WO2004111854A1 - File access management method, file access management device, and recording medium - Google Patents

Abstract

A file A consists of, for example, blocks B0, B1, and B2. When a process (update process 15) requests update (write) of the file A, the block to be updated (block B0 in this case) is copied to generate a copy block B0’. The update processing is executed to the original block B0. During the update process, if a reference (read) request is generated from another process (reference process 16), the copied block B0’ is referenced.

Description

 Description File access management method, file access management device, recording medium

 The present invention relates to a system for accessing one file from a plurality of processes, and more particularly to a method for enabling a process to refer to the same file even when a certain process is updating the file. Kyogi W

 In conventional file access, if a process updates the file while another process refers to the same file, the file may refer to the incomplete and meaningless content. Previously, file locks existed as a way to avoid this. File locking is a mechanism that prevents other processes from accessing and updating while the file is locked.

 When the above file lock is used, the process by other processes is suppressed until the process that has acquired the lock first releases the lock. For this reason, there is a problem that the longer the process of the process that has acquired the lock, the longer the waiting time of the other process.

However, if it is sufficient that the file can be referred to with meaningful contents even if the content of the file is not the latest, there is a request that the file before update should be referred to even if the file is being updated. Another workaround is transaction management of a database system (DB). This is a transaction update This is a mechanism that performs processing in units of application and excludes transactions. When using such a DB management system, it is necessary to register the data to be used in the form of a DB table instead of a file, and use a DB-specific access method (such as SQL) when accessing. There is a need.

 Here, conventionally, the following technology has been proposed.

 For example, in the document registration and retrieval method described in Patent Document 1, in a document database in which a block is stored at a fixed number of registered documents and the storage area of the block data is managed by a management block, the document database is searched. When a document registration request is issued to the server, the contents of the corresponding block are copied and the registered contents are reflected in the copy, so that when a system failure occurs and the system is restarted, the block before registration is blocked. Determine if you want to use the password or refer to the post-registration block, and restore the document database before the failure.

 Furthermore, when a registration and a search request are issued for the same block, the registration refers to the management block after registration, and the search refers to the management block before registration. The feature is that it can be performed in parallel.

Also, for example, in the control method of the microphone port processor system described in Patent Document 2, the control method when the contents of a specific block of a memory area is changed is described. Has proposed a control method that can access this without waiting and read a series of consistent block data. More specifically, in the invention described in Patent Document 2, partial block data of the basic database stored in the variable memory is changed. If it becomes necessary, a process to prepare the changed block data for the work block provided using an area different from the area storing the basic database is performed, but a read command is received during this process. In this case, the block data before the change is read. Next, when the changed block data is completed in the work block, while the contents of this work block are sequentially transferred to the original basic-database area block, when the read command is received, the work block is received. Read the changed block data from. .

Patent Document 1

 Japanese Patent Application Laid-Open No. Hei 11-259595

 Patent Document 2

 SUMMARY OF THE INVENTION In both of the prior arts 1 and 2 described above, update processing is performed on the copy block, and when the update is completed, the updated data is written to the original block. To return, it took time to process extra processing.

 An object of the present invention is to provide a file access management method or the like that enables a process to refer to another process even when a process is updating a file, and makes the update process during normal operation much faster. It is to be. Disclosure of the invention

In the file access management method according to the present invention, when an arbitrary file is requested to be updated from an arbitrary process, the file becomes an update target in the file. In addition to creating a copy block of the block to be updated, the update process is executed on the original block, and if another process requests a reference to the file during the update process, the copy block is referred to.

 In the above file access management method, the block to be updated is the original block, and the block to be referred to is the copy block. There is no need to rewrite, and the update performance during normal operation is much faster. Also, by performing block-by-block copying, unlike the method of making a copy of the entire file, a copy can be made in a fixed time regardless of the file size.

 In the above file access management method, for example, when the update process is completed, or after the update process is completed, if there is no process that refers to the copy block, the copy block is released. You may make it.

 Further, in the above file access management method, for example, for each file, whether or not to apply the file access management method when an update Z reference request is made to the file may be designated. Yo o

 As described above, it is possible to specify whether or not to apply the file access management method according to the present invention for each file, not for all files. When the method of the present invention is not applied, for example, a conventional file access method may be applied.

The program may be read from a computer-readable storage medium storing a program for causing a computer to perform the same control as the function performed by each configuration of the present invention, and executed by the computer. Can solve the aforementioned issues. Wear.

BRIEF DESCRIPTION OF THE FIGURES

 The present invention will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

 FIG. 1 is a diagram showing an example of a system to which a file access management method according to the present embodiment is applied.

