JP4715777B2 - Replication method, storage apparatus, and program - Google Patents

Description

  The present invention relates to a replication method, and more particularly to a replication method in which a plurality of storage apparatuses are connected and data stored in one storage apparatus is replicated to a storage apparatus on the other replica side.

  In recent years, with the widespread use of information processing systems, various types of information are often stored and managed as digital data. For important data, replication such as backup to another computer is indispensable for disaster countermeasures. Replication is also widely used for the purpose of load distribution to computers. When performing replication, it is required to complete synchronization at high speed in order to reduce the risk of data loss and to distribute a large amount of data to a plurality of storages.

  Here, as one of the conventional replication methods, as shown in Patent Document 1 below, a file that does not exist on the replica side or a file that has been updated on the original side is copied from the storage on the original side. There is a method of transferring one file at a time (file itself or file difference). In other words, there is a method of performing synchronization in units of files. This method has an advantage that the replica side can be used while performing synchronization because synchronization is performed in units of files.

  As another replication method conventionally performed, there is a method of synchronizing by transferring RAW data written on a disk as shown in Patent Document 2 below. That is, there is a method of performing synchronization in block units. In this method, since the unit of data to be transferred becomes large, it is possible to create a replica at a higher speed than the above-described method of transferring files one by one.

JP-A-8-110840 JP 2005-182220 A

  However, when a new replica is to be created for a storage in which a large number of files are already stored, in the method of transferring one file at a time in units of files as described above, it takes time to read the file, or the unit of transfer is There is a problem that it takes time to synchronize because it becomes smaller. In addition, the above-described block unit method requires synchronization in a state where the original file system is unmounted, and requires a procedure for temporarily suspending the file system during replication. there were.

  Therefore, an object of the present invention is to provide a replication method capable of improving the inconveniences of the above-described conventional example, and particularly capable of performing replication at high speed without stopping the operation of the storage apparatus.

Therefore, one aspect of the present invention is
A block unit duplication step of duplicating data stored in the master side storage device at a predetermined time to the replica side storage device in block units;
In this block unit duplication process, when an update occurs to predetermined file data among the data stored in the master side storage device, it is possible to identify the order in which the file update information indicating the update contents is updated. A file update information storage step to be stored in the master storage device;
After completion of the block unit replication process, the file unit replication process for replicating the file update information to the replica storage device in the updated order;
It is characterized by having.

  According to the above invention, when replication is started, first, data in a predetermined state on the master side is replicated to the replica side in block units, and data synchronization is executed. If a file update occurs on the master side during that time, file update information representing the update content is accumulated. At this time, the update order of the file update information is stored so that it can be identified. When synchronization in units of blocks is completed, synchronization is performed in units of files in the update order of the accumulated file update information. Therefore, while performing high-speed replication by block-unit synchronization, file data can be updated during this time, and then file-unit synchronization can be performed to ensure data consistency. Replication can be performed.

  The file update information storage step is characterized in that update history information generated at the time of file update in the master storage device is stored in association with the file update information. For example, the file update information storage step stores update identification information that can identify the update order of the update history information included in the update history information in association with the file update information. Furthermore, the file update information storage step stores the update identification information in the file update information and deletes the update history information from the master storage device when the file update is completed in the master storage device. Yes.

  As a result, since the file update information is stored together with information indicating the order in which the file data is updated, the update order of the file update information can be more accurately and easily recognized at the time of the subsequent file unit synchronization. Can perform more accurate file synchronization. Further, by deleting the update history information, it is possible to suppress an increase in the amount of accumulated data.

  The file update information storage step is a journal registration step of registering file update information and update history information in the journal queue in the order in which the update of the file data stored in the master-side storage device occurs during the block unit replication step. And a journal deletion step for deleting the file update information and update history information from the journal queue, and the update identification information is associated with the file update information before and after the journal deletion step and stored in the master storage device. And a file update information registration step.

