JP2013037611A - Storage device, storage method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify the operation of the backup of data.SOLUTION: A flash part 305 instructs a client device 100 which performs access to flash data on the basis of connection information. A memory 302 temporarily stores the data flashed in response to the instruction. A data protection part 307 stores mapping information which has captured the still point of the data stored in the memory 302 on the basis of the connection information. A snap shot part 309 creates a snap shot on the basis of the mapping information stored by the data protection part 307. A storage part 400 stores the snap shot created by the snap shot part 309.

Description

  The present invention relates to a storage device, a storage method, and a program that store data, and more particularly, to a storage device, a storage method, and a program that store data transmitted via a communication network.

  A snapshot function exists as a backup function of a general NAS (Network Attached Storage).

  The snapshot function is a function that is generally used to save a data image of a file at a static point of data on the NAS (mapping information such as data and storage area blocks; hereinafter referred to as mapping information). is there. After that, when there is an update to the file, the mapping information of the saved data is largely unchanged, the update data is written to the new area, the file mapping information is changed, and the update data is written to the existing area In some cases, the pre-update data is written to another area, and the mapping information of the saved snapshot data is changed. In the following description, the former is a write-once type and the latter is an extrusion type. Which is used depends on the specifications of each NAS, but the data at the time of snapshot creation can be accessed even after the data is updated.

  Since the load at the time of creating a snapshot is small, in recent NAS, there are many that use a snapshot in which the former update data is added to a new area, and in the following explanation, it is described as a snapshot without any particular description. In this case, the former update data is a write-once-type snapshot in which data is added to a new area. Other detailed specifications differ depending on each NAS, but detailed description is omitted here.

  The snapshot function is a function that can perform backup without imposing a heavy load. Therefore, although it is a function generally used in NAS, there is one major drawback. In general, it does not guarantee client data.

  In general, the processing on the NAS is performed in a stateless manner. However, if all processing from the client to the NAS is performed in a stateless manner, the access performance is poor, and particularly if the network quality is poor, a large performance degradation occurs.

  In order to eliminate this drawback, the file system does not directly write the update information directly to the HDD (Hard disk drive) or the like, but temporarily stores the data on the memory buffer, and then stores the data in a block of a predetermined size. When data is written on the NAS by update processing of an application on the client or the like so that the processing speed can be increased by collectively writing to the HDD or the like, the data is once buffered on the client, and then the data is predetermined. The performance is improved by writing the data to the NAS in blocks of the same size.

  This is basically common to NAS access protocols such as NFS (Network File System) and CIFS (Common Internet File System), which are representative protocols for NAS access.

  Here, it is necessary to pay attention to the fact that the snapshot is executed by paying attention only to the data on the NAS because the stationary point of the data on the NAS and the storage of the data are executed from the viewpoint of the NAS device. is there.

  When creating a snapshot on the NAS, the client side is generally not notified of the creation of the snapshot. Therefore, the update processing of the buffering data at the time of NAS writing on the client and the data on the memory waiting to be written to the buffering is completed, and the data (hereinafter referred to as buffer data) Such as writing) is not performed. As a result, the snapshot data has a quiesce point for the NAS file system, but the file itself is a buffer depending on the buffering status of the client and the status of the packet written to the NAS on the network. There is a possibility that a snapshot may be taken in a state where a quiesce point of the file is not secured, such as during data writing. In this case, the file data may be incorrect, or the file may be damaged due to inconsistency between the metadata and the actual data.

  In addition, even if the quiesce point of the file is secured, if the buffer data has not yet been written, the data that has been updated on the client before creating the snapshot is saved as a snapshot. It also happens that it is not done.

  In particular, the occurrence of an abnormality in the state of a file as in the former can be said to be a fatal defect in the snapshot function that is a backup performed for the purpose of ensuring normal data.

  Usually, these drawbacks are avoided by creating multiple generations of snapshots.

  As described above, snapshots only need to secure storage capacity for data updates, and the snapshot creation load is lower than the general replication function that copies the entire area, and more than 100 generations with many NAS. It is possible to create a snapshot. Therefore, it is common to create multiple generations when using snapshots.

  At this time, if 100 generations are created, the file is updated at every timing of creating a snapshot for a general text file or the like that is directly operated by a human being, and the quiesce point of the file is not secured. The probability is very low, and even if the data cannot be restored from the desired timing, it is possible to minimize the damage by restoring from the nearest point.

  In addition, the processing of the file system on the NAS is converted into a transaction so that the data in the write command is not reflected, or dedicated software is installed in the client to secure a quiesce point of data on the client side, and then the snapshot is taken There is also a NAS that performs such as creation, and a means for preventing file destruction as described above is prepared (for example, see Patent Document 1).

