JP3050194B2 - A system for dynamically adding a shared memory file between hosts, a method for dynamically adding a shared memory file between hosts, and a recording medium storing a program for dynamically adding a shared memory file between hosts - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inter-host shared memory file dynamically expanding a shared memory file area by dynamically increasing the number of shared memory files shared between multiprocessor systems. The present invention relates to an additional system, a method for dynamically adding a shared memory file between hosts, and a recording medium on which a program for dynamically adding a shared memory file between hosts is recorded.

[0002]

2. Description of the Related Art Japanese Patent Laid-Open Publication No.
A technique for changing the size of one shared memory area shared by a plurality of application programs in one host computer is described. The host computer includes a shared memory area accessible from a plurality of application programs, a memory management table for managing the shared memory area, and a shared memory area in response to a request for extending the shared memory area from the plurality of application programs. A required maximum size determination function for determining the maximum size of the expansion request, a memory area expansion function for expanding the shared memory area according to the maximum size,
In response to a request for reducing the shared memory area from a plurality of application programs, a required minimum size determination function for determining a minimum size of the reduction requests, and a memory area for reducing the shared memory area according to the minimum size It consists of a reduction function.

[0003]

However, in this conventional technique, each of a plurality of application programs does not notify another application program of an enlargement / reduction request to the shared memory area issued by itself. Therefore, even if each application program requests the reservation of the shared memory area of a desired size, there is a possibility that the shared memory area is smaller than the desired size and the desired size cannot be reserved frequently. In such a case, the shared memory area must be expanded each time, so that there is a problem that access to the shared memory area becomes very inefficient.

When this conventional technique is applied to the management of a plurality of shared memory areas, that is, each of a plurality of application programs manages a plurality of different shared memory areas without notifying another application program. In the case of enlargement and reduction, there is a problem that a large amount of fragmentation occurs in a storage area including the shared memory area.

An object of the present invention is to enable a plurality of shared memory files to be efficiently shared between multiprocessor systems.

Another object of the present invention is to make it possible to efficiently use a storage area including a plurality of shared memory files shared between multiprocessor systems.

[0007]

A first system for dynamically adding a shared memory file between hosts according to the present invention comprises: an external storage device including a shared memory file composed of a plurality of data blocks of the same size; It has a hash table composed of a plurality of entries corresponding to any of the data blocks, and when storing data in each of the plurality of data blocks, hashing the identifier of the data and performing hashing in the hash table A plurality of host computers for determining a corresponding entry and storing the data in a data block corresponding to the entry; and inter-host lock means for exclusively controlling access to the hash table of the plurality of host computers. Wherein the external storage device further comprises: A control file for storing correspondence information between each data block of the shared memory file and each entry of the hash table, wherein each of the plurality of host computers further includes its own host computer prior to the other host computers. When started, the host-to-host locking means exclusively controls access to the hash table of the other host computer by the other host computer, and acquires the number of all data blocks of the shared memory file, Creating the hash table including at least the number of entries of the data block or more, storing correspondence information between each entry of the hash table and each data block of the shared memory file in the control file; The other host computer by another host computer by means The exclusive control of access to the hash table held by the host computer is released, and when the host computer is started after the other host computer, the inter-host lock means causes the other host computer to execute the other host computer. Exclusively controls access to the hash table possessed by the computer, reads the correspondence information from the control file, creates the hash table in the host computer based on the correspondence information, A shared memory file access unit that releases exclusive control of access by the host computer to the hash table of the other host computer.

In a second dynamic inter-host shared memory file adding system according to the present invention, in the first dynamic inter-host shared memory file adding system, each of the plurality of host computers further includes an external storage device. Shared file adding means for adding the shared memory file to;
After the shared memory file is added to the external storage device by the shared file adding means, the inter-host locking means exclusively controls access to the hash table of the other host computer by the other host computer by the inter-host locking means. Associating any one of the plurality of entries of the hash table corresponding to a specific data block of the existing shared memory file with any of the plurality of data blocks of the added shared memory file, If the associated entry corresponds to the result of hashing the data identifier of the specific data block, the data of the specific data block is added to the associated data block in the added shared memory file. Transported and the pair stored in the control file Update information, characterized in that it comprises a hash table updating means for releasing exclusive control of access to the hash table with the said other host computer by other host computers by a locking means between said host.

