JP2004054906A - Memory access method and computer system for executing the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data sharing method which easily accesses data shared by systems by means of the same procedure as adopted in accessing a region in one's own system without making a program, which operates on each computer system, be aware of the presence of the sharing device. <P>SOLUTION: The inside of a global memory 212 is divided into fixed units, and each of the divided units is managed with a unique identifier. When an inter-system excluding means 202 designates the region of an optional address space in a processing device 100 with the identifier to secure exclusion, an updating status confirming means 203 checks the existence of the updating of data within the corresponding unit in a sharing device 110. When the data is updated, a data forwarding means 204 forwards the data in the corresponding unit in the sharing device 110 to the address space. When the exclusion is eliminated, a data sharing method forwarding means 204 forwards the data to inside of the corresponding unit in the sharing device 110, when the data in the address space has been updated. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a data sharing system, and more particularly to a data sharing system in which data shared by a plurality of computer systems is shared by a sharing device.
[0002]
[Prior art]
Conventionally, when sharing data in an address space among a plurality of computer systems, the shared input / output device is occupied, the data is written once at a predetermined position, and then the occupation state is released. After that, when another system uses the data, similarly occupy the I / O device, read the data at the determined position, and transfer the data to the area secured in advance in the address space. , The occupation state is released. That is, each computer system temporarily occupies the input / output device to read or write data, and then releases the occupation. Here, examples of the input / output device shared by a plurality of systems include a magnetic disk device, a semiconductor disk device, and an independent extended storage device.
[0003]
Regarding a method of sharing data between a plurality of systems using an independent extended storage device, see, for example, a manual “Program Product VOS3 / FS Center Operation-JSS3 Edition” issued by Hitachi, Ltd., published in October 1995. p. 110-117, "Extended storage device".
[0004]
[Problems to be solved by the invention]
However, in the conventional system, since the I / O device that is a shared device is occupied, the exclusive overhead of the I / O device is large, and if relatively small data is frequently accessed, the exclusive overhead and the access overhead cannot be ignored. was there.
An object of the present invention is to allow a program running on each computer system to easily access shared data between systems by the same procedure as accessing an area in the own system without being aware of the existence of a shared device. It is to provide a data sharing method.
[0005]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides a data sharing method in which data commonly used among a plurality of computer systems is shared in a shared memory in a shared device, wherein the shared memory is divided into fixed units, Each of the divided units is managed with a unique identifier when viewed from the plurality of computer systems and the shared device, and an area of an arbitrary address space in the computer system is designated by the identifier and exclusively secured. Sometimes, the presence or absence of update of the data in the corresponding unit in the shared memory is checked, and when the data is updated, the data in the corresponding unit in the shared memory is transferred to the address space, and when the exclusion is released. When data in the address space is updated, the data is stored in a corresponding unit in the shared memory. With this method, programs running on each computer system can easily perform the same procedure as accessing an area in their own system without being aware of the existence of shared devices. Can access shared data between systems.
In the data sharing method, preferably, the computer system accesses the plurality of identifiers in a group, and secures exclusion of the shared device by using a representative identifier among the grouped identifiers. With this method, exclusive overhead can be reduced.
In the data sharing method, preferably, the computer system makes an access time for referring to and updating data on the computer system and an access time for reflecting data on the shared device to the computer system asynchronous. With this method, the overhead to the shared device can be reduced.
[0006]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, a data sharing method according to an embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a configuration diagram of a compound computer system that realizes a data sharing method according to an embodiment of the present invention.
[0007]
The processing apparatus 100 includes a local area allocating unit 201 that manages a memory area on an address space of a local memory 206 in its own system, which corresponds to the global memory 212 managed by an identifier on the shared apparatus 110 on a 1: 1 basis. I have. The local area allocating unit 201 manages a memory area in a table format including items of “identifier”, “physical memory address”, and “length”.
[0008]
In the local area allocating means 201, for example, an area of a length "1000" X from the address "10002000" X to the address "10002FFF" X of the local memory 206 is managed by "identifier" = "000001". It is assumed that an area of a length “1000” X from the address “0F001000” X to the address “0F001FFF” X of the memory 206 is managed by “identifier” = “000002”.
The memory area managed by “identifier” = “000001” and “identifier” = “000002” is the area illustrated in the local memory 206.
[0009]
[0010]
Further, the processing device 100 checks the inter-system exclusion means 202 for securing and releasing exclusion from other systems when transferring data to the shared device 110, and checks whether data on the shared device 110 has been updated by another system. The update state confirmation unit 203 transfers data from the local memory 206 on the processing device 100 to the global memory 212 on the shared device 110 or from the global memory 212 on the shared device 110 to the local memory 206 on the processing device 100. The system includes a data transfer unit 204, an update vector setting unit 205 for setting an update vector for recording an update state of another system on the shared device 110, and a local memory 206 that can be referred to in an address space on the own system.
[0011]
The processing device 120 has a configuration similar to that of the processing device 100, and includes a local area allocating unit 221, an inter-system exclusion unit 222, an update state confirmation unit 223, a data transfer unit 224, and an update vector setting unit 225. , A local memory 226. Each of these units has the same function as the corresponding unit in the processing apparatus 100, and a description thereof will be omitted.
In the local area allocating means 221, for example, an area having a length "1000" X from the address "0E001000" X to the address "0E001FFF" X of the local memory 226 is managed by "identifier" = "000001". Further, it is recorded that the area of the length “FE000” X from the address “28100000” X to the address “281FFFFF” X of the local memory 226 is managed by “identifier” = “000003”.
The memory area managed by “identifier” = “000001” and “identifier” = “000003” is the area illustrated in the local memory 226.
The shared device 110 is connected to the processing device (SYS A) 100 and the processing device (SYS B) 120, holds a memory therein, and stores the memory therein in accordance with an instruction from the connected processing devices 100 and 120. It has a function to operate the data. The shared device 110 includes, for example, a magnetic disk device, a semiconductor disk device, and an independent extended storage device.
[0012]
The shared device 110 includes, as means for dividing and managing the memory in the shared device, an identifier allocating unit 210 for recording information as to which system each identifier is allocated to, and a memory in the address space of the global memory 212. A shared memory management unit 211 for managing an area by an identifier is provided, and a global memory 212 is provided as a place where actual data is arranged.
The identifier assigning unit 210 manages the assignment of identifiers in a table format including items of “identifier” and “assigned system name”. In the identifier allocating means 210, for example, “identifier” = “000001” is allocated to “SYS A” which is the processing device 100 and “SYS B” which is the processing device 120, and “identifier” = ” “000002” is assigned to “SYS A”, and it is recorded that “identifier” = “000003” is assigned to “SYS B”.
Further, the shared memory management unit 211 manages memory allocation to identifiers in a table format including items of “identifier”, “physical memory address”, and “update vector”. In the shared memory management unit 211, for example, the area from the address “00000000” X to the address “00000FFF” X of the global memory 212 is managed by “identifier” = “000001”, and the “update vector” is “100”. Is recorded. Although the details of the “update vector” will be described later, the “update vector” having a 3-bit configuration is set to flag corresponding to three columns in the “assigned system name” of the identifier assigning means 210. . That is, of the “update vector” = “100” for “identifier” = “000001”, “1” corresponds to “SYS A”, “0” corresponds to “SYS B”, and “0” corresponds to “ -(No allocation system) ". Since the bit is set for “SYS A” (“1”), “identifier” = “000001”
This indicates that the data in the managed memory area has been updated by the processing device 100 (SYS A).
Similarly, the shared memory management unit 211 manages the area from the address “00001000” X to the address “00001FFF” X of the global memory 212 with “identifier” = “000002”, and stores the “update vector”. “000”, which has not been updated by any of the processing devices, and manages an area from the address “00002000” X to the address “000FFFFF” X of the global memory 212 with “identifier” = “000003”. The update vector is “010” and has been updated by the processing device 120 (SYS B), and the area from the address “00100000” X to the address “001FFFFF” X of the global memory 212 is “identifier” = “000004”. And the "update vector" is "000". It is recorded that the data has not been updated even from the processing device of the shift.
The memory area managed by “identifier” = “000001”,..., “Identifier” = “000004” is the area illustrated in the global memory 212.
Next, a procedure for referring to a common memory area between systems defined between a plurality of systems at the time of a lock request will be described with reference to FIG.
FIG. 2 is a flowchart showing a procedure of lock request processing in the data sharing method according to one embodiment of the present invention.
[0013]
Hereinafter, the lock request processing will be described with reference to FIG. 1 in accordance with each step shown in FIG. 2. Here, nothing is recorded in the column of “allocation system” of the identifier allocating means 210 in FIG. It is assumed that the “update vector” column of the shared memory management unit 211 is “000”. In step 1 of FIG. 2, the program on the processing device 100 allocates an area from “address” = “10002000” X to “length” = “1000” X of the local memory 206 in the previously reserved address space. In order to access as a common memory between systems, a lock request is issued by a program on the processing device 120 and a predetermined “identifier” = “000001”.
In step 2, the inter-system exclusion unit 202 of the processing device 100 secures exclusion using the specified identifier. That is, the inter-system exclusion unit 202 sets “identifier” = “000001” to the exclusive state.
Next, in step 3, the update state confirmation unit 203 of the processing device 100 determines whether the specified identifier has been updated in another system. That is, the update state check unit 203 checks whether the global memory 212 on the shared device 110 associated with the specified identifier is being accessed from another system. If it has been updated, the process proceeds to step 4, and if it has not been updated, the process proceeds to step 13. Steps 4 and 5 will be described later.
Here, it is assumed that the access to “identifier” = “000001” from the processing device 100 is the first access, and the process proceeds to step 13.
In step 13, the identifier allocating unit 210 of the shared device 110 records that an allocation request has been received from the processing device 100 (SYS A) for “identifier” = “000001”. The global memory 212 on the shared device 110 corresponding to the identifier = “000001” is secured for a length = “1000” ×.
Next, in step 6, the processing device 100 does not transfer data because the access is the first access in the lock request processing on the processing device 100 and there is no update by the other system on the "update vector". Complete the lock request.
[0014]
Through the processing of the lock request described above, the program on the processing device 100 can access the area associated with the identifier. The program can easily refer to and update this area in the same procedure as accessing the area in its own system without using special instructions.
According to the present embodiment, it is possible to easily refer to and update the same procedure as accessing an area in the own system without using a special instruction.
Next, a procedure for making an unlock request for a common memory area between systems defined between a plurality of systems will be described with reference to FIG.
FIG. 3 is a flowchart showing a procedure of an unlock request process in the data sharing method according to one embodiment of the present invention.
In step 7 of FIG. 3, when the program on the processing device 100 completes the access to the common area, the program issues an unlock request with “identifier” = “000001” as an input.
[0015]
In step 8, the processing device 100 determines whether an update instruction has been issued. If there is an update instruction, the process proceeds to step 9; otherwise, the process jumps to step 11. When the program on the processing device 100 only refers to this memory area, no update instruction is issued.
Here, if the access to “identifier” = “000001” from the program on the processing device 100 is the first time, the data is not created on the global memory 212, and the processing device 100 issues an update instruction. Do.
Next, in step 9, data is transferred from the local memory corresponding to the identifier to the global memory. That is, the data transfer unit 204 transfers the contents of “address” = “10002000” X to “address” = “10002FFF” X of the local memory 206 corresponding to the identifier = “000001” to “address” = Transfer from “00000000” X to “address” = “00001000” X.
[0016]
Next, in step 10, an update vector is set. That is, the update vector setting unit 205 changes the bit of the update vector managed by the shared memory management unit 211 from “0” B to “1” B so that the other system can recognize that the update from the own system has been completed. Set. Here, the “update vector” corresponding to “identifier” = “000001” of the shared memory management unit 211 is set from “000” to “100”. Since “SYS A”, which is the processing device 100, is assigned first at the beginning of the record in the “assigned system name” column of the identifier assigning means 210, the most significant bit of the update vector is changed from “0” B. Set to "1" B.
Next, in step 11, the inter-system release unit 202 performs exclusive release using the specified identifier. That is, “identifier” = “000001” is exclusively released.
In step 12, the program on the processing device 100 completes the unlock request.
If there is no update instruction in step 8, the data in the local memory 206 is not transferred to the global memory 212, and in step 11, the inter-system exclusion means 202 sets the Is executed, and in step 12, the program ends the unlock request processing.
Next, the lock request processing when another system, for example, the processing device 120 refers to and updates the shared memory allocated by the processing device 100 as described above will be described with reference to FIG.
In step 1 of FIG. 2, the program on the processing device 120 allocates an area from “address” = “0E001000” X to “length” = “1000” X in the local memory 226 in the previously reserved address space. In order to access as a common memory between systems, a lock request is issued by a program on the processing device 100 and a predetermined “identifier” = “000001”.
In step 2, the inter-system exclusion unit 222 of the processing device 120 secures exclusion using the specified identifier. That is, the inter-system exclusion unit 222 sets the “identifier” = “000001” to the exclusive state.
Next, in step 3, the update state confirmation unit 223 of the processing device 120 determines whether the specified identifier has been updated in another system. Here, the update state confirmation unit 223 checks the contents of the update vector managed by the shared memory management unit 211, and the “identifier” = “000001” has already been accessed from the processing device 100. Proceed to 4.
In step 4, the data transfer unit 224 transfers data from the global memory 212 corresponding to the identifier to the local memory 226. That is, the data transfer unit 224 transfers the contents of “address” = “00000000” X to “00000FFF” X of the global memory 212 corresponding to “identifier” = “000001” to the “address” of the local memory 226 on the processing device 120. "=" 0E001000 "X to" 0E001FFF "X. By this data transfer process, the “identifier” = “000001” on the processing device 100 matches the data on the “identifier” = “000001” on the processing device 120.
[0017]
In step 5, the update vector setting means 225 of the processing device 120 resets the update vector. That is, the update vector setting unit 225 resets the “update vector” from “1” B to “0” B so that the data is not read again after the data is transferred on the processing device 120. As a result, the “update vector” for “identifier” = “000001” is reset from “100” to “000”.
In step 6, the lock request is completed, and the process returns to the request program source. Then, the program on the processing device 120, for which the lock request has been completed, refers to and updates the data on the local memory 226 associated with “identifier” = “000001”.
Next, the unlock request process shown in FIG. 3 is executed.
[0018]
In step 7, an unlock request is issued, and in step 8, it is determined whether there is an update instruction. If the reference / update execution is only reference, the exclusive release is performed in step 11 without performing the update instruction, and the unlock process is completed in step 12.
[0019]
If it is determined in step 8 that there is an update instruction, data is transferred from the local memory to the global memory in step 9, the update vector is reset in step 10, and exclusive release is performed in step 11. In step 12, the unlock process is completed.
[0020]
By repeating the lock → reference / update → unlock according to the processing procedure described above, an area common to the systems can be accessed. The procedure of lock → reference / update → unlock is the same as the procedure for referring / updating a common area from all spaces in a system having a plurality of virtual address spaces. The common memory between the systems can be accessed with the same ease as the conventional common area.
[0021]
As described above, according to the present embodiment, the program can easily refer to and update the area of the common memory in the same procedure as accessing the area in the own system without using a special instruction. become able to.
Next, with reference to FIG. 4, a description will be given of a method in which a plurality of identifiers, which are the minimum units of exclusion, are grouped together for access.
FIG. 4 is a configuration diagram of a compound computer system that realizes a data sharing method according to another embodiment of the present invention. Note that the same reference numerals as those in FIG. 1 indicate the same parts, and have the same configuration and function, and thus detailed description will be omitted.
In the present embodiment, “identifiers” = “000001” and “000002” are collectively accessed as one group, and “identifiers” = “000003” and “000004” are accessed collectively as one group. ing. The processing devices 100 and 120 recognize that “identifiers” = “000001” and “000002”, respectively, as being grouped as one group. When accessing the group of “identifier” = “000001” and “000002”, “000001” is used as the representative “identifier”, and when accessing the group of “identifier” = “000003” and “00004”, "000003" is used as the representative "identifier".
[0022]
Here, in the local area allocating means 201, for example, the area of the length "1000" X from the address "10002000" X to the address "10002FFF" X of the local memory 206 is managed by "identifier" = "000001". Then, an area of a length “1000” X from the address “10003000” X to the address “10003FFF” X of the local memory 206 is managed by “identifier” = “000002”, and the address from the address “10004000” X of the local memory 206 is An area having a length of “1000” X up to “10004FFF” X is managed by “identifier” = “000003”, and an area of a length “1000” X from the address “10005000” X to the address “10005FFF” X in the local memory 206 is managed. Specify the area as “identifier” = “000004” In assumed to manage. The memory area managed by “identifier” = “000001”,..., “000004” is the area illustrated in the local memory 206.
In the local area allocating means 221, for example, an area having a length "1000" X from the address "0E003000" X to the address "0E003FFF" X of the local memory 226 is managed by "identifier" = "000001". The area of the length “1000” X from the address “0E004000” X to the address “0E004FFF” X of the local memory 226 is managed by “identifier” = “000002”, and the address “0E001000” X to the address ”of the local memory 226. An area having a length of “1000” X up to 0E001FFF ”X is managed by“ identifier ”=“ 000003 ”, and an area of a length“ 1000 ”X from address“ 0E002000 ”X to address“ 0E002FFF ”X in the local memory 226. Is "identifier" = "000004" It is assumed that the management. The memory area managed by “identifier” = “000001”,..., “000004” is the area illustrated in the local memory 206.
[0023]
Further, in the identifier allocating means 210, for example, “identifier” = “000001” is allocated to “SYS A” which is the processing device 100 and “SYS B” which is the processing device 120. = “000002” is assigned to “SYS A” and “SYS B”, “identifier” = “000003” is assigned to “SYS B” and “SYS A”, and “identifier” = ” "000004" is recorded as being assigned to "SYS B" and "SYS A".
In the shared memory management unit 211, for example, the area from the address “00000000” X to the address “00000FFF” X of the global memory 212 is managed by “identifier” = “000001”, and the “update vector” is “ 000 "is recorded, an area from the address" 00006500 "X to the address" 000074FF "X of the global memory 212 is managed by" identifier "=" 000002 ", and the" update vector "is" 010 ". The area from the address “00004000” X to the address “00004FFF” X of the global memory 212 is managed by “identifier” = “000003”, and it is recorded that the “update vector” is “000”. , Address “00008000” of the global memory 212 An area of from X to the address "00008FFF" X managed by the "identifier" = "000001", it is "update vector" is "100" is recorded.
The process in the case of requesting a lock by grouping a plurality of identifiers will be described again with reference to FIG.
In step 1 of FIG. 2, the program on the processing device 100 includes an area from “address” = “1004000” X to “length” = “1000” X in the local memory 206 in the previously secured address space. In order to access an area from “address” = “10005000” X to “length” = “1000” X in the local memory 206 as a common memory between systems, it is determined in advance as a program on the processing device 120. A lock request is issued by “identifier” = “000003”. Here, the area managed by “identifier” = “000003” and “identifier” = “000004” is locked. Here, a lock request is made with a representative identifier.
In step 2, the inter-system exclusion unit 202 of the processing device 100 secures exclusion using the specified identifier. That is, the inter-system exclusion unit 202 sets “identifier” = “000003” to the exclusive state. As a result, “identifier” = “000004” is not in an exclusive state, but both of the processing devices 100 and 120 change “identifier” = “000003” and “identifier” = “000004”. Since it is agreed that grouping is handled, when “identifier” = “000003” is in the exclusive state, no exclusion request is made for “identifier” = “000004”, and “identifier” = “000004” , The exclusion is substantially secured.
Next, in step 3, the update state confirmation unit 203 of the processing device 100 determines whether the specified identifier has been updated in another system. That is, the update state check unit 203 checks whether the global memory 212 on the shared device 110 associated with the specified identifier is being accessed from another system.
Here, even when a lock request is received from the processing device 100 with the identifier = “000003”, the update state confirmation unit 203 does not respond to the representative identifier but to both “identifier” = “000003” and “000004”. Check and confirm the update status. In the state shown in FIG. 4, the update vector for the identifier = “000003” in the shared memory management unit 211 is “000” and is not updated, but the update vector for the identifier = “000004” is “100”. Since it is known from the processing device 120 (SYS B) that the information has been updated, the process proceeds to step 4.
In step 4, the data transfer unit 204 transfers data from the global memory 212 to the local memory 206.
Next, in step 5, the update vector setting means 205 resets the update vector for the identifier = "000004" in the shared memory management means 211.
Next, in step 6, the processing device 100 completes the lock request, and refers to and updates the read data.
[0024]
Through the processing of the lock request described above, the program on the processing device 100 can access a plurality of areas associated with the identifier. Even if a plurality of units (areas) to be referred to / updated are provided, only one exclusive overhead is required, so that the exclusive overhead can be reduced.
[0025]
The program can easily refer to and update this area in the same procedure as accessing the area in its own system without using special instructions.
The process for the unlock request is as described above with reference to FIG. 3. Since the exclusive exclusion is performed for the representative identifier, the exclusion release in step 11 is also performed only for the exclusively secured identifier. Be released.
According to the present embodiment, it is possible to easily refer to and update the same procedure as accessing an area in the own system without using a special instruction.
In addition, it is possible to reduce exclusion overhead when exclusion is secured for a plurality of areas.
