JP2016177407A - Data management device, data management method and data management program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data management device and the like capable of shortening time for accessing a duplicated data storage area of a lower hierarchy.SOLUTION: A data management device for storing original data in a duplicated data storage area with update of the original data, comprises: update means for storing the original data as differential data in the duplicated data storage area of a first hierarchy held as a sequential generation with update of the original data, and generating a difference management table showing the data presence/absence state in the duplicated data storage area of a self-generation and upper generation in association with the duplicated data storage area; and access means for performing access on the basis of the data presence/absence state shown by the difference management table according to access to the duplicated data storage area.SELECTED DRAWING: Figure 1

Description

  The present invention relates to data access in a virtual machine using snapshot technology.

  In recent years, virtualization technology for creating virtual machines has become widespread from the viewpoint of cost reduction and ease of operation management. In the virtualization technology, for example, products using Linked Clone technology of VMWare (registered trademark) have been released. Linked Clone technology, as disclosed in, for example, Patent Document 1, takes a snapshot of a disk image (machine image) of a shared machine and manages only the difference data from the shared machine image, thereby managing the disk capacity. Is a technology to reduce

  Further, as disclosed in, for example, Patent Document 2, a snapshot technique is known as a technique for performing generation management of a copy of a master virtual machine volume (master volume). Also in this case, since the pre-update data is held in the snapshot volume only for the address where the master volume has been updated since the snapshot was created, the disk capacity can be reduced.

  Here, the snapshot is a technique for holding the state of a predetermined volume at a specified time point or a result of capturing the state. Hereinafter, the volume that stores the original data is referred to as a “master volume”, and the volume that becomes the duplicate data storage area defined to hold the pre-update data of the volume that stores the original data is referred to as the “snapshot volume”.

  An example of the snapshot technique will be described with reference to FIG. When taking a snapshot of a master volume storing data as shown in FIG. 17A, first, a snapshot volume having a storage capacity equivalent to this master volume is stored in the storage device as shown in FIG. Is generated.

  When the data “EE” is written to the storage area (address) of the data “BB” in the master volume, the data “BB” before the update is stored in the snapshot volume as shown in FIG. Stored at the same address. This snapshot volume functions as a first generation snapshot.

  Subsequently, when taking a snapshot of the master volume again, a snapshot volume having a storage capacity equivalent to the master volume is newly created in the storage device as shown in FIG. This snapshot volume is a second generation snapshot.

  Subsequently, when the data “FF” is written to the storage area of the data “CC” in the master volume, the data “CC” before the update is used as a second generation snapshot as shown in FIG. Stored at the same address of the functioning snapshot volume. At this time, the contents of the snapshot volume that functions as the first-generation snapshot are retained as shown in FIGS. 17B and 17C.

  Note that “(AA)”, “(BB)”, and the like mean that data A1 and B1 can be read from the corresponding page of the higher generation or master, although no data exists in the volume.

  In a disk array subsystem that manages data replication using the snapshot technique as described above, only data at an address that has been updated to the master volume is held in the snapshot volume. The snapshot volume has a retained data management table that manages for each address whether or not the own volume retains data. Therefore, when the data access means of the disk array subsystem receives a read command to the snapshot volume, it checks the information in the retained data management table corresponding to the address of the target data. As a result, when it is determined that there is data, the data access means reads the data from the snapshot volume. On the other hand, when it is determined that there is no data, the data access means reads the data from the master volume.

  When providing a virtual machine to a plurality of users, it is possible to perform version management of a machine image using the snapshot technology as described above. Further, it is possible to further manage the sub version by using a snapshot technique for one version. Furthermore, a plurality of sub-version snapshots can be created by the Link Clone technology and provided to the user.

  As described above, by using the snapshot technology including the Linked Clone technology hierarchically, a machine image suitable for each of a plurality of users can be provided with a small disk capacity, and operation management is facilitated.

  An example of OS (Operating System) image version management using the snapshot technique as described above and a method of providing an OS image to a user will be described.

  When the OS is updated, a snapshot of the volume of the OS image before update is created. Here, the volume of the OS image before being updated by version upgrade or patch is referred to as “common OS image volume”.

  As shown in FIG. 17B, the snapshot volume (hereinafter referred to as “each version OS volume”) having the common OS image volume as a master is connected to each version OS volume or common OS image volume of a higher generation. Only the difference is retained. The common OS image volume corresponds to the “master volume” described above, and each version OS volume created using the common OS image volume as a master corresponds to the “snapshot volume” described above.

  The user is provided with a snapshot (hereinafter referred to as “user volume”) with each version OS volume as a master. The user volume has the same image as each version OS volume that is the master at the time of creation, and then holds only the data at the address at which each version OS volume was updated.

  Here, for example, Patent Document 1 discloses a data storage system that holds only update data in a snapshot operation in a disk array subsystem.

  Patent Document 2 discloses a storage management system that can realize deduplication of disk images of existing virtual machines by simple processing.

JP 2011-248742 A JP 2005-208950 A

  In the disk array subsystem that manages user volumes by the snapshot technique as described above, access to user volumes is performed in the following procedure. As described above, the user volume does not hold data when there is no update in each master version OS volume. Therefore, in this case, the data access means of the disk array subsystem checks whether data exists in each version OS volume as a master when accessing the user volume.

  If no data exists in the master volume, the data access unit checks whether data exists in each version OS volume that is a higher generation of the snapshot.

  As described above, in the volume management based on the snapshot, it is necessary to trace the generation to the volume in which the data exists for the access to the user volume (lower-level replicated data storage area). Therefore, when there are a plurality of OS versions, there is a problem that the access time to the user volume is delayed.

