JP2003241901A - Method and device for disk sharing control - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce a necessary resource quantity in case of connection of a plurality of accessing devices by consolidating the same data without adding an exclusive control mechanism or the like to the accessing device. <P>SOLUTION: This device is provided with a table for making a physical address of a disk device correspondent with a logical address allocated for each accessing device for management. For a write request including the logical address from the accessing device, if the logical address of the write request and the logical address of the other accessing device are in a state of sharing the same physical address, the logical address of the write request is made correspondent with the physical address in a non-shared state and data writing to the physical address in the non-shared state is allowed. If it is in the non- shared state, the data writing to the physical address corresponding to the logical address of the write request is allowed. After the data writing ends, the data on the physical address corresponding to the logical address in the non-shared state are compared with each other and if the data are the same, the logical address is made correspondent with the same physical address. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk sharing control method and apparatus for sharing one disk device among a plurality of access source devices such as computers.

[0002]

2. Description of the Related Art Conventionally, since a hard disk device connected to an access source device such as a personal computer or a server computer is low in reliability, the hard disk devices are individually connected to a plurality of access source devices. Create a database copy between multiple servers,
The data consistency in the hard disk drive was maintained by software.

However, recent hard disk devices have been improved in reliability due to the development of reliability improving technology, and are now connected and shared by a plurality of access source devices. However, unless dedicated software is introduced, the shared disk is shared. Even in the case of the configuration, it is a configuration in which the disks are simply collected in one housing, and is logically the same as the existence of a plurality of hard disk devices.

[0004]

Therefore, there is a demand for a technique that allows one hard disk device to be shared by a plurality of access source devices without using dedicated software. However, when the data in the hard disk device is used repeatedly by a plurality of access source devices, the same data is written by a plurality of access source devices at the same time, which causes a resource conflict problem in which consistency cannot be obtained.

In order to avoid this problem, it is necessary for each access source device to execute the update processing of the same data under exclusive control. However, for that purpose, there is a problem that a mechanism for performing exclusive control must be added to the access source devices.

An object of the present invention is to add an exclusive control mechanism or the like to an access source device using a disk device,
It is an object of the present invention to provide a disk sharing control method and device that can reduce the required resource amount when a plurality of access source devices are connected by consolidating the same data.

[0007]

In order to achieve the above object, a disk sharing control method according to the present invention provides a physical address indicating a physical read / write position of a disk device and a logical read / write position assigned to each access source device. Is a disk sharing control method for sharing one disk device by a plurality of access source devices, and includes a logical address from any of the access source devices. For a write request,
A first step of searching the table, and a second step of determining whether or not the logical address in the write request and the logical address of another access source device are in the shared state of the same physical address based on the search result. In the shared state, the logical address of the write request is associated with the physical address in the non-shared state, the table is updated, and data writing to the associated physical address in the non-shared state is permitted. A third step of permitting data writing to a physical address corresponding to the logical address of the write request, and searching the table after the data writing is completed, in a non-shared state that is not associated with the same physical address. Detects the logical address, compares the data at the physical address corresponding to the logical address in the non-shared state, and determines whether they are the same. A fourth step of determining, when the same data is characterized by comprising a fifth step of updating the table in association with the logical address into the same physical address.

