JP2014089573A - Back-up device, back-up program and back-up system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To create a snapshot only when consistency is maintained between backed-up data when the data is backed up in a continuous snapshot file system.SOLUTION: When consistency between entity data and a corresponding property file is obtained in a continuous snapshot file system 54 of a back-up server 5 regarding data requiring consistency such as a property file recording entity data to be backed up and meta-data belonging to the entity data, snapshot creation instructed by a back-up client 1 can prevent a snapshot from being created in the state that the consistency is impaired in the continuous snapshot file system.

Description

  The present invention relates to a technology for backing up data on a client to a backup server connected via a network.

  There is known a backup system that backs up data on a client computer while accumulating past data as a history in a backup server connected via a network. For example, Patent Document 1 applies a file system capable of acquiring continuous snapshots (for example, NILFS of Non-Patent Document 1) to automatically and implicitly generate a large number of snapshots, and history each snapshot. And a method for configuring a backup system that protects data on a client computer. Specifically, in Patent Document 1, whenever a change occurs in various states such as file contents (data), file attributes, names, and folder configurations on the client computer, the changed data is continuously transmitted to the backup server. By storing the file in the snapshot file system and repeatedly executing the same change operation on the files or folders in the continuous snapshot file system on the backup server, files and folders on client computers that do not have a history storage function can be created. It describes how to save and protect on a backup server while keeping the past state.

  In a normal file system, files edited and saved by the application are overwritten on the disk, and no past version data remains. On the other hand, in NILFS, the previous data remains on the storage device without being overwritten, and the state of the past file system can be referred to as a snapshot. NILFS can make the state (latest state or past state) of a file system at any point in time as a snapshot, and restore lost data (files and folders) from this snapshot.

  Since a backup system using a continuous snapshot file system provides functions such as history generation and integration by a file system that is part of the OS, the application program that manages it is an image file generation type backup method. The management of a plurality of backup images, the function of creating a differential or incremental backup, the function of integrating backup images, and the like as required are unnecessary. The management of backup data is consolidated into operations such as creating or deleting snapshots.

  On the other hand, there are functions that are lacking when a continuous snapshot file system is applied as a backup system. In order to be able to properly restore the data to be backed up on the client computer, not only the actual data, but also local information such as attribute information, timestamps such as creation date and update date, security information, owner information, and group information It is required to be able to save and restore file system specific metadata. Since the type and format of the metadata differ depending on the OS platform, if the local file system storing the data to be backed up and the OS of the continuous snapshot file system are different, there is a difference in the metadata that can be saved.

  In order to make it possible to restore metadata regardless of the type of OS or file system installed in the client computer, a method of storing the content of the metadata as data on a backup server is used. As a method of storing metadata in a backup server, there are a method of using an extended file attribute provided by a file system and a method of storing metadata of a single file or a plurality of files in a file. The latter is more convenient for efficiently extracting changed files.

JP 2011-204008 A

Sato et al., “Design and Implementation of Log Structured File System NILFS”, IPSJ Transactions Computing System (acs), IPSJ, 2009, Vol. 2, No. 1, pp. 110- 122

  However, if the backup server stores the file received from the client computer and the property file and creates a snapshot of the past file and past folder tree using the continuous snapshot file system, the file is being written If a snapshot is created in the above state, or a snapshot is created with only one of the file and property file written, the snapshot becomes inconsistent as a backup. When restoring files and folders on a client computer from data saved on a backup server, the correct state cannot be restored if files are restored from an inconsistent snapshot .

  The present invention has been made in view of the above, and an object of the present invention is to create a snapshot only when consistency is maintained between backed up data when backing up data to a continuous snapshot file system. And

  A backup device according to a first aspect of the present invention is a backup device that transmits data to a backup server including a continuous snapshot file system that creates and accumulates past history as a snapshot, and backs up the data to be backed up. Creation means for creating a property file describing accompanying metadata, transmission means for sending data to be backed up and a property file corresponding to the data to the backup server, the data and the property file in the backup server, And instructing means for instructing the backup server to create a snapshot when consistency is established.

  In the backup device, the instruction unit may add or not add a tag that prompts the creation of a snapshot to the data or the property file when transmitting the data or the property file. The creation of a snapshot is instructed by giving or not giving a tag that suppresses creation.

