JP3481391B2 - File data distributed storage method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a file data storage system in data management of a computer system, and more particularly to a file data distributed storage system for storing data in a plurality of secondary storage devices connected to a plurality of computers. Regarding

[0002]

2. Description of the Related Art In an OS normally used in a current computer system, the storage area of one secondary storage device is divided into several parts, which can be treated as independent secondary storage areas. This logical secondary storage area is called a file. Being able to create files makes it possible to use multiple independent secondary storage areas without being constrained by the number of actual secondary storage devices. further,
As long as the total capacity of all files does not exceed the capacity of the secondary storage device, it is possible to increase the capacity of individual files and vice versa, as necessary. That is, the storage capacity allocation to each file can be set variously.
It was not possible to manage such a flexible secondary storage area by directly using the hardware secondary storage device.

However, in the file management function of an ordinary OS, a data structure for managing files (usually called a file system) is constructed on one secondary storage device, and files are managed in the data structure. , The storage area of one file was completely contained in one secondary storage device. Therefore, each file could not exceed the capacity of the secondary storage device to which it belongs. Also, when performing multi-tasking, if multiple tasks access different files in the same file system almost simultaneously,
Since the file access can be processed only in order, there is a problem that the overall processing performance becomes slow. That is, the throughput of file access is limited to the throughput of one secondary storage device.

Therefore, the "striping area allocation method" (Japanese Patent Laid-Open No. 3-259320) and the "File data multiplexing method and data processing system" (Japanese Patent Laid-Open No. 6-58242).
A method of allocating the storage area of one file from a plurality of secondary storage devices has been proposed, as in claim 6 of Japanese Patent No. 3332625). In these methods, a virtual secondary storage device composed of multiple secondary storage devices is defined,
Build a large file system on top of this. Then, it becomes possible to allocate storage areas on a plurality of secondary storage devices to the files in the file system. As a result, there are advantages such as being able to create a file that exceeds the capacity of one secondary storage device, and operating a plurality of secondary storage devices at the same time to speed up reading and writing of file data. Further, file access in the same file system becomes access to different secondary storage devices with a certain probability. Since access to different secondary storage devices can be processed at the same time, the overall file access throughput is improved accordingly.

However, I / O that can be connected to one computer
Since the number of control devices and secondary storage control devices is finite, there is an upper limit to the number of actual secondary storage devices that make up this virtual secondary storage device. So, with these methods,
You cannot create a file that exceeds the total capacity of the secondary storage device that can be connected to one computer. In addition, the main memory bus and I / O
File access throughput is also limited due to finite bus performance.

In order to exceed these capacity and access throughput limits, it is basically necessary to use a plurality of computers. As a file system capable of managing a plurality of secondary storage devices connected to a plurality of computers, there is a distributed file system such as NFS (Chapter 6 of "NFS &NIS" issued by ASCII Publishing Bureau). Distributed file system
The name space of the file system built on the secondary storage of each computer is connected to realize a large network-wide name space. Further, it enables remote access to files in any computer from an application program on any computer.

However, in these systems, since the file management function in the OS on each computer is used as it is, the storage area of each file is completely stored in the secondary storage device connected to one computer. include. That is,
Individual files cannot exceed the total capacity of the secondary storage device connected to one computer. On the other hand, as long as the file accesses occurring in the system are evenly distributed to the files belonging to each computer, the overall file access throughput can be the throughput of all the secondary storage devices configured in the system. However, when access concentrates on files belonging to a specific computer,
Only the file access throughput of the computer to which they belong can be output. As described above, in the control method of the distributed file system, if the accessed files are biased, the throughput becomes low.

On the other hand, "disk striping device"
As disclosed in Japanese Patent Laid-Open No. 7-226019, a method has been proposed in which striping storage in a plurality of disk devices is directly controlled by hardware. By using this method, if the scale of hardware is increased, the secondary storage capacity can be increased accordingly. Further, since the striping is performed, the secondary storage accesses that occur almost at the same time are distributed to the accesses to the different secondary storage devices, and the access throughput can be increased according to the number of the secondary storage devices.

However, this control method cannot handle a logical file. That is, it becomes one secondary storage area as a whole. It may be possible to define multiple fixedly partitioned secondary storage areas by adding some control. However, it is not possible to expand or contract each secondary storage area. That is, the file management system is very inflexible.

[0010]

The first problem is that even if a large number of computers are connected by a communication mechanism to increase the scale of the system, the capacity of the secondary storage device connected to one computer can be reduced. It is impossible to create a file that exceeds the limit. The reason is that in the conventional distributed file system, the management of the storage area of each file is entrusted to the file management function in each computer, and further, these conventional file management functions can only be used for file management that is closed in one computer. This is because it has not been realized.

A second problem is that even if a large number of computers are connected by a communication mechanism to increase the system scale, if access concentrates on a file group belonging to one computer, the overall file access throughput will be that computer. The secondary storage access performance is limited. The reason is, as pointed out in the first problem, in the conventional combination of the distributed file system and the file management function,
The storage area of each file is confined in the secondary storage device group connected to a specific computer, and access to these file groups must be done by that computer without fail. is there.

The third problem is that even if a large number of computers are connected by a communication mechanism to increase the performance of the entire system, individual file access is not accelerated. The reason is, as pointed out in the first problem, in the conventional combination of the distributed file system and the file management function,
This is because the storage area of each file is confined in the secondary storage device group connected to a specific computer, and one file access is limited by the performance of this subsystem.