  Patent Documents 1 and 2 do not disclose a technique for preventing such a delay in access time to a user volume.

  The present invention has been made in view of the above problems, and has as its main object to provide a data management device and the like that can shorten the access time to a lower-level replicated data storage area.

  The first data management device according to the present invention is held in a data management device that stores the original data in a replicated data storage area as the original data is updated, and is continuously generated as the original data is updated. The original data is stored as differential data in the replicated data storage area of the first hierarchy, and a differential management table indicating the data presence / absence status in the replicated data storage areas of its own generation and higher generation is stored in the replicated data storage area And updating means for generating the data in association with each other, and access means for performing access based on the data presence / absence status indicated in the difference management table in response to access to the replicated data storage area.

  The first data management method of the present invention is a data management method for storing original data in a replicated data storage area as the original data is updated, and is retained as successive generations as the original data is updated. The original data is stored as differential data in the replicated data storage area of the first hierarchy, and a differential management table indicating the data presence / absence status in the replicated data storage areas of its own generation and higher generation is stored in the replicated data storage area Access is made based on the data presence / absence situation shown in the difference management table in response to access to the duplicate data storage area.

  This object is also achieved by a computer program that realizes the data management method having the above-described configurations by a computer, and a computer-readable storage medium that stores the computer program.

  According to the present invention, it is possible to shorten the access time to the user volume in the data storage device.

1 is a block diagram showing a configuration of a disk array subsystem according to a first embodiment of the present invention. It is a figure which shows the example of the volume corresponding | compatible table of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the holding | maintenance data management table of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the upper generation difference management table of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the volume update permission / prohibition table of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the volume structure of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure explaining the normal read process to the user volume of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure explaining the read process to the user volume in the volume structure which increased the common OS image volume of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a flowchart which shows operation | movement of the volume management part of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure explaining the update of the upper generation difference management table of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a flowchart which shows the flow of the write processing of the disk array subsystem which concerns on the 1st Embodiment of this invention. 4 is a flowchart showing a flow of read processing of the disk array subsystem according to the first embodiment of the present invention. It is a figure explaining the read processing of the disk array subsystem which concerns on the 1st Embodiment of this invention. It is a figure explaining the read processing of the disk array subsystem which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the data management apparatus which concerns on the 3rd Embodiment of this invention. It is a figure which shows an example of the hardware constitutions which implement | achieve the apparatus which concerns on each embodiment of this invention. It is a figure explaining an example of a snapshot technique.

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

First Embodiment FIG. 1 is a block diagram showing a configuration of a disk array subsystem 100 according to a first embodiment of the present invention. As shown in FIG. 1, the disk array subsystem 100 is connected to one or a plurality of hosts 200-1 to 200-n.

  The host 200 writes to the volume, reads from the volume, creates a snapshot, updates the OS, suppresses the OS update, and permits the OS update to the disk array subsystem 100 having a volume group for performing data management. Send commands (commands). Hereinafter, the one or more hosts 200-1 to 200-n are collectively referred to as “host 200”.

  As shown in FIG. 1, the disk array subsystem 100 includes a volume management unit 110, an I / O (Input / Output) processing unit 130, a volume group 150, and a table storage unit 170.

  The volume management unit 110 executes a process for managing the volumes included in the volume group 150 in response to a volume management command from the host 200, and returns a response to the host 200 when the process is completed. The volume management unit 110 includes a volume management command acquisition unit 111, a snapshot execution unit 112, an OS update suppression unit 113, and an OS update permission unit 114.

  The I / O processing unit 130 executes read processing or write processing on the volumes included in the volume group 150 in response to an I / O command from the host 200, and returns a response to the host 200 when the processing is completed. The I / O processing unit 130 includes an I / O command acquisition unit 131, a write command execution unit 132, and a read command execution unit 133.

  The volume group 150 includes a plurality of volumes. Specifically, the volume group 150 includes a common OS image volume 151, each version OS volume 152, a user volume 153, and a pool volume 154.

  The table storage unit 170 includes a table for managing the volumes included in the volume group 150. Specifically, the table storage unit 170 includes a volume correspondence table 171, a volume update permission / refusal table 172, a retained data management table 173, and a higher generation difference management table (difference management table) 174.

  An outline of each component of the disk array subsystem 100 will be described.

  The volume management command acquisition unit 111 of the volume management unit 110 receives a volume management command from the host 200 and passes the command to the snapshot execution unit 112, the OS update suppression unit 113, or the OS update permission unit 114 according to the command.

  When the snapshot execution unit 112 receives a snapshot creation command from the volume management command acquisition unit 111, the snapshot execution unit 112 creates a snapshot from the master volume specified by the command. Then, the snapshot execution unit 112 adds information to the table included in the table storage unit 170 regarding the created snapshot.

  The OS update inhibiting unit 113 sets a flag in the volume update permission / denial table 172 according to the OS update inhibiting command from the volume management command acquiring unit 111. Further, the OS update suppression unit 113 adds information to the upper generation difference management table 174 in response to the OS update suppression command.

  The OS update permission unit 114 clears the flag of the volume update permission / denial table 172 in response to the OS update permission command from the volume management command acquisition unit 111. Further, the OS update permission unit 114 clears the flag of the higher generation difference management table 174 in response to the OS update permission command.