  In the backup apparatus, the instruction unit has a plurality of policies for instructing creation of a snapshot, and determines the timing of creating a snapshot by switching the policy according to a required backup type. .

  A backup program according to the second aspect of the present invention is characterized in that a computer is operated as the backup device.

  A backup system according to a third aspect of the present invention is a backup system having a backup server that holds a backup of data and a backup device that transmits data to the backup server for backup, and the backup device is a backup target. Creation means for creating a property file describing metadata attached to the data; transmission means for sending data to be backed up and a property file corresponding to the data to the backup server; and the data and the property in the backup server And an instruction means for instructing the backup server to create a snapshot when consistency with the file is obtained. The backup server creates and stores a past history as a snapshot. A continuous snapshot file system, storage means for receiving data and property files to be backed up from the backup device and storing them in the continuous snapshot file system, and receiving a snapshot creation instruction from the backup device And a management unit that causes the continuous snapshot file system to create a snapshot.

  In the backup system, when the instruction unit transmits the data or the property file, the instruction unit may or may not add a tag that prompts the creation of the snapshot to the data or the property file. The creation of a snapshot is instructed by giving or not giving a tag that suppresses creation, and the management means takes a snapshot based on the tag given to the data received by the storage means or the property file It is made to create.

  In the backup system, the instruction unit has a plurality of policies for instructing creation of snapshots, and determines the timing of snapshot creation by switching the policies according to the type of backup requested. .

  According to the present invention, when data is backed up to a continuous snapshot file system, a snapshot can be created only when consistency between the backed up data is maintained.

  According to the present invention, it is possible to avoid the overhead of specifying the timing for creating a snapshot.

  According to the present invention, it is possible to reduce the specification of useless snapshot creation timing in consideration of backup conditions, and to reduce the processing of the backup server.

It is a figure which shows the structure of the backup system in this Embodiment. It is a figure which shows the example of a property file. It is a flowchart which shows the flow of a process in which the backup system in this Embodiment backs up the property file corresponding to the actual data of backup object. It is a figure for demonstrating the timing which assign | provides the tag which urges | creates a snapshot, when performing a backup whenever a file is updated. It is a figure for demonstrating the timing which provides the tag which accelerates | stimulates snapshot creation in the case of backing up several files collectively.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram showing a configuration of a backup system according to the present embodiment. The backup system shown in FIG. 1 includes a client computer 3 and a backup server 5, and the client computer 3 and the backup server 5 are connected via a network 100. The client computer 3 has a backup target data storage area 2, and a backup client 1 that transmits the backup target data stored in the backup target data storage area 2 to the backup server 5 is incorporated. Hereinafter, the backup client 1 and the backup server 5 will be described.

  The backup client 1 includes a backup management unit 11 and a server communication unit 12.

  The backup management unit 11 includes a property file generation unit 111 and a backup timing management unit 112. The backup management unit 11 acquires entity data (file or folder) to be backed up and metadata attached to the entity data from the backup target data storage area 2, The server communication unit 12 transmits the backup data to the backup server 5 for backup. Metadata attached to the actual data to be backed up is described in a property file and transmitted to the backup server 5 for backup. In addition, the backup management unit 11 transmits to the backup server 5 a deletion request for deleting the entity data deleted from the backup target data storage area 2 and the property file corresponding to the entity data from the backup server 5. Further, the backup management unit 11 receives the result of the transmission / deletion request of the entity data or property file from the backup server 5 and retries if the request is not successful due to a communication error.

  The property file generation unit 111 acquires an existing property file stored in the backup server 5, acquires metadata attached to the actual data to be backed up from the backup target data storage area 2, and updates the property file. . If there is no existing property file in the backup server 5, a new property file is generated based on the acquired metadata. Details of the property file will be described later.

  The backup timing management unit 112 performs snapshot creation timing so that the consistency between the entity data to be backed up and the property file describing the metadata of the entity data is maintained in the snapshot created by the backup server 5. Is notified to the server communication unit 12. The backup timing management unit 112 determines a snapshot creation timing based on a backup policy described later. The backup timing management unit 112 has a plurality of policies for determining snapshot creation timing, and switches the policy to be applied according to the type of backup required (such as backup at any time, batch backup, etc.).