A fourth problem is that when a storage area for one file is allocated from a plurality of secondary storage devices connected to a plurality of computers, the file management overhead increases according to the scale of distribution. . The reason is that the management data of the files to be distributed and stored are also distributed to the plurality of secondary storage devices, and the cooperative operation of the plurality of computers is required for the management of each file.

Therefore, an object of the present invention is to make it possible to increase the storage capacity of individual files depending on the scale of the system.

[0015] Another object of the present invention is to reduce the number of accesses to a specific file, depending on the scale of the system.
It is to be able to increase the file access throughput.

Another object of the present invention is to speed up large-sized single file access.

Yet another object of the present invention is to reduce the overhead of file management operations for files with distributed secondary storage.

[0018]

According to a first file data distributed storage method of the present invention, one or a plurality of secondary storage devices (D in FIG. 1) for recording file data and file management data are connected. A plurality of secondary storage control computers (C in FIG. 1), one or a plurality of file management computers (B in FIG. 1) capable of communicating with any of the secondary storage control computers, and any of the secondary storage One or more application execution computers (A in FIG. 1) capable of communicating with the storage control computer and any of the file management computers, and a communication mechanism group (N in FIG. 1) that enables communication between these computers. A method for distributed storage of file data in a multi-node computer system having: a part of a partial file configured as a set of storage areas in one secondary storage device and managing the partial file. It is a management data of the file parts
Minute file management data and a set of storage areas corresponding thereto are stored in the same secondary storage device, and one logical file is stored as a set of the partial files distributed and stored in the plurality of secondary storage devices. And a file management data, which is management data for each file, is stored in an arbitrary secondary storage device, operates on the secondary storage control computer, and has the secondary storage control computer. I / O issuing means (12 in FIG. 1) for issuing an I / O request to the secondary storage device connected to the secondary storage device, and operating on the secondary storage control computer,
Partial file opening means (8 in FIG. 1) for reading specified partial file management data on the secondary storage device connected to the secondary storage control computer and enabling access to the partial file; A part that operates on the storage control computer and writes the modified partial file management data of the specified partial file on the secondary storage device connected to the secondary storage control computer, and ends access to the partial file. File closing means (9 in FIG. 1),
Address conversion that operates on the secondary storage control computer and converts the partial file offset of the specified partial file on the secondary storage device connected to the secondary storage control computer to the physical address on the secondary storage device Means (11 in FIG. 1)
If, works with the file management computer machine, the read pre-notated Airu management data by I / O issuing unit from the secondary storage device that stores the file management data, the I / O issued by using the file management data It reads the partial file management data from the secondary storage device for storing the partial file management data by means the partial file
And file opening means for opening all parts files of the file by the partial file open means with the management data (4 in FIG. 1), operating in the file management computer machine, all the files by the partial file close means File closing means for closing the partial file and writing the changed file management data to the secondary storage device by the I / O issuing means (5 in FIG. 1)
And a partial file determining means (3 in FIG. 1) that operates on the application execution computer and calculates the corresponding partial file and the offset within the partial file from the offset of the specified file, and operates on the application execution computer The partial file is selected by the file determination means, the physical address is calculated by the address conversion means on the secondary storage control computer to which the secondary storage device to which the selected partial file belongs is connected, and the partial file in charge of the partial file is calculated. File read / write means for reading / writing data by the I / O issuing means on the next storage control computer (see FIG. 1).
2) and, on the application execution computer, determine the file access request specified by the application program, and if the file access request is open, issue a file open processing request to the file open means of the file management computer. Send, when the file access request is closed, sends a file close processing request to the file close means of the file management computer, and when the file access request is read or write, sends the file access request to the file read / write means. Send, file access required
Calculated open, outside closed, when the read or write, the said file access request file management computer sends to the application execution computer, the file access control means (Fig. Back to the wait for a response file access request application program 1) 1) and a file data distributed storage method.

According to a second file data distributed storage system of the present invention, in the configuration of the first file data distributed storage system, the address conversion means is replaced by an operation on a secondary storage control computer, A means for converting an in-partial-file offset of a specified partial file on a secondary storage device connected to a storage control computer to a physical address on the secondary storage device, and allocating an area when converting an address of an unallocated storage area A file size enlargement means (21 in FIG. 17) that has an extended address conversion means (21 in FIG. 17) having a function of performing the above operation and that operates on the file management computer to change a specified file to a specified file size. 7) and operates on an application execution computer instead of the file read / write means,
A partial file is selected by the partial file determination means, a physical address is calculated by the extended address conversion means on the secondary storage control computer to which the secondary storage device to which the selected partial file belongs is connected, and the partial file is calculated. Is a means for reading and writing data by the I / O issuing means on the secondary storage control computer that is in charge of the above, and in the case of additional writing, expanding the file size in advance by the file size expanding means and expanding the address conversion. An automatic expansion type file read / write means (22 in FIG. 17) for calculating a physical address by means is provided.

The third file data distributed storage system of the present invention operates on a secondary storage control computer in addition to the configuration of the first file data distributed storage system described above, and a specified partial file is specified. Partial file reducing means (10 in FIG. 22) for reducing to size and file reducing means (6 in FIG. 22) operating on the file management computer and reducing each partial file of the file designated by the partial file reducing means. Have and.