  FIG. 2 is a diagram illustrating an example of the volume correspondence table 171 to which information is added by the snapshot execution unit 112. The volume correspondence table 171 is a table including a correspondence relationship between a master volume in which a snapshot is created and a snapshot volume. The volume correspondence table 171 shown in FIG. 2 corresponds to the volume configuration shown in FIG. As shown in FIG. 2, the volume correspondence table 171 includes a master volume number and a snapshot volume number. In FIG. 2, for example, “MV_OS1” is the master volume number, and “SV_OSVer3”, “SV_OSVer2”, and “SV_OSVer1” are created as snapshot volumes of this master volume as the third generation, second generation, and first generation, respectively. Indicates that In this way, snapshots are held as successive generations as the master volume (original data) is updated.

  The snapshot execution unit 112 also adds information to the retained data management table 173 when a snapshot is created. FIG. 3 is a diagram illustrating an example of the retained data management table 173. As shown in FIG. 3, the retained data management table 173 is a table for managing whether or not the snapshot volume has data corresponding to the data stored in the master volume.

  The retained data management table 173 manages whether or not the corresponding data exists with a flag for each snapshot I / O processing unit (hereinafter referred to as a page unit) that is a unit for performing data replication processing. . For example, when the flag is “1”, it indicates that the corresponding data exists, and when the flag is “0”, it indicates that the corresponding data does not exist. In FIG. 3, for example, since the flag is set to “1” for “page0” of the snapshot volume “SV_OSVer1”, it indicates that data exists in “page0” of the snapshot volume “SV_OSVer1”.

  Here, when there is a write to the master volume, the snapshot execution unit 112 replicates the data before the master volume is written to the snapshot volume in units of pages (hereinafter, this operation is referred to as eviction copy).

  The snapshot execution unit 112 also adds information to the higher generation difference management table 174 when the master volume of the created snapshot volume is each version OS volume.

  FIG. 4 is a diagram illustrating an example of the upper generation difference management table 174. As shown in FIG. 4, the upper generation difference management table 174 is a table associated with each version OS volume, and data exists in either the own volume (own generation) or each version OS volume of the snapshot higher generation. Whether or not (data presence / absence status) is managed by a flag for each page. In FIG. 4, for example, “page0” of volume “SV_OSVer1” has the flag set to “1”. Show. Here, the higher generation refers to a snapshot volume created after the own volume. For example, when the first generation, second generation, and third generation snapshot volumes are created from the master volume, the first generation The higher generations refer to the second and third generation snapshot volumes.

  FIG. 5 is a diagram illustrating an example of the volume update permission / refusal table 172 managed by the OS update suppression unit 113. As shown in FIG. 5, the volume update permission / refusal table 172 includes a flag indicating whether or not write processing to a volume is prohibited. In FIG. 5, since the flag of the volume “MV_OS2” is set to “1”, it indicates that the write process to the volume “MV_OS2” is prohibited. In the operation mode in this embodiment, only the flag related to the common OS image volume is set in the volume update permission / refusal table 172, but in another operation mode, a flag related to each version OS volume or the like may be set.

  Next, the volume group 150 will be described. The common OS image volume 151 in the volume group 150 is a volume that holds a common OS image, and is a snapshot master volume of each version OS volume 152.

  Each version OS volume 152 is a snapshot volume having the common OS image volume 151 as a master, and holds only difference data with respect to the master volume for an image at a predetermined point in time in the common OS image volume 151. The actual data of each version OS volume 152 is stored in the pool volume 154, and the presence / absence of data for each address is managed by the retained data management table 173.

  The user volume 153 is a volume used by the user, and is a snapshot volume having each version OS volume 152 as a master. The user volume 153 has the same image as each master version OS volume immediately after creation, and holds only data updated by the user. The actual data of the user volume is stored in the pool volume 154, and the presence / absence of data for each address is managed by the retained data management table 173.

  The pool volume 154 holds actual data of each version OS volume and user volume that are snapshot volumes.

  FIG. 6 shows a configuration in which the version OS volumes “SV_OSVer1”, “SV_OSVer2”, and “SV_OSVer3” are created in order using the common OS image volume “MV_OS1” as a master. Also, each version OS volume “SV_OSVer1” is a master, user volumes “SV_USR1” and “SV_USR2” are each version OS volume “SV_OSVer2” is a master, user volume “SV_USR3” is each master, and each version OS volume “SV_OSVer3” is a master. The user volume “SV_USR4” indicates the created configuration. Here, the version OS volumes “SV_OSVer1”, “SV_OSVer2”, and “SV_OSVer3” are referred to as first generation version OS volumes, second generation version OS volumes, and third generation version OS volumes, respectively.

  In FIG. 6, “A1”, “B1”, and the like indicate data, and “173” and “174” are pages (addresses) for storing each data in the retained data management table 173 and the upper generation difference management table 174, respectively. The flag corresponding to is shown. “(A1)”, “(B1)”, and the like mean that data does not exist in the volume, but data A1 and B1 can be read from the corresponding page of the higher generation or master.

  For example, the data B0 will be described. When the snapshot execution unit 112 creates the snapshot “SV_OSVer2”, the address corresponding to B0 of “SV_OSVer2” indicates (B0). Thereafter, when there is an update to the common OS image volume and B1 is written to the address corresponding to B0, the snapshot execution unit 112 performs eviction copy, and the data at the address corresponding to B0 of “SV_OSVer2” is stored from (B0). Change to B0. At this time, the snapshot execution unit 112 sets “1” to the flag of the corresponding page of the retained data management table 173. Subsequently, when the snapshot execution unit 112 newly creates a snapshot “SV_OSVer3”, the address corresponding to B0 of “SV_OSVer3” becomes the value (B1) of the common OS image volume.