A disk sharing control device according to the present invention is a disk sharing control device for sharing one disk device by a plurality of access source devices, and has a physical address indicating a physical read / write position of the disk device. A table that manages a logical address indicating a logical read / write position assigned to each access source device in association with each other, and a first request to search the table for a write request including a logical address from any access source device. Means and second means for determining whether or not the logical address in the write request and the logical address of another access source device are in the shared state of the same physical address based on the search result; and in the shared state, the write request Of the physical address in the non-shared state by updating the table by associating the logical address of To allow the writing of data to the dress,
In the non-shared state, the third means for permitting data writing to the physical address corresponding to the logical address of the write request, and the table being searched after the data writing is completed, are associated with the same physical address. A fourth means for detecting a non-shared logical address, comparing data at physical addresses corresponding to the non-shared logical address, and determining whether or not the data is the same as the fourth means. And a fifth means for updating the table by associating the logical address with the same physical address. The physical address is assigned so that all read / write positions of the disk device can be designated, and the logical address is assigned so that read / write positions less than the physical addresses can be designated.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, one embodiment for carrying out the present invention will be described in detail with reference to the drawings. Figure 1
1 is a block configuration diagram of a disk device showing an embodiment of the present invention. The disk device 100 of this embodiment is shown in FIG.
As shown in FIG. 2, interfaces (1) 101, (2) 102 that accept requests (read requests, write requests) to a disk (storage area) for each access source device configured by a server computer and data in the device. Control central processing unit 103 and disk unit 100 for controlling
Table expansion area memory 104 for expanding the read / write position in the table and logically connecting the logical read / write position of each interface (each connection access source) to the physical read / write position, and mitigating the difference in speed of physical access. In order to improve the availability of the disk, a disk cache memory 105, an array controller 106 that controls the RAID configuration, a hard disk group 107, and an internal bus 108 through which data and control signals flow. In the block diagram, the components other than the table development area memory 104 are the mechanisms that are usually mounted in the existing large-sized disk device. Therefore, the present invention realizes a new function in the disk device 100 by adding and using the table expansion area memory 104 on the device.

The control central processing unit 103 is requested via the table of the physical read / write position allocated to the disk unit 100 in the table expansion area memory 104 and the input / output interfaces (1) 101, (2) 102. It has a table that manages the logical read / write position for each interface, and controls the correlation between the physical read / write position and the logical read / write position. A personal computer or a server computer that is a data access source device is connected to the interfaces (1) 101 and (2) 102.

FIG. 2 shows a table expansion area memory 104.
It is a figure which shows the structure and concept of the address table 200 provided inside. In FIG. 2, the disk device 100
The physical read / write position of the block is the block number LB that is assigned considering the disk as a continuous block of a certain unit.
It is given as a physical address 203 by A (Logical Block Address). When there is an input / output request to the disk device 100, the disk device determines the physical read / write position from the LBA and executes input / output (data read / write). Therefore, the LBA corresponds to an address table that specifies a physical read / write position in the disk device 100. This physical address 203 is the disk device 100.
A value that can specify all read / write positions in is assigned. FIG. 2 shows an example in which all the read / write positions of the disk device are divided into four blocks for easy understanding of the description. In this case, the physical address 20
“3” is assigned “1” to “4”. On the other hand, to the interfaces (1) 101 and (2) 102, logical addresses 201 and 202 indicating logical read / write positions of the disk device 100 are assigned for each interface. Then, the logical address for each interface is set to the above L
A virtual disk 204 connected to the interface 101 and a virtual disk 205 connected to the interface 102 are configured by linking with the physical address 203 based on the BA.

The read / write position on the disk indicated by the logical addresses 201 and 202 is converted into a physical read / write position by the physical address 203, and data is read / written on the physical disk 206 indicated by the converted read / write position. Therefore, 204 and 205 are virtual disks having only read / write positions. Since the logical addresses of the virtual disks 204 and 205 store different data, it is necessary to associate the physical addresses separately. Therefore, it is allocated less than the physical address.
That is, the logical addresses 201 and 202 are assigned so that the read / write positions that are less than the read / write positions that can be specified by the physical addresses can be specified. In the example of FIG. 2, two logical addresses are assigned to each of the interfaces 101 and 102 for four physical addresses.

FIG. 3 is a process outline diagram showing an outline of the process in the disk device 100. Disk device 10
0 performs an input / output process of accepting an input / output request (data read / write request) through the interfaces (1) 101 and (2) 102 (step 301), and sends it to the disk duplication control process which constitutes the main part of the present invention ( Step 30
2). In the disk duplication control processing, the processing request received from each of the interfaces (1) 101 and (2) 102 is executed as input / output request processing that transfers the logical address so that conflict (contradiction) of the read / write position does not occur. Further, after the data writing is completed, a data duplication control process is performed to check whether the same data is associated with a different logical address for each interface.

Thereafter, through a disk cache process (step 403) for reducing the difference between the physical input / output time of the disk processing device and the processing time with the input / output request source, RAID (Redundant Array of Inexpensive Disk d) is set.
A physical read / write operation is executed by the disk array control processing (step 304) that constitutes rivers).