According to a fourth file data distributed storage method of the present invention, in the first file data distributed storage method, an operation is performed on an application execution computer instead of the file read / write means, and the partial file determination means is used A file is selected, a physical address is calculated by the address conversion means on the secondary storage control computer to which the secondary storage device to which the selected partial file belongs is connected, and on the secondary storage control computer in charge of the partial file. Of the I / O issuing means for reading and writing data, dividing a specified read / write request into a plurality of small read / write requests, calculating the physical address by the address converting means for a plurality of read / write fragments, and executing the I / O. /
It has a file parallel reading / writing means (13 in FIG. 26) for simultaneously performing reading / writing by the O issuing means.

According to a fifth file data distributed storage method of the present invention, in the second file data distributed storage method, the automatic file extension read / write means operates in place of a secondary storage control computer. The offset in the partial file of the specified partial file on the secondary storage device connected to the secondary storage control computer is converted to the physical address on the secondary storage device, and area allocation is performed when the address of the unallocated storage area is converted. It has a function and further divides a specified read / write request into a plurality of small read / write requests, calculates the physical address by the extended type address conversion means for a plurality of read / write fragments, allocates a new secondary storage area and the I / O. An automatic expansion type file parallel read / write means for executing reading / writing of the secondary storage device by the issuing means at the same time (see FIG. With a 23).

According to a sixth file data distributed storage method of the present invention, in the third file data distributed storage method, a file management computer is operated instead of the file reduction means and designated by the partial file reduction means. A means for reducing each partial file of the created file, and a file parallel reducing means (16 in FIG. 33) for simultaneously activating a plurality of the partial file reducing means.

According to a seventh file data distributed storage method of the present invention, in the first file data distributed storage method, a file management computer is operated instead of the file opening means, and the I / O issuing means operates. reads the parts content off § <br/> yl management data from the secondary storage device, and means for opening all parts files of the file by the partial file open means, activates a plurality of the partial file open means simultaneously It has a file parallel opening means (14 in FIG. 36).

An eighth file data distributed storage system according to the present invention is the same as the first file data distributed storage system, which operates on a file management computer in place of the file closing means, and operates by the partial file closing means. Close all subfiles of
The file management data of the partial file changed by the / O issuing means is a means for writing to the secondary storage device, and has a file parallel closing means (15 in FIG. 39) for simultaneously activating a plurality of the partial file closing means.

It is to be noted that a configuration for enabling communication between arbitrary secondary storage control computers and operating each of the above means operating on a part of the file management computers on the secondary storage control computer or execution of an arbitrary application It is also possible to adopt a configuration in which communication between computers is enabled and each of the above-mentioned means operating on a part of the file management computer is operated on the application executing computer. It also includes a dual-purpose computer that can execute application programs and is connected to a secondary storage device, and enables communication between any dual-purpose computer and any other management computer. It is also possible to employ a configuration in which both of the above-mentioned means and the above-mentioned means operating on the secondary storage control computer are operated.

According to the first file data distributed storage method of the present invention, a logical file in which storage areas are distributed and arranged in a secondary storage device connected to a plurality of secondary storage control computers is defined. It will be possible. Also, by allowing the file management data of one file to be distributed to a plurality of secondary storage devices, the management data of each partial file is stored in the secondary storage device to which the partial file belongs, and each partial file is stored. All of the management tasks described above can be executed by the secondary storage control computer to which the secondary storage device is connected.

On the other hand, the file opening means operating on the file management computer controls the opening of the entire file,
Based on this commander, open the secondary part file open means of the storage control computer by performing the open processing part file of the file, the file section file management data is dispersed into a plurality of secondary storage devices Will be processed correctly.

Similarly, the file closing means operating on the file management computer controls the closing of the entire file, and based on this command, the partial file closing means on each secondary storage control computer closes the partial file of the file. processing by performing, closing files that partial file management data is distributed and stored in a plurality of secondary storage device will be able to process correctly.

Furthermore, the file read / write means on the application execution computer controls the entire read / write of the file, and the partial file determination means in the application execution computer determines the read / write position in the designated file as the target partial file and the partial file. Of the secondary storage control computer, the address conversion means on each secondary storage control computer finds the target area on the secondary storage device from the read / write position in each partial file, and also the I / O on each secondary storage control computer. By issuing and reading the target area by the issuing means, a virtual area in the file is distributed to a file whose storage area is distributed to a plurality of secondary storage devices connected to a plurality of secondary storage control computers. Reading and writing by specification will be performed correctly. The address conversion information is held for each partial file and included in the management data of the partial file. That is, by reading this data into the secondary storage control computer by the partial file opening means,
The subsequent address conversion means can be processed.

As described above, by using the first file data distributed storage method of the present invention, it becomes possible to read and write the files distributedly stored in the plurality of secondary storage devices connected to the plurality of computers.

According to the second distributed file data storage method of the present invention, when a write to an area larger than the current file size is requested, the automatic extension type file read / write means on the application execution computer detects this. However, the file size expansion means on the file management computer expands the file size in the file management data, and the extended address conversion means in the secondary storage control computer has not yet allocated the secondary storage area in the partial file. Allocate a new secondary storage area at the position and write the specified data. By this control, it becomes possible to expand the files distributedly stored in the plurality of secondary storage devices of the plurality of computers.