In a third dynamic inter-host shared memory file adding system according to the present invention, in the second dynamic inter-host shared memory file adding system, the hash table updating means may further include the control by another host computer. When the correspondence information stored in the file is updated, the hash table of the host computer system is updated based on the updated correspondence information.

In a first method for dynamically adding a shared memory file between hosts according to the present invention, each of a plurality of host computers comprises:
When storing data in each of a plurality of data blocks of the same size constituting a shared memory file shared by the plurality of host computers, the data identifier is hashed to correspond to any of the plurality of data blocks. Determining a corresponding entry in a hash table composed of a plurality of entries, and storing the data in a data block corresponding to the entry; and When the host computer is started before the other host computer, exclusive access control of the other host computer to the hash table of the other host computer is performed, and the number of all data blocks of the shared memory file is obtained. And the hash comprising at least as many entries as the number of data blocks. And creates correspondence information between each entry of the hash table and each data block of the shared memory file in a control file shared by the plurality of host computers. The exclusive control of access to the hash table having the same is released, and when the own host computer is started after the other host computer, the other host computer accesses the hash table of the other host computer. Exclusively controls access, reads the correspondence information from the control file, creates the hash table in the host computer based on the correspondence information, and sends the hash table to the hash table of the other host computer by another host computer. Table creation step to release exclusive access control Characterized in that it comprises and.

In a second method for dynamically adding a shared memory file between hosts according to the present invention, in the first method for dynamically adding a shared memory file between hosts, each of the plurality of host computers adds the shared memory file. A shared file addition step, and each of the plurality of host computers exclusively controls access to the hash table of the other host computer by another host computer, and to a specific data block of the existing shared memory file. A corresponding one of the plurality of entries of the hash table is associated with one of the plurality of data blocks of the shared memory file added in the shared file adding step, and the associated entry in the hash table is Hash the data identifier of the specific data block. Transfer the data of the specific data block to the associated data block in the added shared memory file, and update the corresponding information stored in the control file. A hash table updating step of releasing exclusive control of access to the hash table of the other host computer by the other host computer.

In a third method for dynamically adding a shared memory file between hosts according to the present invention, in the second method for dynamically adding a shared memory file between hosts, the step of updating the hash table is further performed by another host computer. When the correspondence information stored in the file is updated, the hash table of the host computer system is updated based on the updated correspondence information.

In the first recording medium of the present invention, when data is stored in each of a plurality of host computers, each of a plurality of data blocks of the same size constituting a shared memory file shared by the plurality of host computers. Hashing the identifier of the data to determine a corresponding entry in a hash table composed of a plurality of entries corresponding to any of the plurality of data blocks, and storing the data in the data block corresponding to the entry. When the host computer is started earlier than the other host computer, exclusive control of access to the hash table of the other host computer by the other host computer is performed, Obtain the number of all data blocks of the shared memory file, at least the number of data blocks or more Creating the hash table composed of entries, and storing correspondence information between each entry of the hash table and each data block of the shared memory file in a control file shared by the plurality of host computers; Release exclusive control of access to the hash table held by the other host computer, and if the own host computer is started after the other host computer, the other host computer Exclusive control of access to the hash table owned by the host computer, reads the correspondence information from the control file, creates the hash table in its own host computer based on the correspondence information, and executes the other host by another host computer. Release exclusive control of access to the hash table held by the computer And characterized by recording a program for causing a hash table creation processing and that.

According to a second recording medium of the present invention, in addition to the program recorded on the first recording medium, a shared file adding process for adding the shared memory file to each of the plurality of host computers, The exclusive control of access to the hash table of the other host computer by the host computer, and any one of the plurality of entries of the hash table corresponding to a specific data block of the existing shared memory file is stored in the shared file. In the case where the entry is associated with any of the plurality of data blocks of the shared memory file added in the adding step, and the associated entry in the hash table corresponds to a result of hashing the data identifier of the specific data block. In the shared memory file added Data of the specific data block is transferred to the data block, the correspondence information stored in the control file is updated, and exclusive control of access to the hash table of the other host computer by another host computer is performed. A program for performing a hash table updating process to be canceled is recorded.