Next, the input / output request processing when an input / output processing request is received through the interface (1) 101 and the interface (2) 102 will be described with reference to the flowchart of FIG. Eventually, there are two input / output requests to the disk device 100: a read request and a write request for the hard disk group 107. First, it is determined whether the request is a read or a write (step 401). In the case of a read request, the address table 20 of FIG.
0 is searched for the physical address corresponding to the logical address, the data in the storage location indicated by the physical address is read, and returned to the request source (step 402). In the case of a write request, the address table 200 of FIG. 2 is searched to check whether the logical address of the write request is shared with the logical address of another interface (step 4).
03). The shared state is a state in which the logical address of each interface is associated with a common physical address on the address table 200, as shown in FIG. In FIG. 5A, the interface (1) 1
The logical addresses “1” and “2” of 01 and (2) 102 are associated with the common physical addresses “1” and “2”.

In such a shared physical address state, if data write requests from the respective interfaces occur simultaneously, a situation occurs in which a plurality of write requests compete for the same data. Therefore, in a shared state in which logical addresses are associated with a common physical address, a physical address whose logical address is not correlated with any of the interfaces is correlated with a logical address for which a write request has been issued ( Step 404). As a result, the contention state of write requests for the same data is released. In FIG. 5B described later, the interface (2)
When there is a write request from 102, an example in which the logical address “1” of the interface (2) is transferred to the physical address “3” is shown. After this, for the physical address correlated to the logical address for which a write request was made,
Data writing processing is performed (step 405).

FIG. 5 shows each interface (1) 101,
(2) When a write request including a logical address is input to 102, the logical addresses 201 and 20 in the address table 200 of FIG.
2 is a diagram showing a transition example of the contents of 2 and the physical address 203. FIG. FIG. 5A shows an initial state in which the disk device 100 is ready for use. The interface (1) 101 and the interface (2) 102 have exactly the same physical address, and the interfaces (1) 101 and (2) 102 are the same. The access source device connected to the disk device 100 has the same data in the disk device 100. FIG. 5B shows a case where a write request is made to the interface (2) 102 after the initial state. In this case, the physical address "1" associated with the logical address "1" on the interface (2). Physical address “3” different from the physical address “1” associated with the interface (1) 101
And the data is written to the transferred physical address “3”. Incidentally, the logical address before the transfer on the interface (2) 102 corresponding to the physical address "1" where the transfer has occurred becomes blank after the transfer.

FIG. 5C shows the interface (1) 101.
This is the case when a write request is made to the logical address “2” of In this case, since the logical address "2" is in the shared state of sharing the common physical address "2" with the interface (2) 102, as in the case of FIG. 5B,
The logical address "2" of the interface (1) 101 is transferred to a physical address "4" different from the interface (2), and the data is written to the transferred physical address "4". Even if the same data exists in the interfaces (1) 101 and (2) 102 at this point, the hard disk group 107 will store the data at different physical addresses.

FIG. 5 (d) compares the data in the disk device 100 by the data duplication control process described later, and when the same data is stored in different physical addresses, the same data is stored. It shows that the logical address of is reassigned to the same physical address. In the example of FIG. 5D, since the data of the physical addresses “1” and “3” are the same, the logical address “1” of the interface (2) is reassigned to the physical address “1”, and Since the data of the physical addresses "2" and "4" are the same, an example is shown in which the logical address "2" of the interface (2) is reassigned to the physical address "4". In this case, the logical address to be transferred is blank, as in the case of transferring the logical address generated by the write request. The data of the physical address where the logical addresses are no longer associated becomes so-called "garbage" data. Such data duplication control processing operates completely independently of data input / output requests from the interfaces (1) 101 and (2). Normally, it is set to operate by designating a time when no I / O request is issued from the interfaces (1) and (2).

When a write request for the logical address "1" is issued from the interface (1) 101 after the state shown in FIG. 5B, the common physical address is not in the shared state. As shown in, the logical address is not transferred, and the data is written to the physical address “1”. If the data of the physical addresses "1" and "3" are not the same, the logical address is not reassigned as shown in FIG. 6 (d).