According to the third file data distributed storage method of the present invention, the reduction size of each partial file is calculated from the reduction size of the designated file by the file reduction means on the file management computer, and each secondary storage control is performed. The partial file reducing means on the computer reduces the partial file to this size. By this control, it becomes possible to reduce the files distributedly stored in the plurality of secondary storage devices of the plurality of computers.

According to the fourth file data distributed storage method of the present invention, the file parallel read / write means on the application execution computer divides the requested read / write range into a plurality of fragments, and the partial file corresponding to these read / write fragments is stored in the partial file. Address translation and secondary storage I / O are simultaneously executed by the address translation means and the I / O issuing means on a plurality of secondary storage control computers. By this control, the partial file address conversion for one read / write request and the secondary storage device I / O are executed in parallel.

According to the fifth file data distributed storage system of the present invention, the automatic extension type file parallel read / write means on the application execution computer divides the requested read / write range into a plurality of fragments and requests the address translation for this. And the secondary storage device I / O request are simultaneously sent to the extended address conversion means and the I / O issuing means on a plurality of secondary storage control computers. The extended type address conversion means on each secondary storage control computer allocates a new secondary storage area at that time if the write request is for an unallocated area,
Data is written in this area. That is, by this control, not only the partial file address conversion for one read / write request and the secondary storage device I / O are processed in parallel, but also executed in parallel including the file expansion process.

According to the sixth file data distributed storage method of the present invention, the file parallel reduction means on the file management computer simultaneously activates the partial file reduction means on the plurality of secondary storage control computers. By this control, the file reduction processing operates in parallel.

According to the seventh file data distributed storage system of the present invention, the file parallel open means on the file management computer simultaneously activates the partial file open means of the plurality of secondary storage control computers. As a result, the file opening process operates in parallel. The file open process is mainly to read the file management data from the secondary storage device. That is, by this control, the reading of the file management data distributed to the plurality of secondary storage devices can be executed in parallel.

According to the eighth file data distributed storage system of the present invention, the file parallel closing means on the file management computer simultaneously activates the partial file closing means on the plurality of secondary storage control computers. As a result, the file closing process operates in parallel. The file closing process is mainly to write back the updated file management data to a predetermined position in the secondary storage device. That is, by this control, the write-back of the file management data distributed in the plurality of secondary storage devices can be executed in parallel.

[0039]

BEST MODE FOR CARRYING OUT THE INVENTION Next, an example of an embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 1, in the first embodiment of the present invention, one or a plurality of secondary storage devices D for recording file data and file management data are connected, and this secondary storage device D is connected. A plurality of secondary storage control computers C that can access the server, one or more file management computers B that can communicate with any secondary storage control computer C, any secondary storage control computer C, and any file It includes one or a plurality of application execution computers A capable of communicating with the management computer B, and a communication mechanism group N enabling communication between these computers.

The secondary storage control computer C is a secondary storage device D.
I / O issuing means 12 for issuing an input / output request to the secondary storage control computer 12 and management data of a specified partial file on the secondary storage device D connected to the secondary storage control computer C, and access to the partial file. And a partial file opening means 8 for enabling the above, and a part for writing the management data of the specified partial file on the secondary storage device D connected to the secondary storage control computer C and ending the access to the partial file. The file close means 9 and the offset in the partial file of the specified partial file on the secondary storage device D connected to the secondary storage control computer C are set to the secondary storage device D.
Address conversion means 11 for converting the above physical address.

The file management computer B reads the file management data from the secondary storage device D by the I / O issuing means 12 and the file opening means 4 for opening all the parts of the file by the partial file opening means 8.
The partial file closing means 9 closes all the parts of the file, and the I / O issuing means 12 writes the latest file management data to the secondary storage device D.

The application execution computer A calculates the partial file and the offset within the partial file from the offset of the specified file, and the partial file determining means 3
And a partial file is selected by the partial file determination means 3, the physical address is calculated by the address conversion means 11 on the secondary storage control computer C that is in charge of the selected partial file, and the secondary file in charge of that partial file is calculated. The file read / write means 2 for reading / writing data by the I / O issuing means 12 on the storage control computer C, and the file access request designated by the application program are transmitted to the file open means 4, the file close means 5, and the file read / write means 2. And a file access control means 1 to be executed.

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

Referring to FIG. 2, in each secondary storage device D, secondary storage management data D1 for storing management information of free areas in the secondary storage device and for each file such as size of each file. A file management data table D2 for storing management data, a partial file management data table D3 for storing management data for each partial file such as address conversion information in each partial file, and a data block area D4 for storing file data. Composed of.

A partial file is a set of storage areas allocated to the same secondary storage device D among the storage areas of one file. That is, each file is defined as a set of partial files on the plurality of secondary storage devices D.

Referring to FIG. 3, the management data structure of each file includes a file management data F1 that holds management data for each file such as a file size FS, and a partial file such as an address conversion table FA within a partial file. Partial file management data F2 that holds management data
And a set of data storage blocks FB designated by the address conversion table FA. Therefore, the above-mentioned partial file is a set of data blocks FB of files belonging to one secondary storage device D. Further, the management data F2 of the partial file is connected to the same secondary storage control computer C as the secondary storage control computer C to which the secondary storage device D storing the data block FB group is connected. It is important that the data is stored in the storage device D. As a result, access control to one partial file can be performed by only one secondary storage control computer C. If this condition is not satisfied, it becomes necessary to operate a plurality of secondary storage control computers C in the process of accessing the partial file, and the file read / write processing becomes slow. Generally, assuming that the secondary storage device D is connected to another secondary storage control computer C or the secondary storage device is replaced, as shown in FIG. It is desirable to store it in D. The file management data F1 may be stored in any of the secondary storage devices D without any problem. It is possible to store the file management data F1 of all files collectively in a specific secondary storage device D, but it is often advantageous in terms of performance to store the file management data F1 in a distributed manner in a plurality of secondary storage devices D. .