According to a third recording medium of the present invention, in addition to the program recorded on the second recording medium, the correspondence information stored in the control file by another host computer is used in the hash table update processing. In the case where the hash table has been updated, a program added with a process of updating the hash table of the host computer system based on the updated correspondence information is recorded.

[0016]

Embodiments of the present invention will now be described in detail with reference to the drawings.

Referring to FIG. 1, in a first embodiment of the present invention, an extended memory 100 and hosts 1 (101) and 2 (102) operating under program control and accessing the extended memory 100 in a shared manner. ) And an inter-host lock unit 103 that enables exclusive control between the hosts 1 and 2
It is composed of

The extended memory 100 includes a shared memory file 104 and a shared memory file 110 to be added.
And a control file 109. In addition, the hosts 1 and 2 communicate with the extended memory file access unit 10.
7, a hash table 106, an access unit 105, a hash table updating unit 108, and an application program 112. The host 1 further includes a shared file adding unit 111.

The extended memory 100 is a freely readable and writable external storage device shared by the hosts 1 and 2.

In the present embodiment, a description will be given using an example in which there are two hosts, but a configuration having two or more hosts may be employed. In the present embodiment, the shared file adding unit 1
Although 11 exists only in the host 1, this does not specify the execution host of the shared file adding unit 111.

The inter-host lock means 103 is a device for controlling an exclusive operation between the hosts 1 and 2. However, the lock requests from a plurality of hosts may be program-controlled as software using the extended memory 100.

The shared memory file 104 stores the data shared between the hosts 1 and 2
0 and contains a plurality of data blocks of the same size.

The shared memory file (additional portion) 110 is
This is a file on the extended memory 100 for storing data shared between the hosts 1 and 2, and is added after the shared memory file 104.

The control file 109 is an area for holding the relationship between the shared memory file and the hash table, and is set by the shared memory file access unit 107 of the host that has started first. Also, when the shared memory file 110 is added by the shared file adding unit 111, it is updated by the hash table updating unit 108.

The access means 105 is an extension memory 100
It is a driver that reads and writes data from and to the device.

The hash table 106 is a correspondence table between identifiers of data registered on the shared memory file 104 and positions of the data (data blocks) on the shared memory file 104. Application program 11
The position of the data (data block), that is, the data block address, is stored in the entry when hashing the identifier of the data passed from 2 and mapping it to the total number of entries in the hash table. FIG. 2 shows the relationship between the hash table and the shared memory file.

Shared memory file access means 107
, Based on a request from the application program 112, stores the data in the shared memory file 104 and the data in the added shared memory file 110 in the hash table 10
6 and access. When the host is started, if another host has not been started yet (confirmation of whether or not the other host is started is performed by a well-known function of a computer system composed of multiple hosts), The number of data blocks of all the shared memory files existing in the control file 109 is obtained, the hash table 106 is created based on the value, and the information is stored in the control file 109.
To be stored. If another host was already running,
The information is read from the control file 109, and the hash table 106 is created based on the information.

The application program 112 is a user program for accessing data shared between hosts.

The shared file adding means 111 adds a new shared memory file (additional portion) 110 to the shared memory file 104 shared by the hosts without changing the hash width (the number of entries in the hash table 106). By dynamically updating the data block address in the entry of the hash table 106, the data space of the shared memory file is dynamically expanded.

FIG. 2 shows the relationship between the hash table 106, the shared memory file 104, and the shared memory file (additional) 110. The hash table 106 is
It is composed of a plurality of entries storing data block addresses 201 to 208. The number of entries is a theoretical limit, such as a value obtained by dividing the maximum capacity of the extended memory by the data block length, because the number of entries is a specification of the shared memory space.

In the data block addresses 201 to 208, the addresses of the data blocks of the shared memory file recognized at the time of startup are sequentially stored. In the example of FIG.
Since the number of data blocks is 4 and the number of hash table entries is 8, the association as shown by the arrow in the figure (data block address 1 (201) and data block address 5 (205) are both data block 1 (205) 21
1) store a pointer to the data block address 2
(202) and data block address 6 (206) store a pointer to data block 2).
Is made.