  In this way, each version OS volume holds the data before the update of the higher generation, and is associated with the holding data management table 173 having a flag indicating that the data exists in the own volume. Each version OS volume is associated with an upper generation difference management table 174 having a flag indicating that data exists in either the own volume or an upper generation volume.

  The user volume holds data updated by the user and is associated with a holding data management table 173 having a flag indicating that data exists in the own volume.

  Here, with reference to FIG. 7, a normal read process for the user volume will be described. A unit that executes normal read processing is referred to as a normal read processing unit (not shown). In normal read processing, it is assumed that each version OS volume does not have the upper generation difference management table 174, but has only the retained data management table 173.

  In the configuration shown in FIG. 7, a snapshot of each version OS volume is created using the common OS image volume as a master, and each version OS volume is created from the first generation to the third generation. Further, a user volume 2 is created with each first generation version OS volume as a master, and a user volume 1 is created with each second generation version OS volume as a master.

  Assume that a read command (read a) is made to D0 of user volume 2. In this case, the normal read processing unit checks the retained data management table 173 of the user volume 2. In FIG. 7, since the flag of the retained data management table 173 of the page corresponding to D0 is “0”, no data exists in the user volume 2.

  Therefore, the normal read processing unit checks the retained data management table 173 of each first generation version OS volume that is the master volume of the snapshot pair (indicated by “b” in FIG. 7). Again, since the flag of the page corresponding to D0 is “0”, there is no data. Similarly, the normal read processing unit sequentially checks the retained data management table 173 of the second generation and third generation snapshots (shown as “c” and “d” in FIG. 7), both of which are higher generations. Since the flag is “0”, there is no data.

  As a result, the normal read processing unit reads data D0 from the common OS image volume (indicated by “e” in FIG. 7).

  Here, in OS version upgrades and patches, there are many updates to only a limited area of the volume, so there is no data in the user volume, and the upper generation is traced and the common OS volume is finally referenced as described above. There are many cases. That is, in accessing many user volumes, it is necessary to search a plurality of retained data management tables 173, so that the user volume access time is delayed. Since the retained data management table 173 to be searched increases as the OS version increases, the access performance to the user volume further deteriorates.

  Thus, for example, as shown in FIG. 8, it is conceivable to manage each version OS volume by increasing the common OS image volume. FIG. 8 shows a configuration in which each version OS volume of 1a generation and 2a generation is created using the common OS image volume 1 as a master, and each version OS volume of 1b generation is created using the common OS image volume 2 as a master. Show. Furthermore, a user volume 1 is created using each version OS volume of the 1a generation as a master, and a user volume 2 is created using each version OS volume of the 1b generation as a master.

  Similar to the configuration of FIG. 7, when a read command (read f) is issued to D0 of the user volume 2, the normal read processing unit checks the retained data management table 173 of the user volume 2. In FIG. 8, since there is no data corresponding to D0, the normal read processing unit checks the retained data management table 173 of each version OS volume of the 1b generation that is the master volume (indicated by “g” in FIG. 8). ). Since the flag of the page corresponding to D0 is “0”, the normal read processing unit reads the data D0 from the common OS image volume 2 (indicated by “h” in FIG. 8).

  In the configuration shown in FIG. 8, the number of each version OS volume related to the user volume 2 is one, and the holding data management table 173 to be searched is one, so that the access time to the user volume is not delayed. By increasing the common OS image volume in this way, it is conceivable to prevent a delay in access time to the user volume. However, since the common OS image volume is a volume that holds actual data, it consumes disk capacity. There is a problem of end up.

  Therefore, the disk array subsystem 100 according to the present embodiment reduces the access time to the user volume without increasing the disk capacity consumption as follows.

  First, an operation method of the disk array subsystem 100 will be described. When the OS is updated due to version upgrade or patch, the disk array subsystem 100 holds the OS version before update as each version OS volume 152 by creating a snapshot of the common OS volume image before update. To do. The OS is updated by a write process to the common OS volume image after the snapshot is created.

  When each version is provided to the user, each version OS volume is used as a master, and the snapshot execution unit 112 further creates a snapshot for each user. A user volume that is an OS image of each version is provided for each user. Each user accesses the user volume.

  During normal operation in which there is no update to the common OS image volume, the disk array subsystem 100 suppresses the update to the common OS image volume in response to the OS update suppression command transmitted from the host 200. As a result, the upper generation difference management table becomes valid, so that the read processing to the user volume using the upper generation difference management table is carried out (details will be described later).

  When the common OS image volume is updated by version upgrade or patch application, the disk array subsystem 100 permits the update of the common OS image volume in accordance with the OS update permission command transmitted from the host 200 in advance. To do. At this time, since the upper generation difference management table 174 is cleared and all the flags are “0”, the retained data management table 173 of each version OS volume is searched one by one (details will be described later). When the update of the common OS image volume is completed, an OS update suppression command is sent from the host 200, and the disk array subsystem 100 suppresses the update to the common OS image volume and updates the upper generation difference management table 174 ( Details will be described later).

  Next, details of each component of the disk array subsystem 100 will be described.

  FIG. 9 is a flowchart showing the operation of the volume management unit 110 when a volume management command is sent from the host 200 to the disk array subsystem 100. The operation of the volume management unit 110 will be described with reference to FIG.