 FIG. 2 is a diagram showing a specific example of the block management book.

 FIG. 3 (a) is a diagram showing a state of file management at the time of a file update request, and FIG.

 FIG. 4 is a processing flowchart when there is a request from the update process.

 FIG. 5 is a processing flowchart when there is a request from the reference process.

 FIG. 6 is a processing flow chart in a case where it is specified whether or not the processing of this example is applied to each file.

 FIG. 7 is a diagram illustrating an example of a hardware configuration of a computer that executes the file access management method of the present example.

 FIG. 8 is a diagram showing an example of a recording medium storing a program for causing a computer to execute the file access management method of the present example, and an example of downloading the program.

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings. Figure 1 shows an example of a system that applies the file access management method according to this example. FIG.

 In the illustrated example, a computer 17 (a personal computer (PC), a workstation (WS), etc.) accesses a disk 17 such as a hard disk (HD) as a system for executing the file access management method according to the present example. The configuration is exemplified. When a process accesses a file, the update / reference request is usually first transferred to the file system layer 12 in the operating system (OS) 11. Then, the disk 17 is accessed from the file system layer 12 via the driver layer 13.

 The disk 17 may be internal to the computer 10, may be external, or may be shared with a plurality of other computers (not shown). In the case of a configuration shared by multiple computers, the update process 15 and the reference process 16 may have been requested by another computer.

 The disk 17 has a management block area 18 and a data area 19. The data area 19 stores files and the like. The management area 18 stores management information of each file stored in the data area 19. In the example shown in the figure, the management information is file name 21, date 22, policy 23, and data block No. 24.

 The file name 21 is the name of the file stored in the data area 19, and the date 22 is the newly registered Z update date and time. The policy 23 will be described later.

Here, the data block No. 24 will be described. A non-volatile external storage device such as a magnetic disk or a magnetic tape uses a method of accessing a block unit such as a physical block and a logical block. Data block No. 24 is assigned to each block of the file. The stored number is stored.

 In the file access management method according to the present example, when an update (write) request of an arbitrary file occurs, a block to be updated in each block constituting the file is updated before starting the update process of the block. Make a copy of it. Unlike block-by-block copying, making a block-by-block copy

 -Duplicate in a fixed amount of time regardless of file size

 • File name does not change

There is such a merit.

 Then, information relating to the block in which the copy has been made is managed by the block management book 14. In the implementation, the layer that manages the copy block may be anywhere. For example, a NAS (Network Attached Storage) layer, a file system layer, a driver layer, etc. are considered. Here, a case where the copy protocol is managed in the file system layer 12 will be described as an example. The file system layer 12 manages copy blocks by using a block management book 14 as shown in FIG.

 It is to be noted that, regardless of which layer is managed in the implementation, a configuration for realizing such a file access management method of this example is hereinafter referred to as a file access management device (not shown). Further, in the file access management device, the functional unit performing the processing in FIG. 4 is referred to as an update control unit, and the functional unit performing the processing in FIG. 5 is referred to as a reference control unit (neither is shown).

 FIG. 2 shows a specific example of the block management book 14.

The illustrated block management book 14 is for registering copy block management information, and includes data items of a block number 31, an address number 32, and a reference count 33. The block number 31 stores the data block No. 24 of the block in which the above copy was made. Address number 3 2 contains the block number The address of the storage area to which the data of the block 31 is copied is stored. Reference count 33 stores the number of processes currently referencing the copy block. Here, the file access management method of this example will be described with reference to the image shown in FIG. The reference count 33 will be described later with reference to a flowchart.

 First, as shown in FIG. 3 (a), it is assumed that file A is composed of blocks B0, Bl, and B2 and records meaningful contents. When an arbitrary process (update process 15) issues an update request (write) for file A, the above file access management device does not copy the entire file, but rather updates the block to be updated (here, block B). Control so that only 0) is copied. In the illustrated example, the copy block is represented as B0. Then, the block number of block B0 and the storage address of copy block B0, are stored in block number 31 and address number 32, respectively.

 Then, the update process according to the request of the update process 15 is performed not on the copy block BO 'but on the original block B O (hereinafter, referred to as an original block).

 If a read request from another process (reference process 16) occurs during the update, as shown in Fig. 3 (b), the file access management device refers to the block management book 14 For the block being updated, that is, the block B 0 in which the block number 31 is registered in the block management book 14, the copied block B 0 ′ is referred to. For the blocks Bl and B2 that are not updated in the file A, the original blocks are naturally referred to.