  Thus, by using the journal function of the file system, first, when file data is updated during synchronization in block units, the journal and ID, which are update history information, and file update information are paired into the journal queue. Remembered. When the update of the file data is normally completed, the journal is deleted and the ID and the file update information are stored in association with each other. After that, when the block unit synchronization is completed, the file update information can recognize the update order by referring to the associated ID, and the file unit synchronization can be executed. Therefore, by using the journal function of the file system, high-speed replication can be realized with a simple configuration without stopping the operation. Further, since the file update information is stored together with the deletion of the journal, it is possible to synchronize only the file data that has been successfully updated, and it is possible to improve the reliability of replication.

  The file unit duplication process is characterized in that file update information duplicated by the master storage device to the replica storage device is deleted from the master storage device. As a result, it is possible to prevent file update information from being continuously stored in the master-side storage device, and to suppress an increase in storage capacity.

Moreover, the other form of this invention is:
A master storage device connected to the replica storage device,
A block unit duplicating means for duplicating data stored at a predetermined time to a storage device on the replica side in block units;
A file that stores file update information representing the update contents so that the update order can be identified when an update occurs to predetermined file data among the stored data during replication by this block unit replication means Update information storage means;
File unit replication means for replicating stored file update information to the replica storage device in the order of update after replication by the block unit replication means;
It is characterized by having.

  The master storage device further includes update history generation means for generating update history information when file data is updated, and the file update information storage means stores the update history information in association with the file update information. It is characterized by.

  The update history generation means generates update identification information that can identify the update order of file updates, and the file update information storage means stores the update identification information in association with the file update information. Yes. In addition, the master storage device includes an update history deletion unit that deletes the update history information generated by the update history generation unit after file update is completed.

Then, the update history generating means registers the file update information and the update history information in the order in which the update of the file data occurred in the block unit replication, and registers them in the journal queue.
The update history deletion means deletes the file update information and the update history information that have been updated, from the journal queue,
The file update information storage means stores the file update information deleted from the journal queue by the update history deletion means and the update identification information in association with each other.
It is characterized by that.

  In addition, the master side storage device includes a file update information deletion unit that is stored in the file update information storage unit and deletes the file update information copied to the replica side storage device by the file unit replication unit. It is a feature.

Furthermore, the program which is the other form of this invention is:
To the master storage device connected to the replica storage device,
A block unit duplicating means for duplicating data stored at a predetermined time to a storage device on the replica side in block units;
During the replication by this block unit duplicating means, when an update occurs to predetermined file data among stored data, the file update information indicating the update contents is stored in the apparatus so that the update order can be identified. File update information storage processing means for storing;
File unit duplicating means for duplicating the stored file update information to the replica storage device in the updated order after duplication by the block duplicating means;
It is characterized by realizing.

Moreover, the other form of this invention is:
A replication system in which a master storage device and a replica storage device are connected to each other,
The master side storage device and the replica side storage device comprise block unit duplicating means for duplicating the data stored in the master side storage device at a predetermined time to the replica side storage device in block units,
The order in which the file update information indicating the update contents is updated when the master storage device is updated to the specified file data among the stored data during the block unit replication by the block unit replication unit File update information storage means for storing such that it can be identified,
File unit duplicating means for replicating file update information stored in the master side storage apparatus to the replica side storage apparatus in the order of update after the master side storage apparatus and replica side storage apparatus finish duplication in block units by the block unit duplication means With
It is characterized by that.

  Even the invention of the storage apparatus, the program, and the replication system having the above-described configuration operates in the same manner as the above-described replication method, so that the above-described object of the present invention can be achieved.

  Since the present invention is configured and functions as described above, according to this, first, synchronization in block units is performed, file update information of file data updated during this period is accumulated, and after completion of synchronization in block units Because synchronization is performed on a file basis in the order of file update, high-speed replication can be performed by synchronization on a block basis, and data can be aligned by subsequent synchronization on a file basis. In addition, it has an unprecedented excellent effect of being able to execute highly reliable replication.

  The present invention is characterized in that replication is performed at high speed without stopping operation by combining block unit synchronization and file unit synchronization in replication of a storage apparatus. Hereinafter, a specific configuration and operation for realizing replication will be described in an embodiment.