  When a volume management command is sent from the host 200, the volume management command acquisition unit 111 receives the command (S101). The volume management command acquisition unit 111 determines the type of the received command, and when the command is a snapshot creation command (Yes in S102), passes the command to the snapshot execution unit 112. The snapshot execution unit 112 creates a snapshot volume from the master volume specified by the received command (S103). Then, the snapshot execution unit 112 adds the correspondence between the created snapshot volume and the master volume to the volume correspondence table 171 (S104).

  Subsequently, the snapshot execution unit 112 adds information about the created snapshot volume to the retained data management table 173 (at this time, the flags corresponding to all pages are “0”) (S105).

  Subsequently, when the master of the created snapshot volume is a common OS image volume (Yes in S106), the snapshot execution unit 112 adds information on the created snapshot volume to the higher generation difference management table 174. . For example, when the snapshot execution unit 112 creates each version OS volume “SV_OSVer3” having the common OS image volume “MV_OS1” shown in FIG. 6 as a master, the snapshot execution unit 112 stores “ Information of “SV_OSVer3” is added (at this time, the flags corresponding to all the pages are “0”) (S107).

  On the other hand, as a result of determining the command type in S102, if the command is not a snapshot creation command (No in S102), the command is an OS update inhibition command or an OS update permission command. The OS update inhibition command is sent from the host 200 to the disk array subsystem 100 in response to a response when the update of the common OS image volume shown in S205 of FIG. The OS update suppression command indicates that the upper generation difference management table 174 is valid until the OS has been updated and is updated next time. The OS update permission command is sent from the host 200 to the disk array subsystem 100 before the common OS image volume is updated (write process). Since the common OS image volume is updated according to the OS update permission command, the upper generation difference management table 174 is invalid, that is, the flag is cleared.

  When the command received from the host 200 is an OS update suppression command (Yes in S108), the volume management command acquisition unit 111 passes the command to the OS update suppression unit 113. When receiving the command, the OS update inhibiting unit 113 sets “1” to the flag of the volume update permission / refusal table 172 of the volume designated by the command in order to inhibit the update to the volume (S109).

  Subsequently, when the volume for which update suppression is specified by the OS update suppression command is a common OS image volume, the OS update suppression unit 113 is the upper generation of all the version OS volumes that are snapshots having that volume as a master. The flag of the difference management table 174 is updated (S110). At this time, the OS update inhibiting unit 113 takes a logical sum (OR) of the retained version management volume 173 possessed by each target version OS volume and each version OS volume of all higher generations. When the logical sum is “1”, the OS update inhibiting unit 113 updates the flag of the corresponding page of the higher generation difference management table 174 to “1”.

  An example of updating the upper generation difference management table 174 when the data of the common OS image volume is updated will be described with reference to FIG. As shown in FIG. 10A, a snapshot volume “SV_OSVer2” having “MV_OS1” as a master is created, and thereafter, data B0 of “MV_OS1” is updated to B1 as shown in FIG. 10B. Assuming that At this time, “1” is set to the flag of the page corresponding to the data B1 in the retained data management table 173 of “SV_OSVer2” by the normal snapshot technique (implemented in S205 of FIG. 11 described later).

  At this time, the OS update suppression unit 113 sets the upper level of the corresponding page for all version OS volumes (in this case, “SV_OSVer2” and “SV_OSVer1”) that are snapshots having the updated common OS image volume as a master. The generation difference management table 174 is updated. In other words, the OS update inhibiting unit 113 takes the logical sum (OR) of the retained data management table 173 of the corresponding page for each volume and its higher generation volume.

  As shown in FIG. 10B, for “SV_OSVer2”, since “1” is set in the flag of the corresponding page of the retained data management table 173 as described above, the OS update inhibiting unit 113 sets this page. “1” is set to the flag of the corresponding upper generation difference management table 174 (indicated by “n” in FIG. 10B). As for “SV_OSVer1”, the logical sum of the corresponding pages of the retained data management table 173 of the own volume “SV_OSVer1” and the higher generation “SV_OSVer2” is “1”. “1” is set to the flag of the corresponding page of the upper generation difference management table 174 (indicated by “o” in FIG. 10B). With the above procedure, the OS update inhibiting unit 113 updates the upper generation difference management table 174.

  On the other hand, as a result of determining the command type in S102, if the command is an OS update permission command (No in S108), the volume management command acquisition unit 111 passes the command to the OS update permission unit 114. The OS update permission command is a command for permitting updates to the volume. Upon receiving this command, the OS update permission unit 114 resets the flag of the volume update permission / refusal table 172 for the volume specified by the command (S111). ).

  Subsequently, the OS update permission unit 114 clears the upper generation difference management table 174 of all the version OS volumes that are snapshots whose master is the common OS image volume for which update permission is specified by the OS update permission command ( S112).

  When the above processing ends, the volume management unit 110 returns a response to the host 200 (S113).

  Next, an outline of the operation of the I / O processing unit 130 will be described.

  The I / O processing unit 130 receives an I / O command from the host 200 to the common OS image volume or each version OS volume or user volume, and returns a response when the I / O processing is completed.

  The I / O command acquisition unit 131 receives a read / write command from the host 200, passes the write command to the write command execution unit 132, and passes the read command to the read command execution unit 133.

  When receiving the write command, the write command execution unit 132 checks the flag in the volume update permission / refusal table 172 of the volume to be written, and rejects the write command if the flag is “1”. When the flag is “0”, write processing is performed according to the type of volume to be written. Specifically, when the write command is a write command to the common OS image volume, the write command execution unit 132 performs a write process to the master volume of the normal snapshot technology on the common OS image volume.