[0026]
Next, a method for making access to the shared device and access to the local memory asynchronous when a program on the processing device references / updates a common memory between systems will be described with reference to FIG.
FIG. 5 is a configuration diagram of a compound computer system that realizes a data sharing method according to another embodiment of the present invention. Note that the same reference numerals as those in FIG. 1 indicate the same parts, and have the same configuration and function, and thus detailed description will be omitted.
[0027]
On the processing device 100, the program A 207 operates as a program that operates while referring to and updating the local memory 206 associated with a common memory between systems.
Here, the area allocation of the local memory 206 managed by the local area allocation means in the processing device 100 is the same as that shown in FIG.
[0028]
Also, on the processing device 120, the program B227 operates as a program that operates while referring to and updating the local memory 226 associated with the common memory between the systems.
The allocation of the area in the local memory 226 managed by the local area allocating means in the processing device 120 is the same as that shown in FIG.
[0029]
Further, the name of the assignment system in the identifier assignment means 210 is the same as that shown in FIG.
The management status of the global memory 212 as the shared memory in the shared memory management unit 211 is the same as that shown in FIG.
[0030]
The data shared by the program A 207 and the program B 227 is not data whose integrity must be guaranteed between all systems like user data. For example, the busy rate of the CPUs on the processing devices 100 and 120 and the shared resources This is data that allows some error, such as operation information.
[0031]
When such data is shared, since data exchange occurs periodically, it is desirable to reduce the access overhead or exclusive overhead to the shared device 110 as much as possible. Therefore, the lock / unlock request to the shared device 110 does not operate as an extension of the operation of each program, but performs data access to the shared device 110 periodically at a cycle with relatively little overhead by a completely different program, The data is reflected on the local memories 206 and 226.
[0032]
The program A 207 and the program B 227 operate at a different timing from the reflection of data on the local memories 206 and 226, and refer to and update only the local memories 206 and 226 when necessary.
[0033]
By separating the access to the local memory and the access to the global memory in this manner, it is possible to reduce the overhead to the shared device.
According to the present embodiment, it is possible to easily refer to and update the same procedure as accessing an area in the own system without using a special instruction.
In addition, by dividing the opportunity for referring to and updating data on each computer system and the opportunity for reflecting data on the shared device to the computer system, overhead to the shared device can be reduced.
[0034]
【The invention's effect】
According to the present invention, when a memory is shared between a plurality of computer systems, reference and update can be performed by the same procedure as that for accessing a common area in a conventional system.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of a compound computer system that realizes a data sharing method according to an embodiment of the present invention.
FIG. 2 is a flowchart showing a procedure of a lock request process in the data sharing method according to one embodiment of the present invention.
FIG. 3 is a flowchart showing a procedure of an unlock request process in the data sharing system according to one embodiment of the present invention.
FIG. 4 is a configuration diagram of a compound computer system that realizes a data sharing method according to another embodiment of the present invention.
FIG. 5 is a configuration diagram of a compound computer system that realizes a data sharing method according to another embodiment of the present invention.
[Explanation of symbols]
100, 120 ... processing unit
110 ... shared device
201, 221... Local area allocating means
202, 222 ... inter-system exclusion means
203, 223... Update status confirmation means
204, 224: Data transfer means
205, 225... Update vector setting means
206, 226: Local memory
207 ... Program A
227: Program B
210: identifier assigning means
211: Shared memory management means
212: Global memory

Claims (3)

In a data sharing method in which data commonly used among a plurality of computer systems is shared in a shared memory in a shared device,
Dividing the shared memory into fixed units, managing each of the divided units with a unique identifier as viewed from the plurality of computer systems and the shared device,
When an area in an arbitrary address space in the computer system is designated by the identifier and exclusively secured, the presence or absence of update of data in a corresponding unit in the shared memory is checked. Transferring the data in the corresponding unit in the shared memory to the address space,
A data sharing method wherein, when exclusion is released, when data in the address space is updated, the data is transferred to a corresponding unit in the shared memory. The data sharing method according to claim 1,
The data sharing method, wherein the computer system accesses the plurality of identifiers in a group, and secures exclusion of the shared device by using a representative identifier among the grouped identifiers. The data sharing method according to claim 1,
A data sharing method, wherein the computer system makes an access time for referring to and updating data on the computer system and an access time for reflecting data on the shared device to the computer system asynchronous with each other.

Images