When the shared memory file (additional portion) 110 is added, the hash table updating means 108 causes the data block addresses 5 and 6 (205, 20).
6) is modified to point to data blocks 5 and 6 (215, 216) on the added shared memory file. By this operation, the number of available data blocks, that is, the data space of the shared memory file can be expanded without changing the hash width (the number of hash entries). It should be noted that a data block is usually composed of a plurality of data areas. The correspondence between an application program and a data block is performed by hashing an arbitrary data identifier common to application programs sharing data and mapping the identifier to a hash width.

Next, the operation of the first embodiment of the present invention will be described in detail with reference to FIGS.

First, shared memory file access means 1
The initialization process at the time of starting up the first host by using the command 07 will be described.

First, an inter-host lock on the hash table 106 is acquired by the inter-host lock means 103 (step 301). Next, the total number of data blocks is obtained for all the shared memory files 104 on the extended memory 100 (step 302). Further, the hash table 106 having the number of entries corresponding to the number of data blocks is created. All data block addresses on the created hash table 106 are set so as to equally point to data blocks on the shared memory file 104 existing on the extended memory 100 (step 303). Next, information of the created hash table 106, specifically, the number of hash entries and the number of data blocks are stored in the control file 109 on the extended memory 100 (step 304). Finally, the inter-host lock on the hash table 106 is released (step 30).
5), the initialization process is completed.

The host started secondly reads the number of data blocks and the number of hash table entries of all the shared memory files 104 from the control file 109 instead of the processing of step 302 in the flowchart of the initialization processing. And hash table 1 accordingly
06 is performed. Thereafter, the process of step 304 is skipped, and the process of step 305 is performed.

Next, processing for adding the shared memory file 110 will be described.

First, the shared file adding means 111
The shared memory file (additional portion) 110 is allocated on the extended memory 100 in response to a request to add the shared memory file 110 from S or an application program. Next, the shared file adding unit 111 instructs the hash table updating unit 108 to output the hash table 10 of the newly created shared memory file (added) 110.
Request registration to 6. The flowchart of FIG. 3 shows the subsequent processing.

The hash table updating unit 108 requests the shared memory file access unit 107 to lock the host with respect to the hash table 106 (step 311). Once the lock has been obtained, check the number of data blocks in the shared memory file (additional) 110,
One of the data block addresses on the hash table 106 indicating the same data block is changed to point to the newly added data block (step 312). Next, the identifier of the data on the data block to which the changed data block address points before the change is hashed again, and the data indicated by the data block address stored in the entry of the hash table 106 corresponding to the obtained value The data is stored in the block (step 313). As a result, the storage destination is either the original data block or the added data block indicated by the newly set data block address. Next, information on the additional data block is stored in the control file 109 (step 314). Finally, the inter-host lock on the hash table 106 is released (this release operation notifies other hosts of the release of the lock) (step 315).

The other host (shared file adding means 111) receiving the lock release notification in step 315
The host that does not execute the command maintains the consistency between the hosts for the shared memory file by updating the hash table 106 according to the information in the control file 109. Note that no data transfer occurs in this process.

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

Referring to FIG. 4, the embodiment of the present invention holds a DB 413 in addition to the extended memory 100, the hosts 1 (101) and 2 (102), and the inter-host locking means 103 as in FIG. Consists of an external storage device,
On the extended memory 100, a shared buffer 404 corresponding to the shared memory file 104, and an additional shared buffer 4
10 and a control file 109, on the hosts 1 and 2, access means 105, hash table 1
06, hash table updating means 108, shared memory file access means 107, shared file adding means 11
1 includes a shared buffer adding unit 411 corresponding to the application program 1, a DB registration / snooping program 412 corresponding to the application program 112, and a DB access unit 414.

Next, the operation of the embodiment of the present invention will be described.
This will be described in detail with reference to FIG.

The DB registration / search program 412 uses the DB 413 in common between hosts. DB 413 is
It is built on a large-capacity external storage device such as a magnetic disk.