  When the write command is a write command to each version OS volume, the write command execution unit 132 rejects the write command. This is a process according to an operation in which writing to each version OS volume is not performed because the upper generation difference management table 174 becomes invalid when the writing process to each version OS volume is performed. In other operation modes, write processing to each version OS volume may be performed. As described above, the OS update is performed on the common OS image volume.

  When the write command is a write command to the user volume, the write command execution unit 132 performs a write process to the snapshot volume using the normal snapshot technique.

  On the other hand, when the read command execution unit 133 receives a read command to the common OS image volume, each version OS volume, or the user volume from the I / O command acquisition unit 131, the read command execution unit 133 performs read processing according to the type of volume to be read. To implement.

  Specifically, when the read instruction is a read instruction to the common OS image volume, the read instruction execution unit 133 performs a read process to the master volume of the normal snapshot technology on the common OS image volume. When the read command is a read command to each version OS volume, the read command execution unit 133 performs a read process to the snapshot volume of the normal snapshot technology on each version OS volume.

  When the read command is a read command to the user volume, the read command execution unit 133 checks the retained data management table 173 associated with the user volume, and reads data from the user volume when the flag is “1”. When the flag is “0”, the read instruction execution unit 133 checks the upper generation difference management table 174 of each version OS volume that is the master volume.

  When the flag of the upper generation difference management table 174 is “0”, the read command execution unit 133 reads the corresponding address of the common OS image volume that is the master of each version OS volume. When the flag of the upper generation difference management table 174 is “1”, the read instruction execution unit 133 sets each version OS volume that becomes the master of the user volume, each version OS volume of the upper generation,. The retained data management table 173 is examined in the order of the version OS volume. Then, the read instruction execution unit 133 reads data from the volume in which the flag “1” is first found. If no volume with the flag “1” is found, the read command execution unit 133 reads data from the corresponding address of the common OS volume.

  The I / O process will be described in detail with reference to FIG. 11 and FIG.

  FIG. 11 is a flowchart showing the flow of write processing in the I / O processing by the I / O processing unit 130. FIG. 12 is a flowchart showing the flow of read processing in the I / O processing by the I / O processing unit 130.

  When a read / write command is sent from the host 200, the I / O command acquisition unit 131 receives it (S201). When the read / write command is a write command (Yes in S202), the write command is used as a write command. It is passed to the execution unit 132.

  The write command execution unit 132 determines the type of the write command, and when the write command is a write command to the common OS image volume (Yes in S203), the common OS image volume to be written in the volume update permission / refusal table 172. Check the flags. When the flag is “1” (Yes in S204), the write command execution unit 132 rejects the write command and returns the reject to the host 200 (S207).

  When the flag of the volume update permission / refusal table 172 is “0” (No in S204), the write command execution unit 132 performs the write process to the master volume in the normal snapshot technology on the common OS image volume (S205). ). At this time, as described above, the write command execution unit 132 sets “1” to the flag of the corresponding page in the retained data management table 173 included in each version OS volume having the common OS image volume as a master. Then, the write command execution unit 132 returns a response to the host 200 (S206).

  On the other hand, in S203, when the write command is not for the common OS image volume (No in S203) but for each version OS volume (Yes in S208), the write command execution unit 132 rejects the write command, and the host 200 The reject is returned to (S207).

  If the write command is not for each version OS volume but for the user volume in S208, the write command execution unit 132 performs a write process to the snapshot volume in the normal snapshot technology for the user volume (S209). At this time, the write command execution unit 132 sets “1” to the flag of the corresponding page of the retained data management table 173 of the user volume that has performed the write process. Then, the write command execution unit 132 returns a response to the host 200 (S210).

  Next, an operation when the I / O command acquisition unit 131 receives a read command in S202 will be described with reference to FIG.

  When receiving the read command, the I / O command acquisition unit 131 passes the read command to the read command execution unit 133. The read command execution unit 133 determines the type of the read command, and when the read command is a read command to the common OS image volume (Yes in S301), the read processing to the master volume in the normal snapshot technology is shared. The process is performed on the OS image volume (S302).

  When the read command is a read command to each version OS volume (Yes in S303), the read command execution unit 133 performs the read processing to the snapshot volume in the normal snapshot technology on each version OS volume ( S304).

  If the read command is a read command to the user volume (No in S303), the read command execution unit 133 checks whether there is data in the user volume. Here, with reference to FIG. 13, an operation in response to a read command to the user volume will be described.

  Assume that a read command (read i) is made to the (D0) data of the user volume 2. At this time, the read command execution unit 133 reads the flag of the corresponding address in the retained data management table 173 of (D0) of the user volume 2. If the flag is “1”, since there is data in the user volume 2 (Yes in S305), the read command execution unit 133 reads the corresponding address of the user volume 2 (S306).

  On the other hand, when the flag is “0”, the read instruction execution unit 133 checks the corresponding address in the upper generation difference management table 174 held in the first generation version OS volume that is the master of the user volume 2 (FIG. 13). Indicated by “j” in FIG. When the flag of the corresponding address is “0” (Yes in S307), it can be seen that there is no data in the first generation and higher generation version OS volumes. Therefore, the read instruction execution unit 133 reads data from the common OS image volume as a master (indicated by “k” in FIG. 13) (S308).

  When the flag of the corresponding address is “1” (No in S307), the read command execution unit 133 checks the corresponding address in the retained data management table 173 held in each version OS volume that is the master of the user volume 2 ( S309 to S311). When the flag is “1” (Yes in S311), the read instruction execution unit 133 reads data from each version OS volume (S312).