 By doing so, it is possible to refer from the reference process 16 in parallel with the update process while the contents before the update are guaranteed.

The important and characteristic point here is that the updated block is the original block. The point is that in the lock, the block referred to is a copy block. If the copy block is set as the update side as in the past, it will be necessary to write back to the original block when the update is completed. However, if the copy block is the update side, the file before update can be recovered quickly by simply discarding the unfinished block being updated when a failure occurs. On the other hand, when the original block is the update side as in the method of this example, it is necessary to rewrite the copy block to the original block when a failure occurs.

 However, the update performance during normal operation is much faster with the method of this example. And, in general, the time when a failure occurs is negligible in the total operation time, and the disadvantages associated with the occurrence of a failure are very small.

 A copy block may be released if it has not been referenced by another process when the update of the corresponding original block is completed. However, if the copy block has been referenced when the corresponding original block has been updated, it must not be released until the reference is completed. For this reason, reference count 33 is used to manage whether a block is being referenced. In other words, the value of the reference count 33 is 0 in the initial state, and the value of the reference count 33 is incremented (+1) every time there is a reference request during the update process. Then, every time the reference is completed, the value of the reference count 33 is decremented (1-1). Since the number of other processes requesting reference during update processing is not limited to one, the value of reference count 33 may be 2 or more. Then, when the value of the reference count 33 becomes 0, it means that there is no process being referred to, and the copy block may be released.

FIG. 4 is a processing flowchart when a request from the update process 15 is received, and FIG. 5 is a processing flowchart when a request from the reference process 16 is received. First, FIG. 4 will be described.

 Upon receiving the update request (write) from the update process 15, the file access management device first controls the data of the block to be updated to be copied to another storage area (step S 11). Next, the data block No. 24 of the block to be updated and the address of the storage area to which the data of the block are copied are stored in the block management book 14 in the block number 31 and the end address number 3 respectively. Register in step 2 (step S12). Then, data update processing is performed on the original block to be updated (step S13).

 When there are a plurality of blocks to be updated, steps S1 :! to S13 are repeated and the update is completed for all the blocks to be updated (step S14, YES). Referring to the value of the reference count 33 of the book 14, if the value is not 0 (> 0) (step S 15, NO), the update processing is terminated as it is. On the other hand, if the value of the reference force account .33 is 0 (step S15, YES), the copy block is released (step S16), and the copy block management information is stored in the block management book. Delete from step 14 (step S17).

 Next, the processing of FIG. 5 will be described.

 When the file access management device receives a reference request (read) from the reference process 16, the file access management device accesses the referenced file in units of blocks as described above, so that all blocks related to the referenced file are accessed. Until the block is accessed, the processing of steps S21 to S25 is executed for each block.

First, a copy block is created for the block to be accessed. It is determined whether or not it has been performed (step S21). This is determined based on whether or not the data block No. 24 of the block to be accessed is registered in the block management book 14.

 If a copy block has been created (step S21, YES), first, the value of the reference count 33 of the block management book 14 is incremented (+1) (step S22). . Next, a copy block is referred to based on the address number 32 (step S24), and when the reference is completed, the value of the reference force point 33 is decremented (11) (step S24).

 On the other hand, if the copy block has not been created (step S21, NO), the original block is referred to (step S25). When the above processing is performed for all blocks related to the file to be referred to (that is, when the reference of the file is completed) (step S26, YE S), the reference point of the block management book 14 33 Referring to the value of 3, if the value is not .0 (> 0) (step 7, NO), the process is terminated.

 On the other hand, if the value of the reference force point 33 is 0 (step S27, YES), it is further determined whether or not the original block is being updated (step S28). If the update is being performed (step S28, YES), the process is terminated. On the other hand, if the original block is not being updated (step S28, NO), this copy block is released (step S29), and the management information of this copy block is deleted from the block management book 14 (Step S30).

If the above step S27 is NO, there is another process being referred to, and the process related to the process that has been completed last is referred to. In the process of FIG. 5, the copy block is released. However, if the original block is still being updated at that time, the copy block will be released when the update is completed in the processing in Fig. 4. In other words, the copy block is released when all processes are completed for that block regardless of the update / reference.

 Finally, the policy 23 shown in FIG. 1 will be described.

 The file access management method of this example shown in FIGS. 4 and 5, etc., is not applied uniformly to all the files stored in the data area 19 of the disk 17, but is applied to each file. May be specified, and this specification is stored in policy 23. For example, it is assumed that a flag is stored in policy 23, and is applied in the case of flag ON, and is not applied in the case of flag OFF.