When the number of host computers is one, the number of accesses to the actual DB 413 is reduced by using the memory as a buffer at the time of accessing the DB.
Although access can be speeded up, when the DB 413 is shared by a plurality of host computers, even if a memory unique to each host is used as a buffer, data consistency between the hosts can be maintained. Since the buffer needs to be purged in the commit process, the DB access performance decreases. Therefore, the shared buffer 404 is set on the extended memory 100 which can be accessed at high speed after the memory and can be shared and used by a plurality of hosts.
By using it as a buffer for B access, buffer purging during commit processing becomes unnecessary, and D
B access performance can be speeded up.

However, in this case, since the shared buffer 404 is shared between the hosts, the shared buffer 404
When extending is required, it is necessary to synchronize the hosts, and further, it is necessary to maintain the association between the identifier for managing the buffer area and the buffer area. In order to associate the identifier with the buffer area, a hash is generally used in order to perform a search at high speed.

If the hash width (the number of hash entries) is changed in order to dynamically expand the shared buffer 404, the correspondence between the already secured buffer and the identifier changes. In order to maintain the data, the data in the buffer must be transported. Since this data transfer needs to be performed for data areas belonging to all data blocks, enormous CPU resources are consumed.

Therefore, according to the present invention, when the shared buffer 404 is dynamically expanded, the hash table 106 having entries for the specifications of the shared buffer is created in advance so that the hash width does not need to be changed. During dynamic expansion, the hash table updating means 108 newly adds one of the data block addresses stored in the entry of the hash table 106 that indicates the data block of the same shared buffer 404. By changing the shared buffer 410 to point to the data block, the data area that needs to be transferred is limited to the data area in a specific data block, and the cost for the transfer can be significantly reduced.

The process of adding the shared buffer 410 to the shared buffer 404 will be described below.

For the purpose of improving the performance of DB access, etc.
If the capacity of the shared buffer needs to be increased,
In one of the host computers, the capacity of the shared buffer to be extended is designated, and the shared buffer adding means 411 is activated.

The shared buffer adding unit 411, in which the capacity of the shared buffer to be extended is specified, allocates the shared buffer (additional portion) 410 on the extended memory 100 by using the access unit 105. Next, the hash table updating unit 108 is called.

The hash table updating means 108 called by the shared buffer adding means 411 acquires the lock of the hash table 106 from the inter-host locking means 103 via the shared memory file access means 107. This lock has the effect of locking the hash tables 106 on all hosts. Next, the hash table updating unit 108 sets a shared buffer (additional buffer).
One of the data block addresses of a plurality of entries of the hash table 106 that points to the same data block is assigned to all the data blocks on the 410. As a result, the subsequent securing of a new data area is performed for the shared buffer (additional) 410 (the securing process of the new data area until the inter-host lock on the hash table 106 is released). Does not work). Next, the identifier of the data on the data block to which the changed data block address points before the change is hashed again, and the data indicated by the data block address stored in the entry of the hash table 106 corresponding to the obtained value The data is stored in the block.

Of course, if the data block obtained by hashing again is the same as the current data block,
No data transfer occurs. That is, the reason why the transfer occurs is that the result of the rehash is the shared buffer (additional) 41.
Only when pointing to a data block on 0. Completion of this transfer processing allows access to previously stored data (access to stored data does not operate until the inter-host lock on the hash table 106 is released).

Next, the hash table updating means 108
Writes the updated information of the hash table 106 into the control file 109 using the access unit 105, and unlocks the inter-host lock on the hash table 106 using the shared memory file access unit 107.
Request for

Shared memory file access means 107
Requests the inter-host locking means 103 to unlock the hash table 106.

The inter-host lock means 103 is provided with the hash table 1 from the shared memory file access means 107.
When receiving the unlock request for 06, the lock is released and all other hosts are notified of the release of the lock. Each host receiving the notification,
The contents of the control file 109 are checked, and when it is determined that the shared buffer has been dynamically expanded, the hash table 106 on the memory of the host is updated to the state after expansion. By the above operation, the shared buffer 4
04 and the shared buffer (additional) 410 become available.

Thus, the operation of the first embodiment is completed.

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

Referring to FIG. 1, the second embodiment of the present invention differs from the first embodiment in that the host 1 (10
1) and 2 (102) and a recording medium 200 such as a magnetic disk or a semiconductor memory for recording a program for causing the inter-host locking means 103 to execute the processing according to the first embodiment.