The file name is the number of the secondary storage device D in which the file management data F1 of the file is stored, and the file management data F in the secondary storage device D.
The position of 1 is directly included. In a general file system, a more abstract file name can be used, and the file management data is searched for from the file name specified at the time of opening. However, the resolution of this file name is a function not directly related to the problem to be solved by the present invention. Therefore, for the sake of simplicity, the description will be made on the premise of the file name of the above rule. If you want to use a more abstract file name, combine the function of determining the number of the secondary storage device D in which the file management data F1 is stored and the position on the secondary storage device D from the file name. Good.

Referring to FIG. 4, the process of the file access control means 1 is at the access type judging step 101.
Determine the type of file access request from the application program.

If the file access type is open, in the file access management area securing step 102,
An area for file access management data for controlling the subsequent file access request is secured, and in the file management computer selection step 103, the file management computer B that manages the file is selected from the designated file names, The identifier of the selected file management computer B is stored in the file access management area secured in the file access management area securing step 102, and in the file open request transmission step 104, on the file management computer B selected in the file management computer selection step 103. To the file open means 8 and waits for a response from the file management computer B which sent the processing request in the file open request transmission step 104 in the file open response waiting step 105, and in the file management data storage step 106. Such as identifiers and file size of the file management area of the file management computer in B included in the response to the processing request, the management data required for a subsequent file access, file access management area securing step 1
It is stored in the file access management area secured in 02, and returns to the application program with the identifier of the secured file management area.

When there are a plurality of file management computers B, the range of files managed by each file management computer B is determined at the time of system configuration, and this role division is known to all application execution computers A in advance. To do. That is, from the file name to the file management computer B in charge
Shall be determined.

When the file access type is close, in the file close request transmission step 109, the identifier of the file management computer B that manages the specified file is obtained from the specified file management area identifier, and this file management is performed. A processing request is transmitted to the file closing means 5 on the computer B, and a file close response waiting step 110 waits for a response from the file management computer B, and a file access management area releasing step 11
At 1, the file management area designated by the application program is released, and the process returns to the application program.

If the file access type is other than open or close, in the request sending step 107, the specified access request is sent to the appropriate computer together with the specified parameters, and in the response reception waiting step 108, the requesting computer is sent. Waits for a response from, and when the response comes back, returns to the application program with the processing result. The computer that sends this request changes depending on the access type. Within the scope of the first aspect of the invention described herein, there can be read and write. In the case of these access types, the request is notified to the file reading / writing means 2 on the same application execution computer A as the application program. In addition to this pattern, a file reduction request may come in the subsequent invention. In this case, the identifier of the file management computer B that manages the specified file is obtained from the specified file management area identifier. Then, the file reducing means 6 on the file management computer B is notified of the processing request.

Referring to FIG. 5, the process of the file read / write means 2 is to limit the specified read / write range within the file size in the file access management area in the read / write range restriction step 206, and in the read / write block extraction step 207. One block is extracted from the read / write range, and in the read / write remaining determination step 208, it is determined whether or not the read / write target block has been extracted.

When there is a read / write target block, the partial file determining means 3 is called with the in-file offset of the first byte of the extraction block, the partial file to which the extracted block belongs and the offset in the partial file are calculated, and the address is calculated. Conversion request transmission step 209
In the partial file calculated by the partial file determining means 3, the address converting means 11 on the secondary storage control computer C managing the secondary storage device D to which the partial file determined by the partial file determining means 3 belongs A processing request is transmitted with an offset, a response from the secondary storage control computer C is waited in the address translation response reception waiting step 210, and the I / O issue request transmission step 211 is executed.
The number of the secondary storage device D to which the partial file calculated by the partial file determining means 3 belongs, and the address converting means 11
Of the physical block number included in the response, the size of the block extracted in the read / write block extraction step 207, and the write data passed from the application program when the access type is write, the part determined by the partial file determining means 3 A processing request is transmitted to the I / O issuing means 12 on the secondary storage control computer C that manages the secondary storage device D to which the file belongs, and the access type is determined in the I / O issue response reception waiting step 212. ,
Only when the access type is read, in the read data copying step 214, the read data included in the I / O issue response is written in the area designated by the application program, and the process returns to the read / write block extraction step 207 to read the next block. Repeat the same process. In addition,
Since the connection relationship between the secondary storage device D and the secondary storage control computer C is determined when the system is configured, it is assumed that it is recorded in advance in all computers. That is, both computers
By referring to this information, the identifier of the secondary storage control computer C that manages it can be obtained from the number of the secondary storage device D.

In the read / write remaining determination step 208, when it is determined that there is no read / write target block, the processing is terminated and the process returns to the calling source.

Referring to FIG. 6, the partial file determining means 3 executes in-file block number calculation step 301.
By the following calculation, the in-file block number and the in-block offset are calculated from the in-file offset specified by the request source and the block size defined by the system.

