JP4159506B2 - Hierarchical storage device, recovery method thereof, and recovery program - Google Patents

Description

  The present invention relates to a hierarchical storage device, a recovery method thereof, and a recovery program, and more particularly to a device for recovering a file system of a high-speed access device from backup data of a low-speed access device.

  Hierarchical storage management (HSM) by computer systems is also called HSM (Hierarchical Storage Management), and primary storage devices (for example, memory, magnetic disks, etc.) that are high-speed access devices with high read / write speeds but low read / write speeds. Combining secondary storage devices (for example, magnetic tapes, magneto-optical disks, optical disks (DVD-R, etc.)) that are low-speed access devices at low cost, and constructing large-capacity storage devices (hierarchical storage devices) at low cost The files stored in the storage device are integrated and managed.

  In this hierarchical storage management, when a file requested by the user exists in the secondary storage device, the file is automatically copied from the secondary storage device to the primary storage device, and the file which is no longer used is stored in the primary storage device. A process of automatically saving from the device to the secondary storage device is performed. As a result, the user can use the file without being aware of whether the actual file exists in the primary storage device or the secondary storage device.

  In a computer system that performs such hierarchical storage management, for example, when a failure occurs in a magnetic disk, after removing the cause of the failure, files on the magnetic disk are backed up from backup data that was previously backed up to magnetic tape. Is restored. As a computer system for performing backup and restoration of such a file, an operating system using UNIX (registered trademark, hereinafter omitted) is generally known.

  In a general UNIX operating system, a file system which is a logical device having a structure of a super block, an inode (index node: i-node) list, and a data block is constructed on a magnetic disk device which is a physical device (for example, Non-Patent Document 1). In this case, the super block holds information on the state of the file system such as the size of the file system, the maximum number of files, the location of the free space, etc., and the inode list includes the file type and the file owner. This is an inode sequence that holds meta information such as information, group information, file creation time, and file update time, and a data block holds data (substance) of the file itself. In this file system, an unused inode in the inode list is assigned when a file is created, and the created file can be uniquely identified on the magnetic disk by the inode number. In this case, the meta information of the file is stored in an inode having an arbitrary inode number assigned in the inode list, and the substance of the file is stored in one or more data blocks.

  A computer system on which the UNIX operating system operates is mounted with the above-mentioned hierarchical storage management program (hereinafter referred to as a hierarchical storage management program) and its database. This hierarchical storage management program manages on the database information such as movement of each file on the magnetic disk and magnetic tape constituting the hierarchical storage device using the inode number as a key.

  As a result, the computer system implements backup means and restore means for the file system on the magnetic disk. When restoring and creating a file on the magnetic disk from the backup data on the magnetic tape by this restoration means, the inode number is newly assigned from the vacant number in the inode list. It is different from the inode number. Therefore, in the hierarchical storage management program that operates the database using this inode number as a key, it is possible to operate even with a new inode number after the file system is restored, so that information on the file on the database every time one file is restored Is being processed.

As prior art documents related to the present invention, there are the following.
JP 2001-118365 A Japanese Patent No. 3270216 Maurice J. Bach, Saka text, Yoshikatsu Tada, Jun Murai, “bit separate volume UNIX kernel design”, 1990, Kyoritsu Publishing Co., Ltd., p. 18-20

  However, in the above-described conventional technique, when the file system is restored from the backup data, the inode number that uniquely identifies the file on the file system changes, so the database of the hierarchical storage management program for all the files to be restored And the inode number stored in the database had to be updated. For this reason, when the number of files to be restored increases, it takes time to search and update the database. As a result, there is a problem that the time required for restoration increases exponentially, and a solution has been desired.

  The present invention has been made in consideration of such conventional circumstances, and an object of the present invention is to reduce the time required to restore a hierarchical storage device and to restore the hierarchical storage device at high speed.