The program is read from the recording medium 200 by the hosts 1 (101) and 2 (102) and the inter-host locking means 103, and is read by the hosts 1 (101) and 2 (102).
(102) and the operation of the inter-host lock means 103 are controlled. Under the control of the program, the hosts 1 (101) and 2 (102) respectively access the access unit 105, the shared memory file access unit 107, the hash table update unit 108, and the shared file addition unit 111 in the first embodiment. Perform the same processing.

Thus, the processing of the second embodiment is completed.

[0061]

As described above, the present invention provides a hash table for specifying a data block of a shared memory file on the assumption that a shared memory file is added when the host is started up. When a large shared memory file is added, the added data block and the hash table can be associated with each other without changing the hash width (the number of hash entries). This has the effect that a plurality of shared memory files can be efficiently shared between them. Further, since the shared memory file shared by a plurality of hosts and the additional shared memory file are divided into a plurality of data blocks of the same size and managed, fragmentation occurs in the storage area including the shared memory file. It has the effect of being able to be used efficiently without using.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing a relationship between a hash table 106, a shared memory file 104, and a shared memory file (additional portion) 110 according to the embodiment of the present invention.

FIG. 3 is a flowchart showing the operation of the embodiment of the present invention.

FIG. 4 is a block diagram showing one embodiment of the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 shared memory 101-102 host computer 1, 2 103 inter-host locking means 104 shared memory file 105 access means 106 hash table 107 shared memory file access means 108 hash table updating means 109 control file 110 shared memory file (additional) 111 shared File addition means 201 to 208 Data block addresses 1 to 8 211 to 216 Data blocks 1 to 8 404 Shared buffer 410 Shared buffer (additional) 411 Shared buffer addition means 412 DB registration / search program 413 DB 414 DB access means

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06F 15/177 682 G06F 15/177 672 G06F 12/00 535

Claims (9)