  The server communication unit 12 transmits and receives data to and from the backup server 5 based on instructions from the backup management unit 11. Specifically, a function for transferring the entity data and property file to be backed up to the backup server 5, a function for acquiring the property file from the backup server 5, and a function for transmitting a request for deleting the entity data and property file to the backup server 5 Have In addition, the server communication unit 12 returns the data transmission result to the backup server 5 and the result of the deletion request to the backup management unit 11. The result information returned to the backup management unit 11 includes “success” or “failure” status in addition to “success” or “failure” so that the backup management unit 11 can perform post-processing such as retry as necessary. Contains a number that can identify the reason. The reason for the failure may be when the backup server 5 returns or when the server communication unit 12 detects during communication like a communication failure.

  In addition, the server communication unit 12 includes a timing tag adding unit 121, and creates snapshots for backup target entity data, property files, and deletion requests to be transmitted to the backup server 5 according to instructions from the backup timing management unit 112. Give a tag to encourage.

  Although not shown in FIG. 1, the backup client 1 has a restore management unit connected to the server communication unit 12 and is backed up from the snapshot stored in the backup server 5 through the server communication unit 12. The property file corresponding to the entity data is acquired and restored to the backup target data storage area 2. The restore management unit may cause the user to select a history to be restored by presenting a list of snapshots (history) accumulated in the backup server 5.

  The backup server 5 includes a client communication unit 51, a file management unit 52, a snapshot management unit 53, a continuous snapshot file system 54, and a backup storage area 55.

  The client communication unit 51 transmits / receives data to / from the backup client 1. Specifically, a function for receiving the backup target entity data and property file transmitted by the backup client 1, a function for transmitting the backup target entity data and property file in the specified snapshot to the backup client 1, and a backup It has a function of receiving a delete request transmitted by the client 1. Further, the client communication unit 51 returns a success / failure result and a reason for failure in response to a request from the backup client 1.

  The client communication unit 51 includes a timing tag determination unit 511, determines a tag attached to data received from the backup client 1, and notifies the file management unit 52 of the timing for generating a snapshot. For example, if a tag that prompts snapshot creation is added to the received data, the creation of the snapshot is notified to the file management unit 52, and a tag that prompts snapshot creation is not attached to the received data The file management unit 52 is notified not to notify the file management unit 52 of the creation of a snapshot or to create a snapshot.

  The file management unit 52 accesses the backup storage area 55 through the continuous snapshot file system 54, and creates, writes, deletes, and reads files and folders. Specifically, a function for saving the data received by the client communication unit 51 in the backup storage area 55, a function for creating a folder for saving a file in the backup storage area 55, an acquisition request or a deletion request received by the client communication unit 51 Based on the above, it has a function of acquiring or deleting requested data from the latest state of the continuous snapshot file system 54 and a function of deleting an empty folder.

  Further, the file management unit 52 instructs the snapshot management unit 53 to generate a snapshot based on the timing of generating the snapshot notified from the client communication unit 51.

  The snapshot management unit 53 creates a snapshot at an opportunity given from the file management unit 52. Further, as described in Patent Document 1, a function of integrating a large number of generated snapshots automatically or manually may be provided.

  The continuous snapshot file system 54 stores data in the backup storage area 55 and accumulates past history as snapshots, like NILFS of Non-Patent Document 1.

  The backup storage area 55 is a local storage mounted on the backup server 5 or a network storage connected via a network, and the data (substance data and property files to be backed up) are stored in the past through the continuous snapshot file system 54. Keep the history of

  The units included in the backup client 1 and the backup server 5 may be configured by a computer including an arithmetic processing device, a storage device, and the like, and the processing of each unit may be executed by a program. This program is stored in a storage device included in the backup client 1 and the backup server 5, and can be recorded on a recording medium such as a magnetic disk, an optical disk, or a semiconductor memory, or provided through a network.

  Next, the property file will be described.

  The property file is a file describing metadata attached to the actual data to be backed up. The property file can be created in an arbitrary configuration unit such as a file unit, a volume unit, or a folder unit. Here, an example in which a property file is created in a folder unit will be described.

  FIG. 2 shows an example of a property file. The property file shown in the figure is described in a text format, and holds the names and metadata of files and folders (subfolders) that exist immediately under the target folder. The name and metadata are represented by [ID]. ID based on an ID uniquely assigned to each entity data (file and folder) to be backed up and an identifier indicating the attribute type. The key is expressed in the format of [identifier], and is described as a pair of values indicating the key and the key value.