Block number in file = (offset in file) ÷ (block size) offset in block = (offset in file)%
(Block size) In addition, ÷ is an operator for calculating the quotient obtained by dividing the left side by the right side, and% is the remainder for dividing the left side by the right side.

Further, the partial file number calculation step 3
In 02, the partial file number and the partial file offset are calculated from the value calculated in the in-file block number calculating step 301 and the number of the secondary storage devices D configured in the system by the following calculation.

Partial file number = secondary storage device number =
(Block number in file)% (Number of secondary storage devices) Offset in partial file = {(Block number in file) / (Number of secondary storage devices)} x (Block size) +
(Offset within block)

After that, the process returns to the calling source with the partial file number and the partial file offset.

Referring to FIG. 7, the file opening means 4 searches the file management area of the file designated by the request source in the designated file retrieval step 401, and opens the file of the designated file in the open state determination step 402. It is determined whether the management area is already registered.

If the file management area of the specified file is not registered, it means that no other application program has opened the corresponding file either. In the file management area securing step 403, a new file management area is created. In the file management area initialization step 404, the secured file management area is initialized by setting the number of file references (the number of application programs that open the file) to 1, and the file management data read request is transmitted. In step 405, the secondary storage control computer C that manages the secondary storage device D in which the file management data F1 of the specified file is stored.
A request for reading the corresponding file management data F1 is transmitted to the above I / O issuing means 12, and a file management data read response reception waiting step 406 waits for a response from the secondary storage control computer C, and then the file management data F1. In the storage step 407, the read file management data F1 is stored in the file management area secured in the file management area securing step 403, and in the partial file selection step 408.
Then, select the partial files of the specified file one by one in order.

When a partial file is selected, in a partial file open request transmission step 409, the partial file open means 8 on the secondary storage control computer C that manages the secondary storage device D to which the selected partial file belongs is processed. The request is transmitted, and in the waiting step 410 for receiving the partial file open response, the reply from the secondary storage control computer C is waited for, and when the reply arrives, the partial file selecting step 40
Return to step 8 and select the next partial file.

When all the partial files have been selected in the partial file selection step 408, the file open response transmission step 414 records them in the identifier of the file management area secured in the file management area securing step 403 and the area thereof. The file open response is transmitted to the request source together with some management data such as the file size being set, and the processing is completed.

If the file management area of the specified file is found in the open determination step 402,
This means that another application program has already opened the file, and in the open prohibition state determination step 411, it refers to the detected file management area and determines whether the open from another client is prohibited. .

If the open is prohibited, the file open failure response transmission step 412 transmits the file open response to the request source together with the code indicating the open failure, and the processing is terminated. It should be noted that the state in which the opening is prohibited means that the file is being reduced.

By exclusively allowing the file reduction and the file opening, the integrity of the file read / write operation can be guaranteed at low cost. If the file is reduced during the read / write control, a problem that the block that existed at the time of address translation is released at the timing of the subsequent read / write occurs. In such a case, the integrity of access cannot be maintained unless synchronization control with the file management computer B is performed during reading and writing. Such control causes an increase in communication and causes considerable overhead. Therefore, the function is restricted so that the file reduction and the file opening are exclusively permitted, and in return, the synchronous control at the time of reading and writing is omitted to prevent the performance deterioration.

In the open prohibition state judgment step 411,
If the specified file is permitted to be opened, in the file reference count increment step 413,
One is added to the number of file references in the detected file management area, and in the file open response transmission step 414, a part of the file management area identifier detected in the designated file search step 401 and the file size recorded in that area A file open response is transmitted to the request source together with the file management data, and the processing is terminated.

Referring to FIG. 8, the file closing means 5 executes the file reference number decrement step 501,
1 is subtracted from the file reference number of the file management area of the specified identifier, and the file reference number determination step 502
Then, the number of file references after subtraction is determined.

If the number of file references is zero, it means that there is no other application program that has the file open, and the partial file selection step 5
In 03, the partial files of the specified file are sequentially selected.

While the target partial file is present, in the partial file close request transmission step 504, the partial file close means 9 on the secondary storage control computer C that manages the secondary storage device D to which the selected partial file belongs is A process request is sent to the file, and in step 505, waiting for reception of a partial file close response, the response from the secondary storage control computer C is waited for. When the response comes back, the process returns to the partial file selection step 503 to select the next partial file. To do.

When all the partial files have been selected in the partial file selection step 503, the file management data write request transmission step 506 manages the secondary storage device D in which the file management data F1 of the file should be stored. A file management data write request is transmitted to the I / O issuing means 12 on the secondary storage control computer C, and a file management data write response reception waiting step 507 is executed.
Waiting for a response from the secondary storage control computer C, releasing the file management area of the designated file in the file management area releasing step 508, transmitting the file closing response to the request source in the file closing response transmitting step 509, Finish the process.

If the number of file references is not zero in the file reference number determination step 502, there is an application program that has the file open, so the process is not closed and the process jumps to the file close response transmission step 509.

Referring to FIG. 9, the partial file opening means 8 acquires the partial file management area in step 801.
In step 802, the management area of the designated partial file is acquired, and in step 802, a partial file management data read request transmission step 802 requests the I / O issuing means 12 to read the partial file management data F2 of the designated partial file. In a file management data read response wait step 803, a response is returned from the I / O issuing means 12, and in a partial file management data storage step 804, the partial file management data F2 included in the partial file management data read response is saved. The partial file management area is stored in the partial file management area acquired in step 801 and the partial file open response is transmitted in step 805 in which the partial file open response is transmitted to the request source, thus ending the processing.