FIGS. 7 and 8 are flowcharts of the data duplication control process for integrating the same data existing in the disk device 100 into one. Here, confirmation and integration of the same data are performed based on the correspondence between the logical addresses and the physical addresses of the interfaces (1) 101 and (2) 102 developed in the table development memory 104. First, in order to compare the logical address of the interface (1) 101 as the comparison source and the logical address of the interface (2) 102 as the comparison destination, the logical address of the interface (1) 101 in the address table 200 is set to the logical address n of the first entry. Is set (step 701). Next, the same data is stored in the disk device 10
Whether the logical address of the interface (1) 101 exceeds the maximum value (final address) assigned when the capacity of the disk is determined in order to determine whether or not the check whether it exists in 0 is completed. It is judged whether or not it is over, and if it is over, it is ended, and if it is over, step 703 is executed.
(Step 702).

Next, the interface (1) 1 for comparison and comparison
Interface (2) corresponding to logical address n of 01
It is determined whether the logical address entry 102 is empty (step 703). Interface for comparison (1)
The logical address entry of the interface (2) 102 corresponding to the logical address n of 101 is not empty, and the physical address associated with the logical address n of the interface (1) 101 and the interface (2) 102
If the physical address associated with the logical address n is the same, it means that the logical address n has already been shared with data, and therefore it is not necessary to confirm whether it is the same data. Therefore, the interface (2) already exists in the same entry as the logical address of the interface (1) 101.
If the logical address of 102 is assigned, it is already shared and the logical address n + 1 of the next entry is compared (steps 703 and 714).

If not, that is, if the logical address entry of the interface (2) 102 corresponding to the logical address n of the interface (1) 101 to be compared is empty, the physical address corresponding to the logical address n is changed. The same data as the data is interface (2) 1
The logical address of the interface (2) 102 is set to the leading logical address m in order to check whether it exists in the physical address indicated by any of the logical addresses 02 (step 704).

Next, it is judged whether the logical address of the interface (2) exceeds the final address (step 7).
05) If not, it is determined whether the logical address of the interface (1) 101 corresponding to the logical address m of the interface (2) to be compared is empty.
(Step 706). Interface for comparison (2)
If the logical address entry of the interface (1) 101 corresponding to the logical address m of 102 is not empty,
Since the logical address m is already in the data sharing state, it is not necessary to confirm whether it is the same data. Therefore, the logical address m + 1 of the next entry is compared as already shared (step 713).

However, when it is empty, the data is in a non-shared state, so that the logical addresses n and m of both interfaces (1) 101 and (2) 102, which are the comparison source and the comparison destination, respectively. The data of the associated physical address is read (steps 707 and 708) and compared to see if they are the same (step 709). If they are not the same, the logical address entry of the comparison destination is advanced (step 71).
3) If they are the same, the physical address indicated by the logical address of the interface (2) 102 which is the comparison destination is changed to the physical address indicated by the logical address of the interface (1) 101, and the shared state is set ( Step 71
0). That is, the interface (2) 1 corresponding to the logical address n indicated by the current interface (1) 101.
In the logical address area 02, the logical address m of the interface (2) is set as a comparison reference.

As a result, the same data as shown in FIG.
This means that the logical address is reassigned as shown in (d). Next, the interface (2) 1 that is the comparison target
The physical address associated with the logical address m of 02 is cleared (step 511). Since the data integration has been performed, the logical address m of the interface (2) is set to exceed the final address so that the comparison is completed (step 712).

In this way, the interface (1) 10
1, (2) For the write request including the logical address from 102, the address table 200 is searched, and based on the search result, the logical address in the write request and the logical address of another interface share the same physical address. If it is in the shared state, the logical address of the write request is associated with the physical address in the unshared state, the address table 200 is updated, and data is written to the associated physical address in the unshared state. In the non-shared state, the data writing to the physical address corresponding to the logical address of the write request is permitted, and the address table 200 is set after the data writing is completed.
To detect the unshared logical address that is not associated with the same physical address, compare the data at the physical addresses corresponding to the unshared logical address, and determine whether they are the same or not. When the data is the same, the address table 200 is updated by associating the logical address with the same physical address, so that an exclusive control mechanism or the like is provided to the access source device such as a server computer that uses the disk device 100. It is possible to reduce the required resource amount when a plurality of access source devices are connected by consolidating the same data without adding. Therefore, there is an advantage that the disk capacity can be significantly reduced especially in a system that shares a large number of identical data.