  The identifiers used in the property file shown in FIG. 2 and their meanings are: FT ... file / folder identification, FN ... file name / folder name, ATR ... attribute information, CT ... creation date, AT ... update date, dsNu ... corresponding file Is the number of alternative data streams attached to dsN #, the name of the alternative data stream # (# is the number of 1 to the alternative data stream). The number of alternative data streams is an integer greater than or equal to 0, and is 0 for files and folders that do not have an alternative data stream. The alternative data stream is an auxiliary data stream different from the main data of files and folders, and is managed in the same file together with the main data. The alternative data stream may be called a fork depending on the OS. The attribute information includes attribute information depending on the file system to be backed up, such as “read only”, “hidden file”, “archive”, “compression”, and “encryption”. The time stamp such as the creation date and the update date and the attribute information are used as materials for determining whether or not the backup process and the restore process are to be performed.

  The above is an example of a property file, and any property file format can be used.

  For an empty folder that has no files and subfolders immediately below, only a property file may exist or no property file may exist in the empty folder. In this embodiment, when the folder in the backup target data storage area 2 is an empty folder, it is managed on the continuous snapshot file system 54 of the backup server 5 in association with the state in which the empty folder includes only the property file. doing.

  Next, the processing flow of the backup system in this embodiment will be described.

  In the following, (1) processing for backing up newly created files or changed files, (2) processing for creating folders, (3) processing for deleting files, (4) processing for deleting folders, (5) metadata only A process of reflecting the change of (6) and a process of simultaneously creating, changing, and deleting a plurality of files will be described.

(1) Processing for Backing Up Newly Created Files or Changed Files First, the flow of processing for backing up newly created files newly created in the backup target data storage area 2 and changed files whose data contents have been updated will be described. FIG. 3 is a flowchart showing a flow of processing in which the backup system according to the present embodiment backs up the property file corresponding to the actual data to be backed up.

  The backup management unit 11 detects a newly created file or a changed file as a backup target (step S11). As a change file detection method, a change notification from the backup target data storage area 2 is received and detected, and a property file stored in the backup server 5 is acquired, and the files and folders described in the property file are acquired. A method of comparing and detecting the configuration and metadata (time stamps such as creation date and time and update date and file size) and the current file, folder configuration and metadata stored in the backup target data storage area 2 There is. The former method of receiving and detecting a change notification is suitable for continuously monitoring the presence / absence of a change, and is suitable for backing up a changed file as needed. On the other hand, the latter method of acquiring and comparing property files is suitable for regular backups that are performed after a certain period of time because a change file can be detected regardless of the progress of the process.

  Subsequently, the backup management unit 11 acquires a property file corresponding to the detected backup target from the backup server 5 through the server communication unit 12 (step S12). In this embodiment, since the property file is created in units of folders, the property file of the parent folder in which the newly created file and the changed file exist is acquired. If the property file has already been acquired in step S11, the process in step S12 can be omitted.

  Subsequently, based on an instruction from the backup management unit 11, the server communication unit 12 transmits the detected entity data to be backed up to the backup server 5 without attaching a tag that prompts snapshot creation (step S13). The backup timing management unit 112 of the backup management unit 11 converts the newly created file, the changed file, and the property file into the newly created file and the changed file to be transmitted so that the consistency of the property file is maintained in the snapshot created by the backup server 5. Tells it not to tag. Note that the backup server 5 also transmits information on a path name composed of a combination of a file name and a folder name as information on a location for storing a newly created file and a changed file to be transmitted. There are two methods for transmitting path name information: a method of embedding in a data transfer request and a method of embedding in a resource address such as embedding in an HTTP URL. Either method may be used.

  Then, the client communication unit 51 of the backup server 5 receives the actual data to be backed up, and the file management unit 52 backs up to the backup storage area 55 through the continuous snapshot file system 54 (step S14). The newly created file and the changed file received are newly created or overwritten on the given path name of the continuous snapshot file system 54 based on the simultaneously received path name information. At this time, if the file writing fails, an error is returned to the backup client 1 through the client communication unit 51. When receiving an error, the backup client 1 retries as necessary.

  Note that the entity data received in step S14 is not attached with a tag that prompts the creation of a snapshot, so that no snapshot is created here.