 As a result, the file system layer 12 first, every time there is a request for updating or referencing a file from an arbitrary process, as shown in Fig. 6, first, the policy 2 of the management block for the file to be updated or referenced. Referring to FIG. 3 (Step S41), if the flag is ON (Step S41, YES), the processing of FIG. 4 or FIG. 5 is executed (Step S42), and if the flag is OFF, (Step S41, NO) Performs Conventional Processing (Step S43) FIG. 7 is a diagram showing an example of a hardware configuration of a computer 40 that executes the above file access management method.

The computer 40 shown in the figure has a CPU 41, a memory 42, an input device 43, an output device 44, an external storage device 45, a medium drive device 46, a network connection device 47, and the like, which are connected to a path 48. It has become. The configuration shown in the figure is an example, and the configuration is not limited to this. The CPU 41 is a central processing unit that controls the entire computer 40. The memory 42 temporarily stores a program or data stored in the external storage device 45 (or a portable recording medium 49) when executing a program or updating data. Memory. The CPU 41 executes processing for realizing the file access management method shown in FIGS. 4, 5, 6, and the like by using the program data read into the memory 42.

 The input device 43 is, for example, a keyboard, a mouse, a touch panel, or the like.

 The output device 44 is, for example, a display, a printer, or the like.

 The external storage device 45 is, for example, a magnetic disk device, an optical disk device, a magneto-optical disk device, or the like. The external storage device 45 is used to cause the computer 40 to execute the file access management processes shown in FIGS. Stores programs / data.

 The medium drive unit 46 stores the program stored in the portable recording medium 49.

Read one or more data. The portable recording medium 49 is, for example, F D

Disk), CD-ROM, memory card, DVD, magneto-optical disk, etc. The network connection device 47 is configured to be connected to a network to enable transmission and reception of program Z data and the like to and from an external information processing device.

 FIG. 8 is a diagram showing an example of a recording medium on which the above-mentioned program is recorded and download of the program.

 As shown in the figure, the program Z for realizing the functions of the present invention may be read from a portable recording medium 49 storing the data to the computer 10 side, stored in the memory 42, and executed. The program data is stored in the storage unit 51 of the external server 50 via a network 60 (such as the Internet) connected by a network connection device 47. Data may be downloaded.

In addition, the present invention is not limited to the apparatus / method, and may store the program Z data. The stored recording medium (portable recording medium 49, etc.) can be configured as itself, or can be configured as the program itself. Industrial potential

 As described in detail above, according to the file access management method of the present invention, even when a certain process is updating a file, it is possible to refer to the file from another process and to update the file during normal operation. Processing becomes much faster. References from other processes guarantee the contents before updating, and can be accessed in parallel without changing the programs used so far.

Claims

The scope of the claims 1.When an arbitrary file is requested to be updated from an arbitrary process, the block to be updated in the file is copied to another area, and the update processing is performed on the original block,  A file access management method for referring to the another area when another process requests to refer to the file during the update process.  2. When the update process is completed or after the update process is completed, it is checked whether there is a process that refers to the another area, and if there is no such process, 2. The file access management method according to claim 1, wherein said another area is released.  3. Claims 1 or 2 ', characterized in that, for each file, specify whether to apply the file access management method when there is an update / reference request for that file. File access management method.  4. On the computer,  A function of copying a block to be updated in the file to another area when an update request of an arbitrary file is issued from an arbitrary process, and executing an update process on the original block;  A function of referring to the another area when another process requests a reference to the file during the update processing;  The program to realize. 5. When the update process is completed or after the update process is completed, it is checked whether or not there is a process that refers to the another area, and if no such process exists, The program according to claim 4, further comprising a function of releasing the another area. 6. The program according to claim 4, wherein for each file, whether or not to apply the function of the program when an update / reference request is made to the file is specified.  7. On the computer  A function of copying a block to be updated in a file to another area when an update request for an arbitrary file is issued from an arbitrary process, and executing an update process on the original block;  A function of referring to the another area when another process requests a reference to the file during the update processing;  The computer-readable recording medium storing a program for executing the program.  8. When the update process is completed or after the update process is completed, it is checked whether or not there is a process that refers to the another area, and if no such process exists, 8. The recording medium according to claim 7, further comprising a function of releasing said another area.  9. The recording medium according to claim 7, wherein, for each file, whether or not the function of the program is applied when an update Z reference request is made to the file is specified.  10 0. Update control for copying a block to be updated in another file to another area when an update request for an arbitrary file is issued from an arbitrary process, and executing update processing on the original block. Means for referring to the another area when another process requests to refer to the file during the update processing; and  A file access management device comprising: