target data is the same, the other constituting the storage target data stored in the other storage means in which no failure has occurred Based on the referenced number associated with the fragment data, the number that the storage target data is referred to as the other storage target data is specified, and the other number is referred to in descending order. Regenerating the fragment data constituting the storage target data stored in the storage means where the failure has occurred based on fragment data;
Information processing method.

Images