  A first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a block diagram showing a configuration of a replication system. FIG. 2 shows the configuration of the master storage device, and FIG. 3 is a block diagram showing the configuration of the replica storage device. 4 to 6 are explanatory diagrams showing a state during replication. 7 to 10 are flowcharts showing operations during replication.

[Constitution]
The replication system according to the present embodiment includes a master storage device 1 in which original data is stored, and a replica storage device 2 that is a replication destination of data stored in the master storage device 1, and It is configured to be connected via the communication path 3. It should be noted that more storage devices may be connected to the master storage device 1 as the replica storage device 2. Hereinafter, the configuration of each of the storage apparatuses 1 and 2 will be described.

  As shown in FIG. 1, the master-side storage device 1 includes a processor 1A that is an arithmetic device, a disk 1B that is a storage device that stores data, a memory 1C that temporarily stores data, and a replica-side storage device 2 and a communication unit 1D for transmitting and receiving data.

  As shown in FIG. 2, the processor 1A incorporates a master storage device program, thereby creating a snapshot creation processing unit 11, a block synchronization processing unit 12, a journal registration processing unit 13, and a journal deletion. A processing unit 14, a file update information registration processing unit 15, a file synchronization processing unit 16, and a file update information deletion processing unit 17 are constructed. Further, the memory 1 </ b> C is formed with a storage unit for storing the journal queue 18 and the file update history 19. Hereinafter, each configuration will be described in detail.

  The snapshot creation processing unit 11 has a function of creating a snapshot in block units of data stored in the disk 1B of the master storage device 1 at a certain time as preparation for performing synchronization in block units. Then, the block synchronization processing unit 12 (block unit replicating means) transfers the data to be replicated to the replica storage device 2 in units of blocks based on the created snapshot. As a result, data synchronization is started in block units between the master storage device 1 and the replica storage device 2. Since the block-unit synchronization processing is executed based on the snapshot, even if data is updated after the snapshot is created, the update is not reflected in the block-unit synchronization.

  The journal registration processing unit 13 (update history generating means) displays update history information of the file data when the data stored in the disk 1B is updated in units of files after the snapshot is created. Generate. Specifically, a journal (metadata update information, etc.) that is information representing an update history generated by a journal function that is normally provided in the file system, and an ID that is update identification information for identifying the update process, Are stored in the journal queue 18 as data indicating processing waiting. At this time, the journal registration processing unit 13 also generates file update information including the file name (including the path name) of the file data to be updated and information indicating the update contents of the file, and the ID and journal The data are stored in the journal queue 18 in association with each other. That is, when a file update occurs in the master storage apparatus 1 after the snapshot is created, the journal, ID, and file update information are paired and registered in the journal queue (see FIG. 5 described later). The ID is data that can identify the update order because it is information for identifying the update process of the file data. In addition, since data waiting for processing is registered in the journal queue 18, it is registered in the order in which file data is updated.

  The journal deletion processing unit 14 (update history deletion means) writes the contents of the file update registered in the journal queue 18 to the disk 1B, and at the timing when the file update is completed, reads the ID, journal, and file from the journal queue 18. Delete update information. At the same time, the file update information registration processing unit 15 (file update information storage means) extracts the file update information and the ID from the deleted ID, journal, and file update information, associates them, and updates the file update history. 19

  The journal registration processing unit 13, the journal deletion processing unit 14, and the file update information registration processing unit 15 operate every time file data is updated after the snapshot is created. The update of the file data that occurs until the processing is completed is performed by associating the file update information indicating the update contents with the ID for determining the order of file update related to the journal indicating the update history. Added to the history 19 and stored. That is, the file update information is accumulated so that the update order can be identified.

  Then, after the block unit synchronization is completed, the file synchronization processing unit 16 (file unit replication unit) updates the file update information stored in the file update history 19 based on the associated ID as described above. Determine the order. Then, data representing the updated contents is transmitted to the replica-side storage apparatus 2 in that order, and synchronization is performed so as to replicate with the replica-side storage apparatus 2 in units of files. Thereafter, the file update information that has been synchronized in units of files is deleted from the file update history 19 by the file update information deletion processing unit 17 (file update information deleting means) together with the associated ID.