In order to achieve the above object, a hierarchical storage device according to the present invention has an i-node including attribute information of a file and uniquely identifies the file by its i-node number in the hierarchical storage device operating on the operating system. A first storage device in which a file system is constructed, a second storage device for storing data including backup data of the file system, and the first storage device from backup data on the second storage device When the file system is restored, an i-node number designating unit for designating an i-node number of a file to be restored using an i-node number included in the backup data, and the designated i-node number and a file allocation means for allocating the recovery target file in the file system, the file allocation hands Was designated the i-node number is determined whether unused on the file system, characterized in that it is a means for assigning the i-node number to the recovery target file if unused.

In the hierarchical storage device according to the present invention, before SL i-node number designation means is means for starting a system call by specifying an arbitrary i-node number on the operating system, the file allocation unit, the system calls Means for assigning the i-node number specified in step 1 to a recovery target file of the file system.

The hierarchical storage device recovery method according to the present invention is a hierarchical storage device that operates on an operating system and has an i-node that includes file attribute information and uniquely identifies the file by its i-node number. In a hierarchical storage device recovery method comprising a first storage device in which a system is constructed and a second storage device that stores data including backup data of the file system, backup data on the second storage device Designating the i-node number of the file to be restored using the i-node number included in the backup data when the file system is restored to the first storage device from and a step of assigning a number to the recovery target file of the file system, the file Ri shed steps designated i-node number is determined whether unused on the file system, characterized in that the i-node number if not used a step of assigning to the recovery target file .

In the recovery method of hierarchical storage device according to the present invention, the step of specifying the previous SL i-node number is the step of activating a system call by specifying an arbitrary i-node number on the operating system assigns the file The step may be a step of assigning the i-node number designated by the system call to a recovery target file of the file system.

A recovery program for a hierarchical storage device according to the present invention is a hierarchical storage device that operates on an operating system and has an i-node including attribute information of a file and uniquely identifies the file by its i-node number In a recovery program for a hierarchical storage device having a first storage device in which a system is constructed and a second storage device that stores data including backup data of the file system, the computer stores the data on the second storage device. Designating the i-node number of the file to be restored using the i-node number included in the backup data when the file system is restored on the first storage device from the backup data of Assign the i-node number to the recovery target file of the file system. A program for executing the flop, the step of assigning the file, specified i-node number is determined whether unused on the file system, the i-node number if not used It is a step of assigning to the recovery target file.

In recovery program for hierarchical storage device according to the present invention, the step of specifying the previous SL i-node number is the step of activating a system call by specifying an arbitrary i-node number on the operating system assigns the file The step may be a step of assigning the i-node number designated by the system call to a recovery target file of the file system.

  According to the present invention, when restoring a file to the file system, a file can be created by specifying an arbitrary inode number on the file system, thereby eliminating the need to update the database of the hierarchical storage management program, As a result, the time required for the recovery of the hierarchical storage device becomes proportional only to the number of files, and the recovery of the hierarchical storage device can be executed faster than before.

  Next, the best mode for carrying out the hierarchical storage device, its recovery method, and recovery program according to the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a functional block diagram showing the overall configuration of the hierarchical storage apparatus of this embodiment, and FIG. 2 is a functional block diagram showing the configuration of the main part thereof.

  The hierarchical storage device shown in FIG. 1 is composed of, for example, a computer system capable of running a UNIX operating system, and functionally includes an application program 11, a hierarchical storage management program (HSM) 12, and a UNIX operating system kernel (UNIX kernel) 13. Including a software unit 1, a magnetic disk device 21 as a primary storage device (first storage device) and a high-speed access device unit 2 having a magnetic disk database (DB1) 22, and a secondary storage device (second storage device) ) As a low-speed access device unit 3 having a magnetic tape device 32 and a magnetic tape database (DB 2) 31, and a processing unit 4 having a CPU 41 and a memory 42.

  The software unit 1 is stored on a recording medium such as a ROM or a magnetic disk, and is read and executed on the memory 42 by the CPU 41 of the processing unit 4 at the time of execution.