  Then, the backup management unit 11 acquires the backup target metadata from the backup target data storage area 2, and updates the property file acquired from the backup server 5 in step S12 (step S15). When the backup target is a newly created file, since the information does not exist in the property file acquired from the backup server 5, each item of the acquired metadata is newly inserted. When the backup target is an update file, replace the old metadata information described in the property file with the current metadata information of the backup target update file.

  Subsequently, the server communication unit 12 transmits the updated property file to the backup server 5 with a tag for prompting creation of a snapshot based on the instruction from the backup management unit 11 (step S16). Note that the backup server 5 also transmits path name information as information on the location where the property file to be transmitted is stored. Since the property file is stored in the parent folder where the newly created file and the changed file exist, the folder name of the parent folder and the file name of the property file are combined to obtain path name information.

  Then, the client communication unit 51 receives the property file, and the file management unit 52 backs up to the backup storage area 55 through the continuous snapshot file system 54 (step S17). At this time, if the writing of the property file fails, an error is returned to the backup client 1 through the client communication unit 51.

  When a property file is stored in the backup storage area 55, the received property file is tagged with a tag that prompts the creation of a snapshot. Therefore, the timing tag determination unit 511 instructs the snapshot management unit 53 to create a snapshot, and The shot management unit 53 causes the continuous snapshot file system 54 to create a snapshot (step S18). Note that if the creation of the snapshot fails, the backup server 5 returns an error to the backup client 1.

(2) Processing to Create Folder Next, the flow of processing when a folder is created in the backup target data storage area 2 will be described. The basic flow of the process is the same as (1) the process of backing up a newly created file or a changed file, and therefore the description of the process having the same contents is omitted with reference to FIG.

  First, similarly to the processing of steps S11 and S12 of FIG. 3, the backup management unit 11 detects that a new folder has been created, and the existing property file of the parent folder in which the created new folder exists is stored in the backup server 5. Get from.

  Subsequently, the property file generation unit 111 generates an empty property file corresponding to the generated new folder, and the server communication unit 12 transmits the empty property file to the backup server 5 without attaching a tag for prompting snapshot creation. To do. As a result, the backup server 5 is requested to create a corresponding new folder. The transmitted property file includes the path name information of the folder storing the property file, that is, the path name information of the created folder.

  Then, the client communication unit 51 receives an empty property file requesting creation of a new folder, and the file management unit 52 creates a new folder corresponding to the continuous snapshot file system 54 and receives the received empty file in the new folder. Store the property file. At this time, if creation of a new folder fails, an error is returned to the backup client 1 through the client communication unit 51. When receiving an error, the backup client 1 retries as necessary.

  Since the received property file is not tagged with a snapshot for creating a snapshot, no snapshot is created here.

  Then, the backup management unit 11 acquires the metadata of the new folder created from the backup target data storage area 2 and updates the property file acquired from the backup server 5. Specifically, each item of the metadata of the created new folder is newly inserted into the property file of the parent folder acquired from the backup server 5.

  Then, similarly to the processing in steps S16 to S18 in FIG. 3, the backup management unit 11 sends the updated property file to the backup server 5 with a tag for prompting creation of the snapshot, and the backup server 5 receives the received property file. The property file is backed up to the backup storage area 55, and the continuous snapshot file system 54 creates a snapshot.

(3) Processing for Deleting Files Next, the flow of processing when a file in the backup target data storage area 2 is deleted will be described. The basic flow of the process is the same as (1) the process of backing up a newly created file or a changed file, and therefore the description of the process having the same contents is omitted with reference to FIG.

  First, similarly to the processing in steps S11 and S12 of FIG. 3, the backup management unit 11 detects that a file has been deleted, and acquires the existing property file of the parent folder in which the deleted file exists from the backup server 5. To do.

  Subsequently, the server communication unit 12 sends a deletion request for deleting the file corresponding to the deleted file from the backup server 5 to the backup server 5 without attaching a tag that prompts creation of a snapshot based on the instruction of the backup management unit 11. Send. Include the path name information of the deleted file in the deletion request.

  Then, the client communication unit 51 receives a deletion request for deleting a file, and the file management unit 52 deletes the corresponding file from the continuous snapshot file system 54. At this time, if the deletion of the file fails, an error is returned to the backup client 1 through the client communication unit 51. When receiving an error, the backup client 1 retries as necessary.

  Since the received deletion request is not tagged with a snapshot for creating a snapshot, no snapshot is created here.