  When the flag is “0” (No in S311), the read instruction execution unit 133 checks the corresponding address in the retained data management table 173 of each version OS volume of the one higher generation, and when the flag is “1” Read data from each version OS volume. When the flag is “0”, the read instruction execution unit 133 checks the corresponding address in the retained data management table 173 of each version OS volume of the higher generation. In this way, the read instruction execution unit 133 repeats the investigation of the retained data management table 173 of the higher generation until a volume that holds data is found, and reads data from each version OS volume where the data exists.

  When there is no data in each version OS volume of all higher generations, the read command execution unit 133 reads data from the common OS image volume (S308). When the read process is completed by the above procedure, the I / O processing unit 130 returns the read data to the host 200 (S313).

  As described above, according to this embodiment, the write command execution unit 132 of the disk array subsystem 100 executes a write process to the common OS image volume. When the write process is completed, the write command execution unit 132 generates an upper generation difference management table 174 indicating the presence / absence of data in the own volume or higher generation volume for each version OS volume whose master is the updated volume.

  When there is no data in the user volume in response to a read instruction to the user volume, the read instruction execution unit 133 checks the upper generation difference management table 174 of each version OS volume in the upper hierarchy. If it is found that there is no data in each version OS volume of this hierarchy based on the upper generation difference management table 174, the read command execution unit 133 reads data from the common OS image volume.

  By adopting the above configuration, according to the first embodiment, when there is no data in the user volume, the read command execution unit 133 determines whether or not there is data in the volume of the master tier, by using the higher generation difference management table. It can be determined by searching 174 once. Therefore, the access time to the user volume can be shortened without causing an increase in disk consumption accompanying the increase in the common OS image volume.

Second Embodiment Next, a second embodiment based on the above-described first embodiment will be described. In the first embodiment, as shown in FIG. 6, the disk array subsystem in which the common OS image volume, each version OS volume, and the user volume three-tier volume is arranged is shown, but the present invention is not limited to this. In the second embodiment, examples of different volume configurations will be described.

  FIG. 14 is a diagram showing a configuration of a volume included in the disk array subsystem 100 according to the second embodiment. FIG. 14 shows that the first generation, second generation, and third generation snapshot volumes of the tier 1 (first tier) are created from the master volume. Furthermore, first and second generation snapshot volumes of Tier 2 (second tier) are created with the third generation snapshot volume of Tier 1 as the master. Further, a tier 3 snapshot volume is created with the tier 2 second generation snapshot volume as the master.

  In such a volume configuration having a plurality of hierarchies, the disk array subsystem 100 arranges the upper generation difference management table 174 described in the first embodiment in a snapshot volume having a lower hierarchy snapshot. In the configuration shown in FIG. 14, the snapshot volumes having snapshots in the lower hierarchy are the snapshot volumes in hierarchy 1 and hierarchy 2.

  The operation when a read command is issued to the volume of the lowest hierarchy in the volume configuration as shown in FIG. The read process corresponding to the read command is executed by the read command execution unit 133 described in the first embodiment.

  Assume that a read command (read m) is made to the data C0 of the snapshot volume 3 in the hierarchy 3 shown in FIG. In this case, as described in the first embodiment, the read instruction execution unit 133 refers to the upper generation difference management table 174 of the master volume of the snapshot volume of the tier 3. Since the flag of the corresponding address in the upper generation difference management table 174 is “0”, the read instruction execution unit 133 knows that the data C0 exists in the master volume of the hierarchy 1.

  Here, when it is assumed that the volumes of the tier 1 and the tier 2 do not have the upper generation difference management table 174, the read instruction execution unit 133 operates as follows. That is, when it is determined that no data exists in the snapshot volume of the tier 3, the read instruction execution unit 133 checks the retained data management table 173 of the first generation snapshot volume of the tier 2. Subsequently, until the data exists, the read instruction execution unit 133 stores each snapshot volume of the second generation of the hierarchy 2 → the first generation of the hierarchy 1 → the second generation of the hierarchy 1 → the third generation of the hierarchy 1. All the held data management tables 173 are examined.

  As described above, when the snapshot volume does not have the upper generation difference management table 174, the access time is delayed because it is necessary to search the plurality of retained data management tables 173 when accessing the lower-level snapshot volume. .

  In contrast, in this embodiment, the read instruction execution unit 133 checks the upper generation difference management table 174 of the tier 2 first generation snapshot volume, and if the flag of the corresponding page is “0”, the tier 2 It can be seen that there is no data in the volume. Therefore, the read instruction execution unit 133 checks the upper generation difference management table 174 of the upper tier, that is, the first generation snapshot volume of the tier 1. Also here, since the flag of the corresponding page of the upper generation difference management table 174 is “0”, the read instruction execution unit 133 knows that the data exists in the upper hierarchy, that is, the master volume.

  As described above, according to the second embodiment, the volume configuration is not limited to the three-layer configuration as shown in the first embodiment, but has a snapshot volume generated over a plurality of layers. As in the first embodiment, the access time to the lowest layer can be shortened.

Third Embodiment FIG. 15 is a block diagram showing a configuration of a data management apparatus 400 according to a third embodiment of the present invention. As shown in FIG. 15, the data management device 400 includes an update unit 410 and an access unit 420.

  As the original data is updated, the data management device 400 stores the original data in the duplicate data storage area. The update unit 410 stores the original data as differential data in the first-tier replicated data storage area that is retained as successive generations as the original data is updated, and also includes the self-generation and higher-generation replicated data storage areas. A difference management table indicating the data presence / absence status is generated in association with the duplicate data storage area.