  Next, the replica storage device 2 will be described. The replica-side storage device 2 is a storage device newly introduced to create a replica (replica) of the master-side storage device 1. Similarly to the master-side storage device 1, as shown in FIG. 1, a processor 2A that is an arithmetic device, a disk 2B that is a storage device in which data is copied and stored, and a memory 2C that temporarily stores data The general storage apparatus includes a communication unit 2D for transmitting and receiving data to and from the master storage apparatus 1.

  In addition, as shown in FIG. 3, a block synchronization reflection processing unit 21 and a file synchronization reflection processing unit 22 are constructed in the processor 2A by incorporating the replica storage device program. The block synchronization reflection processing unit 21 (block unit duplicating means) operates corresponding to the block synchronization processing unit 12 of the master side storage apparatus 1, receives the data transmitted in block units, and stores it in the disk 2B. Take synchronization. The file synchronization reflection processing unit 22 (file unit duplicating means) operates corresponding to the file synchronization processing unit 16 of the master storage device 1 and receives data representing the update contents transmitted in file units. The file is updated and stored in the disk 2B, and file synchronization is performed.

[Operation]
Next, the operation of the replication system using the above-described master storage apparatus 1 and replica storage apparatus 2 will be described with reference to FIGS. 4 to 6 are diagrams illustrating the configuration of each of the storage apparatuses 1 and 2 and illustrating a replication operation such as a data flow during replication. FIG. 7 is a sequence diagram showing operations of the storage apparatuses 1 and 2, and FIGS. 8 to 10 are flowcharts showing operations of the master storage apparatus 1.

  First, in preparation for block-unit synchronization processing, the master-side storage apparatus 1 creates a snapshot of data currently stored in the disk 1B at the snapshot creation processing unit 11 ((1) in FIG. 4). , Step S1 in FIG. 7, step S11 in FIG. 8, block unit duplication step).

  At the same time as creating the snapshot, the file system of the master-side storage apparatus 1 starts processing to accumulate file update information in the journal registration processing unit 13 when the stored file data is updated ( Step S2 in FIG. 7 and steps S12 and S13 in FIG. 8). The state at the time of the file update at this time will be described with reference to FIGS.

  First, when a file update process occurs, the journal registration processing unit 13 sets a journal and ID representing an update history and file update information representing the file update content corresponding to this update as a set, and the journal queue 18 (Step (3) in FIG. 5, step S21 in FIG. 9, journal registration step). Note that journal generation and registration in the journal queue 18 are processes that are always executed in the file system. Then, when the file update content is written to the disk and the file update is completed (step S22 in FIG. 9), the journal deletion processing unit 14 performs information corresponding to the update registered in the journal queue 18, that is, update. The journal, ID, and file update information that has been deleted are deleted from the journal queue 18 ((4) in FIG. 5, step S23 in FIG. 9, journal deletion step). At the same time, the file update information registration processing unit 15 additionally registers the file update information to be deleted from the journal queue 18 together with the ID in the file update history 19 ((5) in FIG. 5, step S24 in FIG. 9, file update). Information registration process). Then, the additional registration to the file update history 19 described above accompanying the file update is executed until the block unit synchronization is completed, as will be described later (file update information storage step). As described above, since the file update information is stored at the timing when the journal is deleted from the journal queue 18, it is possible to register only the file update information related to the file data that has been reliably updated. Since it is stored together with the journal ID, the update order of the file update information can be identified.