(57) [Claims] An external storage device including a shared memory file composed of a plurality of data blocks of the same size, a hash table composed of a plurality of entries corresponding to any of the plurality of data blocks, When storing data in each of the plurality of data blocks, hashing the identifier of the data to determine a corresponding entry in the hash table, and storing the data in the data block corresponding to the entry. An inter-host shared memory file dynamic addition system comprising: a host computer; and an inter-host lock unit that exclusively controls access to the hash table of the plurality of host computers. , Each data block of the shared memory file and each entry of the hash table. Each of the plurality of host computers, when the own host computer is started earlier than the other host computers, the host computer locks the other host computer. Exclusive control of access to the hash table of the other host computer by the host computer, obtain the number of all data blocks of the shared memory file, and obtain at least the number of entries equal to or more than the number of data blocks. A table is created, correspondence information between each entry of the hash table and each data block of the shared memory file is stored in the control file, and the other host computer has another host computer by the inter-host locking means. Release exclusive control of access to the hash table and When the own host computer is started later than the other host computers, the inter-host locking means exclusively controls access to the hash table of the other host computer by the other host computer, and the control file And reads the correspondence information from the host computer, creates the hash table in the host computer based on the correspondence information, and uses the inter-host locking means to access the hash table of the other host computer by another host computer. A shared memory file dynamic addition system between hosts, comprising a shared memory file access means for releasing exclusive control. 2. Each of the plurality of host computers,
A shared file adding unit for adding the shared memory file on the external storage device; and a host file locking unit for adding the shared memory file to the external storage device after the shared memory file is added to the external storage device. Exclusive control of access to the hash table of the other host computer by the host computer, and any one of a plurality of entries of the hash table corresponding to a specific data block of the existing shared memory file is added. If the corresponding entry in the hash table corresponds to the result of hashing the identifier of the data of the specific data block, it is added to any one of the plurality of data blocks of the shared memory file. The associated data file in the shared memory file Transfer the data of the specific data block to the host, update the correspondence information stored in the control file, and use the inter-host locking means to store the hash table of the other host computer by another host computer. 2. A system according to claim 1, further comprising: a hash table updating unit for canceling exclusive control of access to the host. 3. The hash table updating means, further comprising: when another host computer updates the correspondence information stored in the control file, based on the updated correspondence information. 3. The system according to claim 2, wherein the hash table of the system is updated. 4. When each of a plurality of host computers stores data in each of a plurality of data blocks of the same size constituting a shared memory file shared by the plurality of host computers, an identifier of the data is used. A data storing step of performing hashing to determine a corresponding entry in a hash table composed of a plurality of entries corresponding to any of the plurality of data blocks, and storing the data in a data block corresponding to the entry; Each of the plurality of host computers, when the host computer is started before the other host computer, exclusively controls access to the hash table of the other host computer by the other host computer. Obtaining the number of all data blocks of the shared memory file,
The hash table including at least the number of entries of the data block is created, and correspondence information between each entry of the hash table and each data block of the shared memory file is transferred to a control file shared by the plurality of host computers. Storing and releasing exclusive control of access to the hash table of the other host computer by the other host computer, and when the own host computer is started after the other host computer, the other host computer Exclusively controlling access to the hash table of the other host computer by a computer, reading the correspondence information from the control file,
A hash table creating step of creating the hash table in the host computer based on the correspondence information, and releasing exclusive control of access to the hash table of the other host computer by another host computer. A method for dynamically adding a shared memory file between hosts. 5. Each of the plurality of host computers,
A shared file adding step of adding the shared memory file, wherein each of the plurality of host computers exclusively controls access to the hash table of the other host computer by another host computer, and the existing shared memory Associating any one of the plurality of entries of the hash table corresponding to a specific data block of a file with any one of the plurality of data blocks of the shared memory file added in the shared file adding step, and If the corresponding entry corresponds to the result of hashing the data identifier of the specific data block, the data of the specific data block is added to the associated data block in the added shared memory file. And store it in the control file A hash table updating step of updating the corresponding information that has been performed, and releasing exclusive control of access to the hash table of the other host computer by another host computer.
A method for dynamically adding a shared memory file between hosts as described in the above. 6. The hash table updating step,
Further, when the correspondence information stored in the control file is updated by another host computer, the hash table of the host computer system is updated based on the updated correspondence information. 6. The method for dynamically adding a shared memory file between hosts according to claim 5. 7. When storing data in each of a plurality of data blocks of the same size constituting a shared memory file shared by the plurality of host computers, an identifier of the data is stored in each of the plurality of host computers. Hashing to determine a corresponding entry in a hash table composed of a plurality of entries corresponding to any of the plurality of data blocks, and storing the data in a data block corresponding to the entry; When the own host computer is started before the other host computer, exclusive access control to the hash table of the other host computer by the other host computer is performed, and all data in the shared memory file are controlled. Before obtaining the number of blocks and before being composed of at least the number of entries of the data blocks or more Create a hash table,
Correspondence information between each entry of the hash table and each data block of the shared memory file is stored in a control file shared by the plurality of host computers, and stored in another host computer by the hash table of the other host computer. When the exclusive control of the access of the host computer is released and the own host computer is started after the other host computer, the exclusive control of the access to the hash table of the other host computer by the other host computer is performed. Reading the correspondence information from the control file,
A hash table creation process for creating the hash table in the host computer based on the correspondence information and canceling exclusive control of access to the hash table of the other host computer by the other host computer is performed. At least one recording medium on which a program is recorded. 8. A shared file addition process for adding the shared memory file to each of the plurality of host computers, and exclusively controlling access to the hash table of the other host computer by another host computer, Associating any one of the plurality of entries of the hash table corresponding to a specific data block of the existing shared memory file with any of the plurality of data blocks of the shared memory file added in the shared file adding step If the associated entry in the hash table corresponds to the result of hashing the identifier of the data of the specific data block, the specified entry is added to the associated data block in the added shared memory file. The data of the data block of the A program that updates the correspondence information stored in the hash table and performs a hash table update process for releasing exclusive control of access to the hash table of the other host computer by the other host computer. The recording medium according to claim 7, wherein: 9. The method according to claim 1, wherein the hash table updating process further includes updating the corresponding information stored in the control file by another host computer based on the updated corresponding information. 9. The recording medium according to claim 8, wherein the hash table of the system is updated.