JP 2009-146389 A

  However, each workaround described above has its own problems.

  Regarding the means for storing a plurality of generations, even if the required capacity is small and it is possible to hold 100 generations or more as a functional specification, storing several tens to more than one hundred leads to an increase in management and operational load. For this reason, most operations are to store only a few generations to a dozen generations at most. In that case, the probability of accessing each time when creating a snapshot is still low, but since the snapshot creation interval is widened, there is a possibility that the most recent restoration is insufficient.

  In addition, since nothing is done with respect to the writing of the buffer data, although the possibility is low, the destruction of the file or the unreflection of the stored data itself occurs.

  In addition, in the case of a workaround by making the file system transactional, data that has not been written for update of the file on the NAS is not accepted as legitimate data. It is possible to prevent snapshot data from being updated and being created as an incomplete file.

  However, if the buffer data write command is divided into blocks that are transferred over the network, etc., and processed as multiple write commands, it may not be said that the quiesce point is still secured for the file. Therefore, it is determined that the portion for which the write command has been completed is normal, and a snapshot is created and incomplete file data is saved. Therefore, it is not a sufficient measure. In addition, data cannot be stored in a state that accurately reflects the update completion data, resulting in data loss.

  In addition, when software for cooperation with NAS is used on the client, data loss and file destruction can be prevented because the buffer data is reflected.

  However, these softwares are mostly linked with functions such as a database that secure a stationary point independently, and are often difficult to use in a file server, which is a general use of NAS.

  Some software can also be used for file servers, but operations are performed from the client side. For this reason, the software for cooperation basically pays attention only to the data of the installed client, and does not have a function to cooperate with other clients. For this reason, if the same file system is used by a plurality of clients, the consistency of data of other clients cannot be guaranteed even if a snapshot is created using the cooperation software. Even if other clients can be linked, the maximum number of clients is the limit. Since there are hundreds to tens of thousands of clients in a general file server, it is practically very difficult to create a snapshot with all the clients securing a quiesce point at the same time. Therefore, it is recommended to use the file system in a divided manner so that it is not necessary to link with other clients when using these linkage software. However, there are many clients like this. In an environment, the number of file systems increases by that number, so it can be said that it is extremely difficult to operate after all.

  The technique described in Patent Document 1 also has the same problem because it does not secure a stationary point from NAS.

  The objective of this invention is providing the memory | storage device, the memory | storage method, and program which solve the subject mentioned above.

The storage device of the present invention
A storage device for storing data,
A storage device for storing data,
A flash unit for instructing the client device holding the data to flush the data to the storage device;
A temporary holding memory for temporarily holding data flushed from the client device in response to the command;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, mapping information with a quiesce point of the data held in the temporary holding memory is obtained. The data protection part to save,
A snapshot unit for creating a snapshot based on the mapping information stored by the data protection unit and the flash completion information;
A storage unit for storing a snapshot created by the snapshot unit;
The flash unit issues the command to the accessing client device based on the connection information.

The storage method of the present invention includes:
A storage method for storing data in a storage device,
A process of instructing a client device that is accessing the storage device among the client devices that hold the data to flush the data to the storage device;
Processing to temporarily hold data flushed from the client device in response to the command;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, the mapping information with the quiesce point of the temporarily held data is saved. Processing to
A process of creating a snapshot based on the stored mapping information and the flash completion information;
A process of storing the created snapshot is performed.

The program of the present invention is
A program for causing a storage device for storing data to be executed,
A procedure for instructing a client device that is accessing the storage device among the client devices that hold the data to flush the data to the storage device;
Temporarily holding data flushed from the client device in response to the instructions;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, the mapping information with the quiesce point of the temporarily held data is saved. And the steps to
A procedure for creating a snapshot based on the stored mapping information and the flash completion information;
And a step of storing the created snapshot.

  In the present invention, it is possible to simplify the operation of data backup.

It is a figure which shows 1st Embodiment of the memory | storage device of this invention. FIG. 2 is a block diagram illustrating an example of an internal configuration of a NAS device and a client device illustrated in FIG. 1. It is a figure which shows an example of the flash map shown in FIG. It is a figure which shows an example of the cold map shown in FIG. FIG. 10 is a schematic diagram when update processing is performed from a client device that has already acquired buffer data during flash processing. FIG. 10 is a schematic diagram when update processing is performed from a client device that has already acquired buffer data during flash processing. It is a flowchart for demonstrating the process from the snapshot preparation start to the flush of the buffer data on a client apparatus among the storage methods in this form. It is a flowchart for demonstrating the update process of the client apparatus in a flash process among the storage methods in this form. It is a flowchart for demonstrating the process which produces a snapshot among the storage methods in this form. FIG. 10 is a schematic diagram when update processing is performed from a client device that has already acquired buffer data during flush processing in the second embodiment. It is a figure which shows an example of the flash map in 3rd Embodiment.

  Embodiments of the present invention will be described below with reference to the drawings.

  The feature of the present invention is that, regarding NAS, which is a storage system used via a network, the quiesce point of data in the backup target area is displayed on the NAS when using a write-once snapshot in an area used by a plurality of clients. The data that has been updated on the client is flushed and written on the NAS to secure a quiesce point of the file data and to store the latest data.

  In the following description, regarding general terms and technologies such as the NAS device and the snapshot function, the omissible portions are the same as the general ones, and a detailed description is not given, and those related to the present invention are mainly described. Explained.

  FIG. 1 is a diagram showing a first embodiment of a storage device according to the present invention.

  In this embodiment, the NAS device 1 that is a storage device of the present invention is realized as a storage device having at least one NAS device controller unit 300 and a storage unit 400.

  The NAS device 1 is connected to at least one client device 100 that uses the NAS device 1 via the network 200.

  The client device 100 is a communication device operated by a user that has data and has a communication function.

  The network 200 is a communication network configured using, for example, the Internet protocol (IP).

  FIG. 2 is a block diagram showing an example of the internal configuration of the NAS device 1 and the client device 100 shown in FIG.

  As shown in FIG. 2, the client apparatus 100 shown in FIG. 1 includes a NAS buffer area 101, a memory 102, an application unit 103, an input / output module 104, and a flash execution unit 105.

  The NAS buffer area 101 is buffer data for NAS access on the client apparatus 100, and the performance is improved by buffering the NAS access data therein.

  The memory 102 is a memory area on the client device 100. The memory 102 is an area used by all applications on the client device 100, and holds buffer data for all accesses of the client device 100 including NAS access to improve performance.

  The application unit 103 is software that runs on the client device 100 and executes access to the NAS area.

  The input / output module 104 performs I / O output / input processing between the NAS device 1 and the client device 100.

  The flash execution unit 105 issues a write instruction to the NAS area for the updated data to the NAS buffer area 101, the memory 102, and the application unit 103 after receiving a flash execution instruction from the flash unit 305, and writes Is completed, the flash unit 305 has a function of notifying the completion of writing. Thereby, the NAS device controller unit 300 can determine whether the data to be written is buffer data or update completion data that is not buffer data.

  The flash execution unit 105 holds the update completion data on the memory 102 so that the update completion data is not written to the NAS area when the update process is newly confirmed during the flash process and writing to the NAS area occurs. When a notification of completion of buffer data acquisition is received from the flash unit 305 after the completion of the flash processing, it has a function of writing to the NAS area. As a result, writing of additional update completion data during the flash process can be prevented, and a quiesce point of the file can be secured.

  In addition, when the client device 100 has a plurality of connections to the NAS device 1, the flush execution unit 105 performs the buffer data flush processing on all the connections, and updates until the flush processing is completed. It has a function of waiting for writing completion data and sending a flash completion notification after flushing is completed for all connections. As a result, when there are a plurality of connections, only some of the connections have been flushed, but a flush completion notification is sent, and the NAS device 1 recognizes that there is no buffer data on the client device 100, and thereafter It is possible to prevent the update prevention process from moving.

  As shown in FIG. 2, the NAS device controller unit 300 shown in FIG. 1 includes an input / output module 301, a memory 302, a main controller 303, a disk controller 304, a flash unit 305, and a flash map 306. A data protection unit 307, a cold map 308, and a snapshot unit 309 are provided.

  The input / output module 301 performs I / O output / input processing between the NAS device 1 and the client device 100.

  The memory 302 is a processing area in the NAS device controller unit 300. The memory 302 is also a temporary storage memory that improves response time by returning a response after temporarily storing the writing from the client device 100 here.

  The main controller 303 executes all basic function processing related to NAS access, and is a control unit of the NAS device controller unit 300.

  The disk controller 304 manages the volume of the storage unit 400, receives an access command to the storage unit 400 from the main controller 303, and executes I / O processing.

  The flash unit 305 has a function of acquiring connection information with the client device 100 from the main controller 303 and creating a flash map 306.

  In addition, the flash unit 305 has a function of confirming the created flash map 306 and issuing a command to write updated information to the flash execution unit 105 of the client device 100.

  Further, when the flash unit 305 receives a buffer data write completion notification from the flash execution unit 105, the flash unit 305 has a function of describing and updating that the update completion data has been received in the information of the client device 100 in the flash map 306.

  Further, the flash unit 305 has a function of transmitting a buffer data acquisition completion notification to the flash execution unit 105 on the client device 100 after the description of the buffer data acquisition information in the flash map 306 is completed.

  Further, the flash unit 305 has a function of instructing the snapshot unit 309 to execute creation of a snapshot when the flash processing of all the client apparatuses 100 is completed.

  In addition, the flash unit 305 has a function of deleting the flash map 306 that is no longer needed upon receiving a notification of completion of snapshot creation from the snapshot unit 309.

  The flash map B6 and the cold map B8 will be described with reference to FIGS. 3 and 4, respectively.

  FIG. 3 is a diagram showing an example of the flash map 306 shown in FIG.

  As shown in FIG. 3, the flash map 306 describes information for performing buffer data flush processing, and describes a connection number, a connection destination IP address, a connection destination port number, and flash completion information.

  In the connection number, a number uniquely assigned to determine the connected client device 100 is described. Thereby, the client device 100 is uniquely identified.

  In the connection destination IP address, the IP address of the currently connected client device 100 is described. The flash processing information is notified to the IP address described here.

  In the connection destination port number, the port number of the currently connected client device 100 is described. Similar to the IP address, the flash processing information is sent to the port number described here.

  The flash completion information describes whether or not the flash process has been completed. For example, as shown in FIG. 3, “ON” and “OFF” may be used to indicate whether or not the flash process has been completed. By confirming this information, the data protection unit 307 can determine whether or not the client device 100 has already been flushed with respect to the update processing from the client device 100 being flushed, and can perform the processing.

  FIG. 4 is a diagram showing an example of the cold map 308 shown in FIG.

  As shown in FIG. 4, the cold map shown in FIG. 2 describes information about still point image information after buffer data acquisition. Specifically, file meta information, file still point mapping information, and snapshot check are described.

  In the meta information of the file, meta information (file name, update information, etc.) of the file on the updated NAS is described. By confirming this, the file that passes the still point mapping information to the snapshot unit 309 is determined.

  The file quiesce point mapping information describes the mapping information of the file with the quiesce point including the metadata after the buffer data is acquired. By passing the mapping information of the quiesce point described here to the snapshot unit 309, the file image with the quiesce point at the time of buffer data acquisition can be saved. Therefore, even if the same file is updated a plurality of times after the creation of the snapshot after obtaining the buffer data, it will not be updated once described.

  The snapshot check describes whether the mapping information of the target file has been passed to the snapshot unit 309. For example, as shown in FIG. 4, “ON” and “OFF” may be used to indicate whether the mapping information of the target file has been passed to the snapshot mechanism. By confirming this information, it is determined which file mapping information should be passed to the snapshot unit 309.

  The data protection unit 307 cooperates with the main controller 303 to check the flash map 306 for the update process during the buffer data flush process. When the client apparatus 100 has not completed the flush process, the data protection unit 307 performs the normal process. It has a function of instructing the main controller 303 to perform update processing.

  Also, the data protection unit 307 checks the flash map 306, and in the case of update processing from the client device 100 that has already received the buffer data, saves the mapping information of the update target file and the meta information of the file, and then updates it. Data has a function of instructing the main controller 303 to write to a new area.

  Further, the data protection unit 307 has a function of creating a cold map 308 for storing the acquired mapping information and meta information, and describes the stored mapping information and meta information of the pre-update file in the cold map 308. It also has a function.

  In addition, when the file described in the cold map 308 is further updated, the data protection unit 307 is similar to the update process after acquiring the buffer data described above when updating to the area where the still image is held. The main controller 303 is instructed to write to the new area, and in the case of re-updating to the area already written in the new area or appending to the new area, the main controller 303 performs the same update and append as usual. 303 has a function to instruct 303.

  In addition, the data protection unit 307 receives an inquiry about static point information when all the flash processing is completed and the snapshot unit 309 creates a snapshot, and after acquiring the static point described in the cold map 308 It has a function of passing the information of the updated file and the file quiescing point mapping information, which is the file mapping information when acquiring the quiesce point, to the snapshot unit 309 and checking that the information has been passed.

  Further, when the data protection unit 307 receives a notification of completion of creation of a snapshot from the snapshot unit 309, the data protection unit 307 has a function of deleting the cold map 308 that is no longer needed when the cold map 308 exists.

  The snapshot unit 309 has a function of creating a snapshot. The creation of the snapshot is the same as a general creation method, so a detailed description is omitted.

  Further, the snapshot unit 309 has a function of instructing the flash unit 305 to execute the buffer data flush processing of the client device 100 when receiving a snapshot creation command from the main controller 303.

  When the snapshot unit 309 receives a notification from the flash unit 305 that all the client devices 100 have completed the flash process, the snapshot unit 309 checks the data protection unit 307 to determine whether or not there has been an update after acquiring buffer data. Acquires the mapping information and file information of the file in the state where the quiesce point has been acquired, and the mapping information acquired from the data protection unit 307 is not the mapping information at the time of creation for the file updated at the time of creating the snapshot. It has a function to create a snapshot using it. At this time, for a file that has not been updated, the same processing as the normal snapshot processing is performed.

  Further, the snapshot unit 309 has a function of transmitting a snapshot creation completion notification to the flash unit 305 and the data protection unit 307 when a snapshot can be created.

  FIG. 5 and FIG. 6 are schematic diagrams in the case where there is an update process from the client device 100 that has already acquired the buffer data during the flush process.

  As shown in FIGS. 5 and 6, when a part of the storage unit 400 is a block from “1” to “15”, file mapping information when “1” to “4” have already acquired buffer data It is the acquired information of the stationary point.

  Thereafter, when there is an update process, the main controller 303 executes a write inquiry to the data protection unit 307. The data protection unit 307 that has received the inquiry from the main controller 303 confirms the flash map 306. If the update is from the client device 100 that has already acquired the buffer data, the mapping information (“1”, “2”, “3”, “4”) and the file meta information are stored, the update prevention to the saved area and the appending instruction to the new area are passed to the main controller 303, the mapping information before the update and the file information Meta information is described in the cold map 308. At this time, if the cold map 308 has not been created, the cold map B8 is created first and then described.

  The main controller 303 that has received an instruction to prevent update to the saved area and an additional write to the new area from the data protection unit 307 writes the update data to the new area and updates the file mapping information.

  Thereafter, when all the flash processes are completed and the process proceeds to the snapshot creation process, the snapshot unit 309 inquires of the data protection unit 307 about the update process after the completion of the flash.

  Upon receiving the inquiry, the data protection unit 307 confirms the cold map 308, and acquires the meta information of the file that has been updated after the completion of the flush process and the mapping information before the update.

  The data protection unit 307 that acquired the information from the cold map 308 passes the acquired information to the snapshot unit 309.

  The snapshot unit 309 that has acquired the meta information of the updated file and the mapping information before the update creates a snapshot based on the received information.

  In the following, an operation of acquiring a quiesce point of a file of a client device and creating a snapshot using the information, which is a storage method in this embodiment, will be described.

  First, processing from the start of snapshot creation to the buffer data flush on the client device will be described.

  FIG. 7 is a flowchart for explaining the processing from the start of snapshot creation to the flushing of buffer data on the client device in the storage method according to this embodiment.

  Here, the snapshot creation and the connection protocol operation between the client and the NAS device are the same as those in the general technique.

  First, in step S1, the main controller 303 issues a snapshot creation instruction to the snapshot unit 309.

  Upon receiving the creation instruction, the snapshot unit 309 instructs the flash unit 305 to flush the buffer data of the client device 100 in step S2.

  Then, the flash unit 305 that has received the instruction acquires connection information of the currently connected client device 100 from the main controller 303 in step S3.

  Subsequently, when the connection information is acquired, the flash unit 305 creates the flash map 306 shown in FIG. 3 in step S4.

  After creating the flash map 306, in step S5, the flash unit 305 instructs each client device 100 connected to the NAS device 1 to flush the buffer data based on the connection information.

  The flush execution unit 105 of the client device 100 that has received the instruction instructs the application unit 103 or the like on the client device 100 to write the buffer data in step S6.

  When the buffer data is actually written, in step S7, when sending the buffer data, the flash execution unit 105 sends a message (flash start notification) to the flash unit 305 to inform the buffer unit 305 that it will send the buffer data. Send.

  After sending the flash start notification, the flash execution unit 105 checks in step S8 whether update completion data is generated during execution of the flash process, and checks whether there is new update completion data.

  If there is new update completion data, in step S9, the data is not written immediately but is temporarily secured in the memory.

  If there is no update completion data, the flash execution unit 105 completes the flash processing as it is, and if there is update completion data, the flash execution unit 105 completes the flash processing while securing data on the memory, and then in step S10, A flash completion notification is sent to the flash unit 305.

  After receiving the flash completion notification, the flash unit 305 describes the completion in the flash map 306 and sends a buffer data acquisition completion notification to the flash execution unit 105 in step S11.

  In step S12, the flash execution unit 105 that has received the buffer data acquisition completion notification checks whether there is update completion data on the client device 100, and if not, ends the processing. On the other hand, if there is, the writing of the update completion data is resumed.

  Next, update processing of the client device 100 during flash processing will be described.

  FIG. 8 is a flowchart for explaining the update process of the client apparatus 100 during the flash process in the storage method according to the present embodiment.

  First, in step S21, it is confirmed whether there is an update process.

  If there is no update process, the process ends because there is no particular need to perform the process.

  On the other hand, if there is an update process, in step S22, the main controller 303 inquires the data protection unit 307 about the update process written, and the data protection unit 307 confirms the flash map 306.

  In step S <b> 23, it is confirmed on the flash map 306 whether or not the flash processing of the client device 100 performing the update processing is completed.

  If the flash process has not been completed, in step S24, the data protection unit 307 instructs the main controller 303 to perform the same process as the normal process.

  On the other hand, if the flash process has been completed, the data protection unit 307 stores the meta information of the file to be updated and the map information before the update in step S25, and sends the update information to the main controller 303 as a new area. To write to. The main controller 303 instructed to write to the new area updates the meta information and map information of the target file.

  In step S26, the data protection unit 307 storing the pre-update meta information and map information confirms whether the cold map 308 has been created.

  If the cold map 308 has not been created, the data protection unit 307 creates the cold map 308 in step S27.

  When the cold map 308 is created, in step S28, the meta information and pre-update map information of the acquired file are described in the cold map 308, stored, and the process is terminated.

  Thereby, it is possible to secure data with a stationary point while continuing the update process.

  Next, a process for creating a snapshot will be described.

  FIG. 9 is a flowchart for explaining a process of creating a snapshot in the storage method according to this embodiment.

  First, in step S31, the flash unit 305 confirms the flash map 306.

  After the confirmation, it is determined in step S32 whether or not the flash processing is completed from all the connections. If it is not completed, snapshot creation is not performed and the completion of the flash processing is waited.

  If the flush processing has been completed for all connections, the snapshot unit 309 starts creating a snapshot in step S33.

  When the creation of the snapshot is started, the snapshot unit 309 confirms whether there is no pre-update information in the data protection unit 307 in step S34.

  In step S35, the data protection unit 307 checks whether the cold map 308 exists.

  If the cold map 308 does not exist, in step S36, the data protection unit 307 instructs the snapshot unit 309 to perform the same processing as normal snapshot creation, and the snapshot unit 309 is currently recognized. Create a snapshot for a file.

  When the snapshot is created, the snapshot unit 309 sends a snapshot creation completion notification to the flash unit 305 and the data protection unit 307 in step S37.

  When acquiring the snapshot creation completion notification, the flash unit 305 deletes the flash map 306 and ends the processing in step S38. At this time, since the cold map 308 does not exist, the data protection unit 307 does not delete the cold map 308.

  On the other hand, if the cold map 308 exists, in step S39, the data protection unit 307 passes the updated file information and the pre-update map information from the cold map 308 to the snapshot unit 309.

  In step S40, the snapshot unit 309 that has acquired the information of the file that has been updated after the flash process and the map information before the update uses the acquired map information for the file that has been updated after the flash process, and other information. For files that are not updated after the flash processing, map information at the time of creation is acquired and snapshots are created in the same way as in normal processing.

  If the snapshot can be created, the snapshot unit 309 sends a snapshot creation completion notification to the flash unit 305 and the data protection unit 307 in step S41.

  When the snapshot creation completion notification is acquired, in step S33, the flash unit 305 deletes the flash map 306, and the data protection unit 307 deletes the cold map 308 and ends the processing.

  Through the processing described above, creation of a snapshot with a stationary point secured is realized.

  Next, a second embodiment of the storage device of the present invention will be described.

  In the present embodiment, the same configuration and the same processing as in the first embodiment are performed. The difference is that the snapshot function to be used is not a write-once type but an extrusion type snapshot. Therefore, the flow of data protection mechanism processing is different.

  FIG. 10 is a schematic diagram when update processing is performed from the client device 100 that has already acquired buffer data during the flush processing in the second embodiment.

  As shown in FIG. 10, when update data is transmitted from a connection in which buffer data has already been written, the update data is not written to another area, but already stored data is written to the snapshot area 410. .

  After that, when creating a snapshot, the mapping information at the completion of writing the buffer data is passed to the snapshot unit 309 as in the write-once type. If the second embodiment is used, the present invention can be used not only in a write-once type but also in an extrusion type snapshot.

  Next, a third embodiment of the present invention will be described in detail with reference to the drawings.

  In the third embodiment, certain information is added to the flash map 306 used in the first embodiment and the second embodiment.