  As described above, after the snapshot is created, the block synchronization processing unit 12 of the master storage device 1 transmits data to the replica storage device 2 in units of blocks based on the created snapshot. Accordingly, the block synchronization reflection processing unit 21 of the replica-side storage apparatus 2 stores the block unit data in the received disk 2B, so that the block unit replication is executed ((2) in FIG. 4 and FIG. 7). Steps S3 and S4, Step S14 in FIG. 8, block unit duplication process). The initial state of the disk 2B of the replica storage device 2 may be empty. During this period (No in step S5 in FIG. 7 and No in step S15 in FIG. 8), when a file update occurs in the master-side storage apparatus 1, processing for accumulating file update information is executed as described above (FIG. 5). , File update information storage step).

  Thereafter, when the synchronization in block units is completed (Yes in step S5 in FIG. 7 and Yes in step S15 in FIG. 8), the master storage device 1 notifies the storage device 2 that the synchronization has been completed (step S6 in FIG. 7). Then, the block-by-block synchronization process is cut off and the process proceeds to the file-by-file synchronization process (step S18 in FIG. 8, file unit duplication process).

  Then, the replica-side storage apparatus 2 confirms the ID of the journal recorded in the file system constructed in the replica-side storage apparatus 2 and notifies the master-side storage apparatus 1 (step S7 in FIG. 7). Note that the journal ID stored in the replica-side storage apparatus 2 is an ID generated by a file-unit update performed in the master-side storage apparatus 1 before synchronization is made in block units. For this reason, the file update to which an ID subsequent to this ID is attached means an update after a snapshot is created in the master storage apparatus 1. Therefore, the master side storage apparatus 1 receives the ID notified from the replica side storage apparatus 2 (step S31 in FIG. 10), and updates the file update information associated with the ID after this ID to the file update history 19. The data is searched for and read (step S32 in FIG. 10). Then, on the basis of the read file update information, the file synchronization processing unit 16 and the file synchronization reflection processing unit 22 execute synchronization in units of files in an order that can be discriminated by the IDs associated with each (see FIG. (6), steps S8 and S9 in FIG. 7, and step S33 in FIG.

  Thereafter, the file update information that has been synchronized in units of files is deleted from the file update history 19 by the file update information deletion processing unit 17 ((7) in FIG. 6, step S34 in FIG. 10). Then, the file synchronization is completed when there is no file update information in the file update history 19 (step S35 in FIG. 10).

  As described above, according to the present invention, first, data is updated by accumulating updates of file data during high-speed replication based on block unit synchronization, and executing file unit synchronization after completion of block unit synchronization. Can be adopted. Therefore, high-speed and highly reliable replication can be executed without stopping the operation. In particular, by accumulating journal IDs in association with file update information, it is possible to recognize the order of file updates during file synchronization while using the journal function provided in the normal file system, so a more accurate file Synchronization can be executed, and there is no need to stop the file system when switching from block unit synchronization to file unit synchronization.

  The replication method of the present invention can be applied to uses such as storage device replication, and can also be applied to uses such as database replication, and has industrial applicability.

It is a block diagram which shows the structure of a replication system. It is a block diagram which shows the structure of a master side storage apparatus. 2 is a block diagram showing a configuration of a replica storage device. FIG. It is explanatory drawing which shows the replication operation | movement in a replication system. It is explanatory drawing which shows the replication operation | movement in a replication system. It is explanatory drawing which shows the replication operation | movement in a replication system. It is a sequence diagram which shows operation | movement of a replication system. It is a flowchart which shows operation | movement of the master side storage apparatus. It is a flowchart which shows operation | movement of the master side storage apparatus. It is a flowchart which shows operation | movement of the master side storage apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Master side storage apparatus 2 Replica side storage apparatus 3 Communication path 11 Snapshot creation process part 12 Block synchronization process part 13 Journal registration process part 14 Journal deletion process part 15 File update information registration process part 16 File synchronization process part 17 File update information Deletion processing unit 18 Journal queue 19 File update history 21 Block synchronization reflection processing unit 22 File synchronization reflection processing unit

Claims (8)