  The application program 11 activates a preset system call, reads a function in the kernel 13, and executes various exchanges with the kernel 13. As shown in FIG. 2, the application program 11 includes restore means (restore program) 111 that restores the file system on the magnetic disk device 21. In this embodiment, a system call 5 for requesting a specific inode allocation (see later) that can specify an inode number is newly set as a system call by the restoration unit 111. The designation of the inode number of the file by the specific inode allocation request system call 5 is performed using the inode number included in the backup data 6 (see later) on the magnetic tape device 32, as will be described later.

  The hierarchical storage management program 12 operates on the kernel 13, manages the magnetic disk database 22 and the magnetic tape database 31 according to a predetermined algorithm, and manages the hierarchical storage management of files on the magnetic disk 21 and the magnetic tape 32. Execute a predetermined operation related to

  The kernel 13 is composed of a program that plays a central role in the UNIX operating system, and calls a function defined in advance according to a system call from the application program 11 and executes predetermined processing by the function. As shown in FIG. 2, the kernel 13 has a file management layer 130 and a device driver layer 131 in terms of function. The file management layer 130 includes an inode list reading unit 1301, an inode allocation unit 1302, and an inode list writing unit 1303, and the device driver layer 131 includes a device access unit 1311. By the operation of these means, in addition to normal known processing, as will be described later, processing according to the system call 5 for requesting specific inode allocation from the restoring means 111 of the application program 11 can be executed.

  The inode list reading unit 1301 performs a process of reading a disk block in which the inode specified by the inode number exists from the inode list 212 on the file system 210 (corresponding to the processes of steps A1 to A5 in FIG. 6 described later). .

  The inode allocation unit 1302 checks whether or not the specified inode number is unused from the inode of the disk block read by the inode list reading unit 1301. If the inode is not used, an error is returned. If not, the inode number is assigned to the file. An allocation process is performed (corresponding to the processes in steps A6 to A10 in FIG. 6 described later).

  The inode list writing unit 1301 performs a process of writing the inode used by the inode allocation unit 1302 on the magnetic disk device 21 (corresponding to steps A11 and A12 in FIG. 6 described later). Thus, when the same inode number is specified by the next specific inode allocation request system call 5, an error can be returned because it is in use.

  As described above, the file system 210 based on a general UNIX operating system is constructed in the magnetic disk device 21. The file system 210 has a super block 211, an inode list 212, and a data block 213 (see, for example, Non-Patent Document 1). In the file system 211, when creating a file, an unused inode is allocated, and meta information such as the owner of the file and creation time is stored in this inode. Also, one or more data blocks 213 having a capacity capable of storing the file entity are allocated.

  Here, in a general hierarchical storage device, when a file is created on the file system 210 constructed in the magnetic disk device 21 which is a primary storage device, and the free capacity of the file system 210 becomes insufficient, the user's explicit The file on the file system 210 constructed in the magnetic disk device 21 by the processing of the hierarchical storage management program 12 is moved to the magnetic tape device 32 as the secondary storage device by the designation or setting, and the data block 213 of the magnetic disk device 21 is moved. Among them, the data block 213 of the file is released. At this time, the information of the moved file is added to the magnetic disk database 22 and the magnetic tape database 31. In the hierarchical storage device, the state in which the file is moved to the low-speed access device unit 3 in this way is called a migration state. As a result, the free capacity of the file system 210 increases. Further, when the application program 11 or the user accesses a file that has been moved from the magnetic disk 21 to the magnetic tape 32, it is automatically referred to the magnetic disk database 22 and the magnetic tape database 31 by the processing of the hierarchical storage management program 12. Thus, the file is copied from the magnetic tape device 32 to the magnetic disk device 21. As a result, the user can access without being aware of where the actual file exists.

  Next, a recovery method using the restoring means of the hierarchical storage device of this embodiment will be described. This restoration method is implemented by the CPU 41 of the processing unit 4 executing the programs 11 to 13 of the software unit 1.

  First, the backup means of the magnetic disk device 21 will be described. This backup means reads the inode list 212, data block 213, and magnetic disk database 22 of the magnetic disk device 21 and outputs them to a file on the arbitrary magnetic disk device 21 or to the magnetic tape device 32. When reading a data block, it is determined whether the file has been migrated. If the file has been migrated, the data block of the file does not exist in the magnetic disk device 21, that is, exists only in the magnetic tape device 32. Therefore, it is not necessary to read the data block, and backup is performed as fast as that. Is possible.