  FIG. 11 is a diagram illustrating an example of a flash map according to the third embodiment.

  As shown in FIG. 11, the information added in the third embodiment is latency information indicating a signal transmission delay time.

  When the flash unit 305 collects connection information, it simultaneously measures the network latency between the NAS device 1 and the client device 100 and describes the information in the flash map 306. Thereafter, when the flash unit 305 notifies the flash execution unit 105 on the client device 100 to start the flash process, a flash command is notified from the client device 100 having a large latency. By doing so, the writing of the buffer data is started from the client device 100 having a large latency, and the flush processing time can be shortened by minimizing the delay of the flush processing due to the buffer data writing waiting. As a result, the snapshot can be quickly created, and the backup time can be shortened.

  As described above, the present invention has the following effects.

  The first effect is that data that has been updated on the client device at the time of creating the snapshot can be saved.

  The reason is that, normally, update processing on the client device is completed, and even if data is stored in a file or the like, it is not immediately reflected on the NAS device, but is buffered like a file system buffer on the client device. However, even if the data has already been updated at the start of snapshot creation, it is not always stored in the snapshot data. However, in the present invention, the buffer data of the client device is stored in the client device at the start of snapshot creation. This is because the data that has been updated on the client device can be saved by flashing it on the side and writing it on the NAS device. Even if there is a possibility that the storage of data is shifted depending on the timing of the flash, the error is still very small compared to the existing snapshot.

  The second effect is that file damage in the snapshot data can be prevented. As a result, when restoring from a file in the snapshot data, it is not possible to restore due to file corruption.

  The reason is that the update of the file is completed, and the update data of the metadata and the actual data is once stored in the buffer of the client device and then written to the NAS device. Since data is not written while considering consistency in writing buffer data to the NAS device, if a snapshot is created with halfway completion of buffer data writing, file data is accessed when the snapshot data is accessed. May be inconsistent and inaccessible. In the present invention, since all the buffer data at the time of creating the snapshot is written on the NAS device, inconsistency between the metadata of the file and the actual data does not occur, and the file in the snapshot data is damaged. Because there is no such thing.

  The third effect is that it is not necessary to divide the area used by the file server and can be used as one volume. This makes it possible to simplify information management such as file system and file management.

  The reason for this is that, according to the present invention, it is possible to secure a quiesce point of all files accessing the file system and perform backup (snapshot), so there is no need to divide the file system so that it can be backed up. This is because management becomes easier. Further, since it is not necessary to divide the file system, management of files stored in the file system is facilitated, and the burden of security management and the like is reduced, so that information management can be simplified.

  The fourth effect is that the NAS device can lead all of the snapshot processing. As a result, it is not necessary to operate a large number of clients on the operation side, and the processing is completed in one place, so that the backup operation can be simplified.

  The reason for this is that, in addition to the creation of the snapshot, all the flush processing of the buffer data is led by the NAS device. Many of the functions linked to the client device are operated on the initiative of the client device side. However, in the present invention, it is not necessary to perform any particular operation on the client device side, and a consistent snapshot can be obtained only by operating the NAS device. I can do it. In the present invention, attention is paid to only one file system. However, the present invention can also be used in a plurality of file systems, and a plurality of files can be obtained by performing the same processing for each file system. Even in an environment where the system is present, the NAS device can take the lead and perform processing, so the backup operation can be simplified.

  Further, the advantages of the present invention will be described below in comparison with a general NAS device.

  In a general NAS device, the instruction may be executed from a client device. However, even in that case, a quiesce point of the file system on the NAS device is secured, but a quiesce point of data on the file system is not secured. For this reason, there is a possibility that data is corrupted in a general process.

  Even if a stationary point can be secured by a general NAS device, the latest data cannot be secured when creating a snapshot if there is data on the server.

  However, in the present invention, when the NAS device creates a snapshot, the NAS device instructs the client device to flush, so that a quiesce point can be taken and the latest information can be backed up.

  In addition, in general NAS devices, there are cases where it is possible to secure a quiesce point of data by using linked software, but these operations back up processes that were originally used in SAN storage. Therefore, there is almost no cooperation function between client devices, and the client (or application) and the storage (or the area where the snapshot is created are not in a one-to-one relationship as in the SAN storage. I can not use it.

  If you try to use it, as described above, client devices usually do not cooperate with each other, so when you create a snapshot, one of them has a stationary point but the other has not, or one of them is stationary. This means that you can make two snapshots with points. In that case, it is of course impossible to manage two cases where a stationary point is not obtained. Even when two can be made, the management becomes complicated. If there are 100 client devices, 100 can be obtained, which is not realistic.