A block unit duplication step of duplicating data stored in the master side storage device at a predetermined time to the replica side storage device in block units;
In this block unit duplication process, when update occurs to predetermined file data among the data stored in the master side storage device, it is possible to identify the order in which the file update information indicating the update contents is updated. A file update information storage step for storing in the master storage device,
A file unit duplication process for duplicating the file update information to the replica storage device in the order of update after the block unit duplication process;
I have a,
In the file update information storage step, when a file update is completed in the master storage device, the update history information included in the update history information generated at the time of file update in the master storage device is updated. Storing identifiable update identification information in association with the file update information, and deleting the update history information from the master storage device,
A replication method characterized by that. The file update information storage step registers the file update information and the update history information in a journal queue in the order in which the file data stored in the master storage device is updated during the block unit replication step. A journal registration step, a journal deletion step of deleting the file update information and update history information for which file update has been completed from the journal queue, and associating the update identification information with the file update information before and after the journal deletion step File update information registration process stored in the master storage device,
The replication method according to claim 1 , further comprising: In the file unit duplication process, the master storage device deletes the file update information duplicated in the replica storage device from the master storage device,
The replication method according to claim 1 or 2 , characterized in that A master storage device connected to the replica storage device,
A block unit duplicating means for duplicating data stored at a predetermined time to a storage device on the replica side in block units;
A file that stores file update information representing the update contents so that the update order can be identified when an update occurs to predetermined file data among the stored data during replication by this block unit replication means Update information storage means;
Update history generating means for generating update history information when updating the file data;
File unit duplicating means for duplicating the stored file update information to the replica storage device in the updated order after duplication by the block duplicating means;
Bei to give a,
The update history generation means generates update identification information that can identify an update order of file updates included in the update history information,
The file update information storage means stores the update identification information in association with the file update information,
Furthermore, the update history deletion unit for deleting the update history information generated by the update history generation unit after file update is completed,
A master-side storage device characterized by that. The update history generation means associates the file update information and the update history information in the order in which the update of the file data occurred during the block unit replication, and registers them in the journal queue,
The update history deleting means deletes the file update information and the update history information that have been updated, from the journal queue,
The file update information storage means stores the file update information deleted from the journal queue by the update history deletion means and the update identification information in association with each other.
The master-side storage apparatus according to claim 4, wherein: Stored in the file update information storage means, and provided with file update information deletion means for deleting the file update information copied to the replica storage device by the file unit replication means,
6. The master-side storage apparatus according to claim 4 or 5, To the master storage device connected to the replica storage device,
A block unit duplicating means for duplicating data stored at a predetermined time to a storage device on the replica side in block units;
During the replication by this block unit duplicating means, when an update occurs to predetermined file data among stored data, the file update information indicating the update contents is stored in the apparatus so that the update order can be identified. File update information storage processing means for storing;
Update history generating means for generating update history information when updating the file data;
File unit duplicating means for duplicating the stored file update information to the replica storage device in the updated order after duplication by the block duplicating means;
Together to realize,
The update history generation means generates update identification information that can identify an update order of file updates included in the update history information,
The file update information storage means stores the update identification information in association with the file update information,
Furthermore, a program for causing the master storage device to realize an update history deleting unit that deletes the update history information generated by the update history generating unit after file update is completed. A replication system in which a master storage device and a replica storage device are connected to each other,
The master-side storage device and the replica-side storage device comprise block unit duplicating means for duplicating data stored in the master-side storage device at a predetermined time to the replica-side storage device in block units,
When the master-side storage apparatus is updated in predetermined file data among stored data during replication in block units by the block-unit replication unit, file update information indicating the update contents is updated. File update information storage means for storing the order so as to be identifiable, and update history generation means for generating update history information when updating the file data ,
After the master-side storage device and the replica-side storage device have completed duplication in units of blocks by the block-unit duplication unit, the file update information stored in the master-side storage device is updated to the replica-side storage device in the order of update. for example Bei the file amplicons means to replicate,
The update history generation means of the master storage device generates update identification information that can identify the update order of file updates included in the update history information,
The file update information storage means of the master storage device stores the update identification information in association with the file update information,
The master-side storage device includes update history deletion means for deleting the update history information generated by the update history generation means after file update is completed,
A replication system characterized by that.

Images