  As the backup means, a dump command which is a general command in the UNIX operating system is used. The dump command backs up the entire file system 210 on the magnetic disk device 21 at once.

  FIG. 3 shows the data structure of backup data. In FIG. 3, the backup data 6 has a structure comprising a set of an inode 61, a data block 62, and a magnetic disk database (HSM DB) 63 for hierarchical storage management. Among these, the inode 61 includes an inode number 611, file owner information (user ID) 612, group information (group ID) 613, file creation time 614, and file update time 615, and a file type (not shown) (normal file, directory) Data such as symbolic links (file types such as files pointing to other files).

  In the backup data 6, in the case of a file moved from the magnetic disk device 21 to the magnetic tape device 32, that is, a migrated file, the data in the data block 62 is missing and the data is moved from the magnetic disk device 21 to the magnetic tape device 32. In the case of a file that has not been migrated, that is, a file that has not been migrated, data in the HSM DB 63 is lacking and is backed up. In addition, in the case of a file copied from the magnetic disk device 21 to the magnetic tape device 32, that is, a file that has been migrated but the data block on the magnetic disk device 21 is not released, as shown in FIG. 62, HSM DB 63 is backed up as a set.

  Next, the restoration unit 111 for restoring the file on the magnetic disk device 21 from the back-up data 6 by the backup unit will be described.

  FIG. 4 illustrates a restoration method by the restoration unit 111 of the conventional hierarchical storage device, and FIG. 5 illustrates a restoration method by the restoration unit 111 of the hierarchical storage device of this embodiment.

  The conventional restoration means 111 shown in FIG. 4 creates a directory on the magnetic disk device 21 based on the inode list 212, creates a file to be restored, and restores the data block 213 as necessary, thereby providing a database for the magnetic disk device. 22 is restored and the magnetic tape database 31 is updated.

  In this case, when the restoration unit 11 restores the directories and files to be restored to the magnetic disk device 21, all directories are restored first, and then the files are restored. However, the inode number assigned by the kernel 13 is free. In other words, since an unused inode is automatically assigned, it is almost the same as the inode number at the time of backup. For this reason, it is necessary to update the magnetic disk device database 22 and the magnetic tape database 31 based on the new inode number for the file to be restored, and the more file information registered in these databases, the longer the search time. There was a problem that took. Furthermore, since these databases need to be updated via the hierarchical storage management program 12, the communication time between the restore means 11 and the hierarchical storage management program 12 cannot be ignored.

  On the other hand, in the restoration means 11 of the embodiment shown in FIG. 5, when restoring the directories and files to be restored to the magnetic disk device 21, all directories are restored first, and then the files are restored. Also, the inode number at the time when the backup is taken for the kernel 13 is designated and restored. The designation of the inode number is performed using the inode number 611 included in the inode 61 in the backup data 6 shown in FIG.

  Here, in the restoration unit 111 of the conventional example, when restoring a backed up file, among the inodes 61 included in the backup data 6, the file owner information 612, the group information 613, the file creation time 614, the file update time 615, etc. The data was used for recovery, but the inode number 611 was not used. This is because it is natural that the inode number changes during recovery.

  On the other hand, the restore means 111 of the present embodiment is devised so that the inode number 611 is used at the time of restoration and the inode number 611 is not different from that at the time of backup creation. Thereby, in this embodiment, since the inode number of the file to be restored is not different from that at the time of backup creation, there is no need to update the magnetic tape database 31, and further, the restore means 11 does not go through the hierarchy management program 12. Can directly rewrite the magnetic disk database 22, and as a result, high-speed recovery is possible.

  Next, with reference to FIG. 6, the operation | movement by the restoration | recovery method of the hierarchical storage apparatus of a present Example is demonstrated in detail.

  First, the restore unit 11 designates the path and inode number of the file to be restored and calls the specific inode allocation request system call 5 to the kernel 13. The inode number of the recovery target file is designated using the inode number 611 included in the backup data 6 as described above.