  Further, regarding cooperation between client apparatuses, it is not realistic from the viewpoint of cooperation processing and operation to cooperate between 100 or more client apparatuses.

  Therefore, it cannot be used in a many-to-one configuration in which two or more client devices simultaneously access one area.

  However, when a NAS device is used for general purposes, a many-to-one configuration is fundamental, so that the use of linked software is limited.

  In the case of the present invention, on the other hand, by flushing from the NAS device that creates the snapshot, the flushing side can control the processing of the stored / updated data. Need not be considered, and only the quiesce point of data on the NAS device needs to be noted.

  As a result, it is possible to secure a quiesce point for all data with a single snapshot, and management is basically simple because it is only necessary to look at the NAS device. A many-to-one configuration can be realized.

  In the present invention, a snapshot can be created regardless of the network to which the client device belongs. Further, by using the connection information, it is possible to execute the processing only on the client device that should secure (access) the quiesce point when creating the snapshot.

  In general, even if it is intended to be realized by cooperation between client devices, especially in cooperation with a PC (Personal Computer) used by a user and another PC or server, also from the viewpoint of network technology and security. , Very difficult and unrealistic.

  However, in the present invention in which flush processing is performed on the accessed path without performing cooperation between clients, if the NAS apparatus can be accessed regardless of the network, It can be secured.

  In addition, unlike a client device that is constantly operating, such as a normal server, whether a PC or the like is activated is undefined, but in the present invention, it is based on information that is actually accessed on a NAS device. There is no need to confirm on the client device side which client device is accessing.

  For this reason, only the client devices that are really necessary can be instructed to flush, so there is no wasteful processing, and there is a client device that has forgotten to perform flash processing due to missing check that may occur when it is executed on the client device side. It does not occur, and it is possible to ensure a stationary point.

  The processing performed by each component provided in the NAS device 1 described above may be performed by a logic circuit that is produced according to the purpose. In addition, a computer program (hereinafter referred to as a program) in which processing contents are described as a procedure is recorded on a recording medium readable by the NAS device 1, and the program recorded on the recording medium is read by the NAS device 1 and executed. It may be what you do. The recording medium readable by the NAS device 1 is a removable recording medium such as a floppy (registered trademark) disk, a magneto-optical disk, a DVD, or a CD, and a memory such as a ROM or a RAM built in the NAS device 1. And HDD. The program recorded on the recording medium is read by a CPU (not shown) provided in the NAS device 1, and the same processing as described above is performed under the control of the CPU. Here, the CPU operates as a computer that executes a program read from a recording medium on which the program is recorded.

DESCRIPTION OF SYMBOLS 1 NAS apparatus 100 Client apparatus 101 NAS buffer part 102,302 Memory 103 Application part 104,301 Input / output module 105 Flash execution part 200 Network 300 NAS controller part 303 Main controller 304 Disk controller 305 Flash part 306 Flash map 307 Data protection part 308 Cold map 309 Snapshot unit 400 Storage unit 410 Snapshot area

Claims (4)

A storage device for storing data,
A flash unit for instructing the client device holding the data to flush the data to the storage device;
A temporary holding memory for temporarily holding data flushed from the client device in response to the command;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, mapping information with a quiesce point of the data held in the temporary holding memory is obtained. The data protection part to save,
A snapshot unit for creating a snapshot based on the mapping information stored by the data protection unit and the flash completion information;
A storage unit for storing a snapshot created by the snapshot unit;
The flash unit is a storage device that issues the command to an accessing client device based on the connection information. The storage device according to claim 1,
When there are a plurality of the client devices, the flash unit measures the transmission delay times between the storage device and the plurality of client devices, and executes the command in order from the client devices having the longest transmission delay times. A storage device. A storage method for storing data in a storage device,
A process of instructing a client device that is accessing the storage device among the client devices that hold the data to flush the data to the storage device;
Processing to temporarily hold data flushed from the client device in response to the command;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, the mapping information with the quiesce point of the temporarily held data is saved. Processing to
A process of creating a snapshot based on the stored mapping information and the flash completion information;
And a storage method for storing the created snapshot. In a storage device that stores data,
A procedure for instructing a client device that is accessing the storage device among the client devices that hold the data to flush the data to the storage device;
Temporarily holding data flushed from the client device in response to the instructions;
Based on the connection information between the storage device and the client device and the flush completion information indicating whether or not the flush has been completed, the mapping information with the quiesce point of the temporarily held data is saved. And the steps to
A procedure for creating a snapshot based on the stored mapping information and the flash completion information;
A program for executing the procedure for storing the created snapshot.

Images