  The access unit 420 performs access based on the data presence / absence status indicated in the difference management table in response to access to the replicated data storage area.

  The data management device 400 corresponds to the disk array subsystem in the first embodiment, the update unit 410 also corresponds to the write command execution unit 132, and the access unit 420 similarly to the read command execution unit 133. Equivalent to.

  By adopting the above configuration, according to the third embodiment, the data presence / absence status of the replicated data storage area of the predetermined hierarchy can be searched based on the difference management table. The access time can be shortened.

  Each part of the disk array subsystem 100 shown in FIG. 1 is realized by the hardware resources illustrated in FIG. That is, the configuration shown in FIG. 16 includes a processor 10, a RAM (Random Access Memory) 11, a ROM (Read Only Memory) 12, an I / O (Input / Output) device 13, a storage 14, and a bus 15 for connecting each component. Is provided.

  In each of the above-described embodiments, as an example executed by the processor 10 shown in FIG. 16, after supplying a computer program capable of realizing the above-described functions to the disk array subsystem 100, the computer program is stored in the processor. A case has been described in which 10 is realized by reading out to the RAM 11 and executing it. However, some or all of the functions shown in each block shown in FIG. 1 may be realized as hardware.

  The supplied computer program may be stored in a computer-readable storage device such as a readable / writable memory (temporary storage medium) or a hard disk device. In such a case, the present invention can be understood as being configured by a code representing the computer program or a storage medium storing the computer program.

10 processor 11 RAM
12 ROM
13 I / O device 14 Storage 15 Bus 100 Disk array subsystem 110 Volume management unit 111 Volume management command acquisition unit 112 Snapshot execution unit 113 OS update inhibition unit 114 OS update permission unit 130 I / O processing unit 131 I / O command Acquisition unit 132 Write command execution unit 133 Read command execution unit 150 Volume group 151 Common OS image volume 152 Version OS volume 153 User volume 154 Pool volume 170 Table storage unit 171 Volume correspondence table 172 Volume update permission table 173 Holding data management table 174 Generation difference management table 200 host

Claims (10)

Along with the update of the original data, in the data management device that stores the original data in the duplicate data storage area,
The original data is stored as difference data in the replicated data storage area of the first tier that is retained as successive generations with the update of the original data, and in the replicated data storage areas of its own generation and higher generations Update means for generating a difference management table indicating the presence / absence of data in association with the replicated data storage area;
A data management apparatus comprising: access means for performing access based on the data presence / absence status indicated in the difference management table in response to access to the duplicate data storage area. The update unit is configured to store the difference data stored in the duplicate data storage area associated with the difference management table in the duplicate data storage area in the second and subsequent hierarchies held as successive generations with the update of the difference data. Storing the difference data;
The access means has no data to be accessed in the duplicate data storage area of the hierarchy based on the data presence / absence situation indicated in the difference management table associated with the duplicate data storage area of a predetermined hierarchy The data management device according to claim 1, wherein when the determination is made, the difference management table associated with the replicated data storage area in a hierarchy higher than the hierarchy is searched. The access means has the data to be accessed in the duplicate data storage area of the hierarchy based on the data presence / absence situation indicated in the difference management table associated with the duplicate data storage area of a predetermined hierarchy The data management device according to claim 2, wherein, when it is determined, the data held in any one of the duplicate data storage areas of the hierarchy is accessed. The access means accesses the original data when it is determined that there is no data to be accessed in the replicated data storage area of all layers based on the difference management table. 3. The data management device according to 3. The update means sets a flag of the difference management table corresponding to an address where the differential data exists when the differential data exists in at least one of the replicated data storage areas of its own generation and upper generation. 5. The data management device according to any one of 4. In the data management method for storing the original data in the replicated data storage area along with the update of the original data,
The original data is stored as difference data in the replicated data storage area of the first tier that is retained as successive generations with the update of the original data, and in the replicated data storage areas of its own generation and higher generations A difference management table indicating data presence / absence status is generated in association with the duplicate data storage area,
A data management method for performing access based on a data presence / absence situation indicated in the difference management table in response to access to the duplicate data storage area. When storing the original data as differential data, the second and subsequent hierarchies held as successive generations as the differential data stored in the duplicate data storage area associated with the differential management table is updated. Storing the difference data in a replicated data storage area;
When performing the access, based on the data presence / absence situation indicated in the difference management table associated with the replicated data storage area of a predetermined hierarchy, the access target data is stored in the replicated data storage area of the hierarchy. The data management method according to claim 6, wherein if it is determined that there is no data, the difference management table associated with the replicated data storage area in a higher hierarchy than the hierarchy is searched. When performing the access, based on the data presence / absence situation indicated in the difference management table associated with the replicated data storage area of a predetermined hierarchy, the access target data is stored in the replicated data storage area of the hierarchy. The data management method according to claim 7, wherein when it is determined that the data is present, the data held in any one of the duplicate data storage areas of the hierarchy is accessed. 8. When performing the access, if it is determined that there is no data to be accessed in the replicated data storage area of all layers based on the difference management table, the original data is accessed. Item 9. The data management method according to Item 8. Along with the update of the original data, the data management device that stores the original data in the replicated data storage area,
The original data is stored as difference data in the replicated data storage area of the first tier that is retained as successive generations with the update of the original data, and in the replicated data storage areas of its own generation and higher generations A process for generating a difference management table indicating data presence / absence conditions in association with the duplicate data storage area;
A data management program that executes a process of performing an access based on a data presence / absence situation indicated in the difference management table in response to an access to the duplicate data storage area.

Images