Referring to FIG. 10, the partial file closing means 9 requests the I / O issuing means 12 to write the partial file management data F2 of the designated partial file in the partial file management data write request transmission step 901. Then, in the step 902 of waiting for reception of the response for writing the partial file management data, it is waited for a response from the I / O issuing means 12 and the step 90 of releasing the partial file management area is waited.
In step 3, the partial file management area of the designated partial file is released, and a partial file close response transmission step 90
In step 4, the partial file close response is transmitted to the request source, and the processing is completed.

Referring to FIG. 11, address conversion means 1
In the address conversion table index number calculation step 1101, the designated offset in the partial file is divided by the block size to calculate the block number in the partial file, and the physical block number acquisition step 1102.
Then, the block number in the partial file is used as an index to look up the address conversion table FA in the partial file management data F2 to obtain the physical block number recorded therein, and an address conversion response transmission step 1107.
Then, the address conversion response is transmitted to the request source together with the block number of the obtained block, and the process is completed.

Referring to FIG. 12, the I / O issuing means 12
In the reception request type determination step 1201, the type of the I / O request from the request source is determined.

If it is a read request, read I
In the I / O issuing step 1202, the read I / O of the specified physical block is issued to the specified secondary storage device D, and in the read I / O completion waiting step 1203,
The completion of the read process is waited for, and in a read response transmission step 1204, a read response is transmitted to the request source together with the read data, and the process is completed.

When the type of the I / O request from the request source is write in the reception request type determination step 1202, the write I / O issue step 1205 writes the specified secondary storage device D. , I / O for writing the passed data to the specified physical block is issued, and in the write I / O completion wait step 1206, the completion of the write I / O is waited, and the write response transmission step 1207
Then, the write response is transmitted to the request source, and the process ends.

Next, the operation of one example of the first embodiment of the present invention will be described in detail.

FIG. 13 shows a time chart of an embodiment of file opening. First, a file open request issued from an application program is passed to the file access control means 1 on the same application execution computer A as the application program (S).
O01). Here, the file management computer B that manages the specified file is selected. Next, an open request is sent to the file open means 4 on the computer (S
O02). Upon receiving this, the file opening means 4 selects the secondary storage device D in which the file management data F1 of the specified file is stored, and on the secondary storage control computer C to which the secondary storage device D is connected. I / O issuing means 1
The file management data F1 is read via S2 (SO0
3 to SO04). Next, the secondary storage device D group to which the partial files of the specified file belong is selected, and the partial file opening means 8 on all the secondary storage control computers C to which those secondary storage devices D are connected, Send a partial file open request (SO05, SO07, SO09). Upon receiving this request, the partial file opening means 8 reads the partial file management data F2 of the designated partial file to the secondary storage control computer C (SO06, SO08, SO
10).

According to this sequence, the file management data F1 of the designated file is assigned to the file management computer B in charge, and the partial file management data F2 of the partial files constituting the designated file is assigned to the entire secondary storage control computer C.
Read to and ready for file access. In particular,
It is important that the address conversion table FA is loaded not on the file management computer B but on the secondary storage control computer C.

FIG. 14 shows a time chart of an embodiment of file closing. First, the file close request issued from the application program is passed to the file access control means 1 on the same application execution computer A as the application program (S).
C01). Here, the file management computer B that manages the specified file is selected. Next, a file close request is sent to the file close means 5 on the computer (SC02). Upon receiving this, the file closing means 5 holds the partial file management data F2 of the partial file of the designated file, and the secondary storage control computer C.
A partial file close request is sent to the upper partial file close means 9 (SC03, SC05, SC07). Upon receiving this, the partial file closing means 9 writes the partial file management data F2 that may have been updated to the secondary storage device D (SC04, SC06, SC08). Finally, the file closing means 5 transfers the file management data F1 held in its own computer to the I on the secondary storage control computer C to which the secondary storage device D in which the data was originally stored is connected. Write to the secondary storage device D via the / O issuing means 12 (SC09 to SC10).

By this sequence, the file management data F1 of the designated file and the partial file management data F2 which are distributed and loaded in the plurality of secondary storage control computers C are surely stored in the predetermined storage of the secondary storage device D. It is written back to the place.

FIG. 15 shows a time chart of an embodiment of file reading. First, the file read request issued from the application program is passed to the file access control means 1 on the same application execution computer A as the application program, and further to the file read / write means 2 (SR01). Here, the read range of the specified file is sliced in block units. Then, for each block, the partial file determining means 3 is used to calculate the partial file to which the block belongs and the offset within the partial file, and the address conversion on the secondary storage control computer C that manages the determined partial file. The means 11 calculates the position of the block on the secondary storage device D from the offset in the partial file (SR02, SR05, SR08), and the I / O on the secondary storage control computer C that also manages the partial file.
The target block is read from the secondary storage device D by the issuing means 12 (SR03, SR04, SR06, S).
R07, SR09, SR10).