Further, it is possible to independently perform the data operation by the input / output request from the access source device connected to the disk device and the data operation for reducing the redundant use of the internal data. Even after storing the data, even if the access source device storing the same data is stored from a different access source device, even if the access source device does not recognize the same data, the same data is stored in the disk device. Can be processed. Furthermore, from the perspective of the access source device, the disk device appears to be connected to each access source device independently, so when the access source device is in a stopped state due to a failure, etc. However, it is possible to prevent other access source devices from being affected, and to prevent the access source device from being affected even if the processing function of the present invention is added to the disk device.
In the above embodiment, in order to simplify the description,
Although the example in which the number of physical addresses is four and the number of logical addresses is two for each interface is shown, the present invention is not limited to this, and it goes without saying that a larger number of physical addresses and logical addresses can be used. Yes. Also, the number of interfaces is not limited to two.

[0029]

As described above, according to the present invention, the same data is consolidated and connected to a plurality of access source devices without adding an exclusive control mechanism or the like to the access source device using the disk device. In this case, the required resource amount can be reduced.

[Brief description of drawings]

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

FIG. 2 is a configuration diagram for explaining an address table expanded in a table expansion area and a concept of use thereof.

FIG. 3 is a processing outline diagram showing an outline of processing in the disk device.

FIG. 4 is a flowchart showing processing of the disk device in response to an input / output request from the outside.

FIG. 5 is a diagram showing an example of state transition of an address table in response to an input / output request from the outside.

FIG. 6 is a diagram showing another example of the state transition of the address table in response to an input / output request from the outside.

FIG. 7 is a flowchart showing a data duplication control process.

FIG. 8 is a flowchart showing a continuation of FIG.

[Explanation of symbols]

100 ... Disk device, 101 ... Input / output interface (1), 102 ... Input / output interface (2), 103
Control central processing unit, 104 table expansion memory, 105 disk cache memory, 106 array controller, 107 hard disk group, 108 internal bus, 200 address table.

Claims (4)

[Claims] 1. A table is provided for managing a physical address indicating a physical read / write position of a disk device and a logical address indicating a logical read / write position assigned to each access source device in association with each other. A disk sharing control method for sharing one disk device, comprising: a first step of searching the table for a write request including a logical address from any access source device; The second step of determining whether or not the logical address in the request and the logical address of the other access source device are in the shared state of the same physical address, and if it is in the shared state, the logical address of the write request is in the non-shared state. Update the table in association with the address,
The third step of permitting data write to the associated physical address in the non-shared state, and permitting data write to the physical address corresponding to the logical address of the write request in the non-shared state, and the data write end After that, the table is searched, a logical address in a non-shared state that is not associated with the same physical address is detected, and data at physical addresses corresponding to the logical address in the non-shared state are compared to determine whether they are the same. A disk sharing control method comprising: a fourth step of determining whether or not, and a fifth step of updating the table by associating the logical address with the same physical address when the data is the same. . 2. The physical address is assigned so that all read / write positions of a disk device can be designated, and the logical address is assigned so that read / write positions less than the physical address can be designated.
The disk sharing control method described in. 3. A disk sharing control device for sharing one disk device by a plurality of access source devices, wherein a physical address indicating a physical read / write position of the disk device and a logical read / write assigned to each access source device. A table that manages a logical address indicating a position in association with each other, a first means for searching the table for a write request including a logical address from any access source device, and a write request based on the search result Second means for determining whether or not the logical address in the above-mentioned logical address and the logical address of another access source device are in the shared state of the same physical address; and in the shared state, the logical address of the write request is set to the unshared physical address. Update the table in association with
Third means for permitting data writing to the associated physical address in the non-shared state, and permitting data writing to the physical address corresponding to the logical address of the write request in the non-shared state, and data writing end After that, the table is searched, a logical address in a non-shared state that is not associated with the same physical address is detected, and data at physical addresses corresponding to the logical address in the non-shared state are compared to determine whether they are the same. A disk sharing control device comprising: a fourth means for determining whether or not the same data and a fifth means for updating the table by associating the logical address with the same physical address when the data are the same. . 4. The physical address is assigned so that all read / write positions of the disk device can be designated, and the logical address is assigned so that read / write positions less than the physical address can be designated.
The disk sharing control device described in 1.