  Then, the backup management unit 11 deletes the metadata of the deleted file from the property file acquired from the backup server 5, and if necessary, changes the time stamp information of the changed folder to the entry of the property file folder. To reflect.

  Then, similarly to the processing in steps S16 to S18 in FIG. 3, the backup management unit 11 sends the updated property file to the backup server 5 with a tag for prompting creation of the snapshot, and the backup server 5 receives the received property file. The property file is backed up to the backup storage area 55, and the continuous snapshot file system 54 creates a snapshot.

(4) Processing for Deleting a Folder Next, a processing flow when a folder in the backup target data storage area 2 is deleted will be described. Since the basic flow of the process is the same as (3) the process of deleting a file, the description of the process having the same content is omitted.

  First, the backup management unit 11 detects that a folder has been deleted, and acquires an existing property file of the parent folder in which the deleted folder exists from the backup server 5.

  Subsequently, the server communication unit 12 makes a deletion request for deleting the property file corresponding to the deleted folder from the backup server 5 based on an instruction from the backup management unit 11 without adding a tag for prompting creation of a snapshot. Send to. Include the path name information of the property file to be deleted in the deletion request.

  Then, the client communication unit 51 receives a deletion request for deleting the property file, and the file management unit 52 deletes the corresponding property file from the continuous snapshot file system 54. At this time, the file management unit 52 detects that the folder storing the property file on the continuous snapshot file system 54 is empty, and deletes the folder as well.

  Since the received deletion request is not tagged with a snapshot for creating a snapshot, no snapshot is created here.

  Then, the backup management unit 11 deletes the metadata of the deleted file from the property file acquired from the backup server 5, and if necessary, changes the time stamp information of the changed folder to the entry of the property file folder. To reflect.

  Similar to the processing in steps S16 to S18 in FIG. 3, the backup management unit 11 transmits the updated property file to the backup server 5 with a tag for prompting the creation of the snapshot, and the backup server 5 receives the received property file. The file is backed up to the backup storage area 55, and the continuous snapshot file system 54 creates a snapshot.

  Note that the files and folders deleted from the latest state of the continuous snapshot file system 54 in the processes (3) and (4) are not removed from the backup storage area 55. Files and folders deleted from the latest state by the history saving function of the continuous snapshot file system 54 also remain in the backup storage area 55. The deleted file or folder information is deleted from the backup storage area 55 after it is no longer referenced from all snapshots held by the continuous snapshot file system 54.

(5) Processing that Reflects Changes in Metadata Only Next, the flow of processing when only metadata is changed will be described.

  When only the metadata is changed without changing the actual data, such as the attribute change or owner change of the file or folder in the backup target data storage area 2, the backup management unit 11 sends the corresponding property file from the backup server 5. Acquire and update the acquired property file to make it current.

  Then, similarly to the processing in steps S16 to S18 in FIG. 3, the backup management unit 11 sends the updated property file to the backup server 5 with a tag for prompting creation of the snapshot, and the backup server 5 receives the received property file. The property file is backed up to the backup storage area 55, and the continuous snapshot file system 54 creates a snapshot.

  If only the property file is transmitted to the backup server 5 by changing only the metadata, there is no timing at which the consistency between the entity data and the metadata is broken. Can be selected on the backup client 1 side.

(6) Processing for simultaneously creating, changing, and deleting a plurality of files Next, the flow of processing when a plurality of files are operated simultaneously will be described.

  When a plurality of files are created or changed, a tag for prompting creation of a snapshot is not attached, and the file is transmitted to the backup server 5 for backup, and the created or changed file is written in the continuous snapshot file system 54. When a plurality of files are deleted, a deletion request for the deleted files is transmitted to the backup server 5 and the corresponding files are deleted from the continuous snapshot file system 54.

  The backup management unit 11 acquires the corresponding property file from the backup server 5 and updates the acquired property file to make it current.

  Then, similarly to the processing in steps S16 to S18 in FIG. 3, the backup management unit 11 sends the updated property file to the backup server 5 with a tag for prompting creation of the snapshot, and the backup server 5 receives the received property file. The property file is backed up to the backup storage area 55, and the continuous snapshot file system 54 creates a snapshot.

  In the above procedure, when a property file is sent, a tag that prompts the creation of a snapshot is added and sent to the backup server 5, but when an updated file is sent, the opposite meaning tag, that is, the creation of a snapshot is suppressed. You may attach a tag. Specifically, in the procedure described above, if no tag is attached, a tag that suppresses snapshot creation is attached, and if a tag is attached, that is, a property file without a tag when sending a property file. Send.

  Further, although the backup opportunity is not mentioned here, the periodic backup activated by the scheduler, the manual backup according to the direct instruction of the user, and the change contents in response to the notification from the backup target data storage area 2 are transferred to the backup server 5. It can be applied to any form of so-called real-time backup that is reflected in the above.

  In addition, for simplification of description, the description of the processing related to the alternative data stream (icon information, etc.) in the file system of the backup target data storage area 2 is omitted. Managed in the property file, the alternative data stream entity data is handled in the same way as the general file entity data. For example, the stream name is added to the end of the original file name, the file name is mapped and handled as a general file.

  In addition, when a file to be backed up is huge, even when the data is divided into a plurality of data files and transmitted to the backup server 5, a snapshot is not created until all the divided files are stored in the backup server 5. As applicable.

  Furthermore, the present invention can also be applied to differential backup in which only a part of a file is transmitted to the backup server 5 to suppress the communication data amount when the backup target is partially changed.

  Next, a policy for giving a tag that prompts creation of a snapshot will be described.

4 and 5 are diagrams for explaining the timing of adding a tag that prompts creation of a snapshot. In each figure, F _j (j = 1, 2, 3,...) Indicates a file to be backed up, and P {F _j } indicates a property file in which metadata of the file j is described. P {F _i , F _j } represents a property file in which metadata of both the files F _i and F _j are described. When an arrow t _n (n = 1, 2, 3,...) Is specified in the property file in the figure, a tag that prompts creation of a snapshot when the property file is transmitted to the backup server 5 Indicates that it is assigned.

  First, with reference to FIG. 4, a description will be given of the timing of adding a tag that prompts the creation of a snapshot when one or several files are updated from time to time and the updated file is backed up each time.

  FIG. 4 shows that every time it detects that a file is updated in the backup target data storage area 2, the updated file is transmitted to the backup server 5 for backup, and the continuous snapshot file system 54 stores the backup target data storage. This is an example of timing for giving a tag that prompts the creation of a snapshot when a snapshot is created at a frequency similar to that of a file update in area 2.

In the example shown in FIG. 4, the file F _j (j = 1, 2, 3,...) To be backed up is not attached with a tag that prompts snapshot creation, and the property files P {F _j , F _k,. ... Only a tag that prompts the creation of a snapshot is attached. As a result, the property file describing the file and the metadata of the file is stored in the continuous snapshot file system 54, and the snapshot is created in a state where the consistency between the two is maintained.

  Next, with reference to FIG. 5, a description will be given of the timing at which a tag that prompts creation of a snapshot when a plurality of files are backed up collectively.

  In FIG. 5, a plurality of updated files in the backup target data storage area 2 are collectively sent to the backup server 5 for backup, and all the updated files and property files are stored in the continuous snapshot file system 54. This is an example of timing for adding a tag that prompts creation of a snapshot when a snapshot is created at the time of creation.

In the example shown in FIG. 5, the backup target file F _j (j = 1, 2, 3,...) And its property file P {F _j , F _k ,. When a series of backups are completed, that is, when the property file P {F ₉₉ , F ₁₀₀ } is transmitted to the backup server 5, a tag that prompts creation of a snapshot is added. As a result, a snapshot is created at the timing when a series of backups are completed.

  The backup server 5 only controls the timing of creating a snapshot based on tag information without being conscious of the content and type of the received file (including the distinction between the actual data and property file to be backed up). Therefore, it does not depend on the OS of the backup client 1 or the file system type of the data storage area 2 to be backed up, and in the sense that application of snapshot creation timing control by other policies will be possible in the future. It is a method with versatility and expandability.

  In this embodiment, the client computer 3 having the backup target data storage area 2 is provided with the backup client 1. However, the client computer 3 that acquires and restores the backup target through a file sharing protocol such as SMB or CIFS. The backup client 1 may be incorporated into another agentless backup device.

  In this embodiment, the data transmitted from the backup client 1 is stored in the backup storage area 55 managed by the continuous snapshot file system 54 in the backup server 5. It has a temporary storage area managed by a normal file system that does not have a creation and storage function. Data that is not assigned a tag that prompts snapshot creation is stored in the temporary storage area, and a tag that prompts snapshot creation is assigned. When the received data is received, it can be stored in the continuous snapshot file system 54 together with the data stored in the temporary storage area to create a snapshot.

  As described above, according to the present embodiment, continuous snapping of the backup server 5 is performed on data that requires consistency, such as a property file that describes entity data to be backed up and metadata attached to the entity data. When consistency is obtained between the entity data and the corresponding property file in the shot file system 54, the backup client 1 issues an instruction to create a snapshot, so that the consistency in the continuous snapshot file system is impaired. A snapshot can be prevented from being created.

  Also, in this embodiment, there is no specific assumption about the configuration between files to be backed up. For example, a property file attached to a file is saved in a file with the same name with a different extension and backed up. Since the server 5 does not handle the data as a single entity, there is an n-to-1 relationship between data that should maintain consistency, as in the case where metadata of a plurality of entity data is managed together in one property file. In some cases, the backup client 1 issues a snapshot creation instruction to create a snapshot in which data consistency is maintained. Since the backup server 5 has only to determine the creation of a snapshot based on an instruction from the backup client 1, there is no need to place a specific premise on the configuration between the data to be backed up or to know the contents of the data.

  In addition, according to the present embodiment, for example, a client computer can be created by adding a tag for prompting snapshot creation to backup target data to be transmitted to the backup server 5 or a deletion request and controlling the snapshot creation timing. 3 If the files changed in 3 above are backed up at any time, and each time a snapshot is created, the number of communications can be reduced compared to the method of independently sending a snapshot creation request, and in a WAN environment with a large communication delay Therefore, it is possible to avoid a decrease in performance due to the communication of the snapshot creation request.

DESCRIPTION OF SYMBOLS 1 ... Backup client 11 ... Backup management part 111 ... Property file generation part 112 ... Backup timing management part 12 ... Server communication part 121 ... Timing tag provision part 2 ... Backup object data storage area 3 ... Client computer 5 ... Backup server 51 ... Client Communication unit 511 ... Timing tag determination unit 52 ... File management unit 53 ... Snapshot management unit 54 ... Continuous snapshot file system 55 ... Backup storage area 100 ... Network

Claims (7)

A backup device that transmits data to a backup server including a continuous snapshot file system that creates and accumulates past history as a snapshot, and backs up the backup server.
A creation means for creating a property file that describes metadata accompanying the data to be backed up;
Transmission means for transmitting data to be backed up and a property file corresponding to the data to the backup server;
An instruction means for instructing the backup server to create a snapshot when the data and the property file are consistent in the backup server;
A backup apparatus comprising: When the instruction means transmits the data or the property file, the tag for giving or not giving the snapshot for creating the snapshot to the data or the property file, or the tag for suppressing the creation of the snapshot The backup apparatus according to claim 1, wherein the creation of a snapshot is instructed by giving or not giving. The said instruction | indication means has multiple policies which instruct | indicate creation of a snapshot, determines the timing of snapshot creation by switching the said policy according to the kind of backup requested | required. The backup device described in 1.   A backup program for operating a computer as the backup apparatus according to claim 1. A backup system having a backup server that holds a backup of data and a backup device that transmits data to the backup server for backup,
The backup device is
A creation means for creating a property file that describes metadata accompanying the data to be backed up;
Transmission means for transmitting data to be backed up and a property file corresponding to the data to the backup server;
An instruction means for instructing the backup server to create a snapshot when the data and the property file are consistent in the backup server;
The backup server
A continuous snapshot file system that creates and accumulates past history as snapshots, and
Storage means for receiving data and property files to be backed up from the backup device and storing them in the continuous snapshot file system;
And a management unit that causes the continuous snapshot file system to create a snapshot when receiving an instruction to create a snapshot from the backup device. When the instruction means transmits the data or the property file, the tag for giving or not giving the snapshot for creating the snapshot to the data or the property file, or the tag for suppressing the creation of the snapshot Instructing the creation of a snapshot by giving or not giving
The backup system according to claim 5, wherein the management unit creates a snapshot based on the data received by the storage unit or the tag attached to the property file. The said instruction | indication means has two or more policies which instruct | indicate creation of a snapshot, determines the timing of snapshot creation by switching the said policy according to the kind of backup requested | required. The backup system described in.

Images