As a result, the inode list reading means 13 0 1 on the kernel 13 side is executed to lock the super block 211 of the file system 210 (step A1). This is to limit reading / writing to the file system 210 to one at a time.

Then, inode list reading unit 13 0 1 After successfully locked superblock 211, is stored in the superblock 211 extracts the maximum inode number (step A2). The maximum inode number is a fixed value set in advance according to the file system 210.

Next, the inode list reading unit 13 0 1 compares the inode number input by the restoration unit 11 with the maximum inode number (step A3). As a result, if the maximum inode number is exceeded, an error is returned. If the maximum inode number is not exceeded, the disk block address on the file system 210 where the inode specified by the inode number exists is calculated. (Step A4). In this calculation, the inode number 0, that is, the address of the disk block in which the head is stored, is a value that is uniquely obtained depending on the size of the file system, and the number of inodes included in one disk block is also determined. Therefore, a disk block address where an inode specified by an arbitrary inode number exists is obtained based on both values.

Next, the inode list reading unit 13 0 1 passes the obtained disk block address to the device access unit 1321, and reads the disk block from the inode list area 212 of the file system 210 (step A5).

  Next, the inode allocation unit 1302 on the kernel 13 side is executed, and the inode having the inode number designated by the restore unit 11 is retrieved from the inode list 212 of the read disk block (step A6), and corresponds to the retrieved inode list 212. It is determined whether the “disk inode” to be used is an unused inode (step A7).

  This determination is made based on whether or not the data in the inode has a value. The data in the inode used in this determination can be exemplified by a data area storing the file type (file type such as normal file, directory, symbolic link, etc.) to which the inode is assigned, and this file type has a value. If so, it is determined that the inode is in use, and if no value is entered, the inode is unused.

  If “disk inode” is in use according to the above determination, an error is returned, and if it is not used, “memory inode” is searched (step A8). “Memory inode” is obtained by caching (temporarily saving) “disk inode” on the memory 42.

  Next, the inode allocation unit 1302 determines whether or not the searched memory inode is an unused inode (step A9). Similarly to the above, this determination is also made based on whether or not the data in the inode, for example, the file type has a value, and if it has a value, it is determined that the inode is in use and the value is If not, it is determined that the inode is unused.

  If the “memory inode” is in use by the above determination, an error is returned, and if it is not used, a file such as “file type”, “user ID”, “group ID”, etc. is stored in the “memory inode”. Is stored (step A10).

  Next, the inode writing means 1303 on the kernel 13 side is executed, and the inode list area 212 of the file system 21 is updated so that the contents of the memory inode are directly reflected in the disk inode (step A11). Finally, the super block 211 is unloaded. Lock (step A12).

  Since the inode data of the recovery target file and the data of the file itself are recovered by the processing of the above steps A1 to A12, the data of the magnetic disk database 22 is subsequently recovered. At this time, since the inode number has not changed, the data in the database (HSM DB) 63 of the backup data 6 is written as it is.

  Therefore, according to the present embodiment, a file and a directory having the inode number designated by the application program 11 can be created on the file system 210, whereby the magnetic disk device 21 is backed up by the backup means. Since the inode number on the file system can be made the same when the restoration means 111 restores it, it is not necessary to rewrite the database of the application having the inode number in the database 22 as in the hierarchical storage management program 12, and the high speed Can be recovered. As a result, when the file system is destroyed due to a failure of the magnetic disk device or the like, the system stop time accompanying the restoration of the file system 210 can be significantly reduced.

  As other embodiments, in addition to normal files and directories that can be created on the file system 210, symbolic link files (files indicating other files) and special files (mainly for outputting I / O to the device). (Special files) can also be created by specifying the inode number, so that the file system 210 including these can be restored.

  Further, any number of sets of the file system 210 and the magnetic disk database (DB1) 22 can be defined in one system as long as the specifications of the operating system and hardware are not exceeded.

  In the above embodiment, the case where the magnetic disk device 21 is used as the primary storage device of the high speed access device unit 2 and the magnetic tape device 32 is used as the secondary storage device of the low speed access device unit 3 has been described. In addition to the above, a primary storage device (first storage device) such as a semiconductor memory is used for the high-speed access device unit 2 and a secondary storage such as a magneto-optical disk, an optical disk (DVD-R, etc.) is used for the low-speed access device unit 3. A device (second storage device) may be used.

  The present invention can also be applied to a file system restoration technique for a computer running a UNIX operating system and a method in which an application program creates a file having an arbitrary inode number using a system call in the UNIX file system.

It is a functional block diagram which shows the whole structure of the hierarchical storage apparatus based on the Example of this invention. It is a functional block diagram which shows the principal part structure of the hierarchical storage apparatus based on the Example of this invention. It is a figure which shows the data structure of backup data. It is a figure explaining the restoration method by the restoration means of the conventional hierarchical storage apparatus. It is a figure explaining the restoration method by the restoration means of the hierarchical storage apparatus based on the Example of this invention. It is a flowchart explaining the restoration method of the hierarchical storage apparatus based on the Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Software part 2 High speed access apparatus part 3 Low speed access apparatus part 4 Processing part 11 Application program 12 Hierarchical storage management program (HSM)
13 Kernel 21 Magnetic Disk Unit 22 Magnetic Disk Database (DB1)
31 Magnetic Tape Device 32 Database for Magnetic Tape (DB2)
41 CPU
42 memory 210 file system 211 super block 212 inode list 213 data block

Claims (6)

In a hierarchical storage device that runs on an operating system,
A first storage device having a file system that has an i-node including attribute information of the file and that uniquely identifies the file by the i-node number;
A second storage device for storing data including backup data of the file system;
When the file system is restored on the first storage device from the backup data on the second storage device, the i-node number of the recovery target file is determined using the i-node number included in the backup data. An inode number designating means to designate;
File allocating means for allocating a designated i-node number to a recovery target file of the file system ,
The file assigning means is means for judging whether or not a designated i-node number is not used on the file system, and assigning the i-node number to the recovery target file if not used. Hierarchical storage device. The i-node number designation means is a means for designating an arbitrary i-node number on the operating system and starting a system call;
The file allocation means, hierarchical storage system according to claim 1, characterized in that the means for assigning the i-node number specified by the system call in the recovery target file of the file system. A tiered storage device running on an operating system,
A first storage device having a file system that has an i-node including attribute information of the file and that uniquely identifies the file by the i-node number;
In a recovery method for a hierarchical storage device having a second storage device for storing data including backup data of the file system,
When the file system is restored on the first storage device from the backup data on the second storage device, the i-node number of the recovery target file is determined using the i-node number included in the backup data. A step to specify;
And a step of assigning the designated i-node number in the recovery target file of the file system,
The step of allocating the file is a step of determining whether or not the designated i-node number is not used on the file system, and allocating the i-node number to the recovery target file if not used. A recovery method of the hierarchical storage device. The step of designating the i-node number is a step of activating a system call by designating an arbitrary i-node number on the operating system,
4. The hierarchical storage device recovery method according to claim 3 , wherein the step of allocating the file is a step of allocating the i-node number designated by the system call to a recovery target file of the file system. A tiered storage device running on an operating system,
A first storage device having a file system that has an i-node including attribute information of the file and that uniquely identifies the file by the i-node number;
In a recovery program for a hierarchical storage device having a second storage device for storing data including backup data of the file system,
On the computer,
When the file system is restored on the first storage device from the backup data on the second storage device, the i-node number of the recovery target file is determined using the i-node number included in the backup data. A step to specify;
Assigning a designated i-node number to a recovery target file of the file system ,
The step of allocating the file is a step of determining whether or not the designated i-node number is not used on the file system, and allocating the i-node number to the recovery target file if not used. A recovery program for the hierarchical storage device . The step of designating the i-node number is a step of activating a system call by designating an arbitrary i-node number on the operating system,
6. The recovery program for a hierarchical storage device according to claim 5 , wherein the step of allocating the file is a step of allocating the i-node number designated by the system call to a recovery target file of the file system.

Images