FIG. 16 shows a time chart of an embodiment of file writing. The file write is different from the file read in the moving direction of the user data. At first,
The file write request issued from the application program is passed to the file access control means 1 on the same application execution computer A as the application program, and further to the file read / write means 2 (SW01). Here, the writing range of the specified file is sliced in block units. Then, for each block, the partial file determining means 3 is used to calculate the partial file to which the block belongs and the offset within the partial file, and the address conversion on the secondary storage control computer C that manages the determined partial file. The means 11 determines the position of the block on the secondary storage device D from the offset in the partial file (SW02, SW05, SW0).
8), data is written in the target block of the secondary storage device D by the I / O issuing means 12 on the secondary storage control computer C which also manages the partial file (SW).
03, SW04, SW06, SW07, SW09, SW
10).

As can be seen from FIG. 15 and FIG. 16, the file read and file write operations are the application execution computer A on which the application program that issued the request operates, and the secondary storage in which the partial file to be read is stored. Only the device D group and the secondary storage control computer C group managing these secondary storage devices D are processed, and the file management computer B does not operate. This means that the communication of user data generated by reading and writing does not go through the file management computer B at all. Generally, user data communication is heavier than request and response communication. That is, the configuration of the present invention is a suitable communication protocol in the sense that the relay of the heaviest user data is small.

It is also possible to prepare a request that is a combination of an address conversion request and a data read request and a combination of an address conversion request and a data write request. By using this, it becomes possible to issue an address translation request and a data read / write request at the same time, the number of times of communication in file read and file write can be reduced, and performance is improved.

Referring to FIG. 17, in the second embodiment of the present invention, instead of the file read / write means 2 on the application execution computer A in the first embodiment, the automatic extension type file read / write means 22 is used. Further, instead of the address conversion means 11 on the secondary storage control computer C, an extended type address conversion means 21 is provided, and further, a file size expansion means 7 operating on the file management computer B is provided.

Referring to FIG. 18, in the processing of the automatic extension type file read / write means 22, in the access type judging step 201, the kind of requested access is judged.

When the access type is write, in the file size comparison step 202, the size relationship between the file size in the file access management area designated by the application program and the designated writing area is determined, and the writing area is the file. If it is within the size range, the process jumps to the read / write block extraction step 207.

In the file size comparison step 202, when the writing area includes an area equal to or larger than the file size, in the file size expansion request step 203, the file size expansion means 7 on the file management computer B recorded in the file access management area 7 Send processing request to
In the file size expansion response waiting step 204, a response from the file management computer B is waited for, and in the file size updating step 205, the latest file size included in the file size expansion response is stored in the specified file access management area and read / written. Jump to block extraction step 207.

If the access type is read in the access type determination step 201, the specified read range is limited within the file size in the file access management area in the read range restriction step 206, and the read / write block extraction step 207 is executed. move on.

In the read / write block extraction step 207, one block is extracted from the read / write range, and in the read / write remaining determination step 208, it is determined whether or not the read / write target block has been extracted.

When there is a read / write target block, the partial file determining means 3 is called with the in-file offset of the first byte of the extraction block, the partial file to which the extracted block belongs and the offset in the partial file are calculated, and the address is determined. Conversion request transmission step 209
In, part extended address converting means 2 1 on the secondary storage control computer C files partial file determined in decision means 3 manages the secondary storage device D belongs portion calculated in partial file determining means 3 The processing request is transmitted together with the in-file offset, the response from the secondary storage control computer C is waited in the address translation response reception waiting step 210, and the partial file determination means 3 calculates in the I / O issue request transmission step 211. the number of the partial files or secondary storage device D, a physical block number contained in the response of the extended address converting means 2 1, the size of the blocks extracted by the read-write block extracting step 207, even if the access type is a write application Determined by the partial file determining means 3 together with the write data passed from the program The secondary storage device D to which the partial file to which the partial file belongs belongs is sent a processing request to the I / O issuing means 12 on the secondary storage control computer C. Waiting for a response from the control computer C, determining the access type in the access type determining step 213, and only when the access type is read, in the read data copying step 214, the read data included in the I / O issue response is set to the application program. To the specified area, the process returns to the read / write block extraction step 207, and the same process is repeated for the next block. Since the connection relationship between the secondary storage device D and the secondary storage control computer C is determined at the time of system configuration,
It shall be recorded in advance on all computers.

In the read / write remaining determination step 208, when it is determined that there is no block to be read / written, the process is terminated and the process returns to the calling source.

Referring to FIG. 19, the file size enlarging means 7 determines in the size comparison step 701 the current file size in the file management area of the designated file,
Compare with the expanded file size specified by the requestor.

If the current file size is smaller than the requested file size, file size change step 702
Then, the enlarged file size designated by the request source is written in the file management area, and in the file size enlargement response transmission step 703, the file size enlargement response is transmitted to the request source together with the latest current file size, and the processing ends.

If the current file size is equal to or larger than the requested file size in the size comparison step 701, the file size is not changed and the process jumps to the file size expansion response transmission step 703.

Note that the file size is cached in the application execution computer A so that only the application execution computer A can judge the process of limiting the read range within the file size at the read access. If this caching is not performed, communication to the file management computer B will occur every read access, and the efficiency will deteriorate.

However, if this caching is performed, the master file size on the responsible file management computer B and the file size cached in each application execution computer A do not match. This mismatch does not occur because only that application program is open when the file is reduced, and occurs when multiple application programs open the same file and any of them perform additional writing. In this case, only the cache value of the application execution computer A and the master value of the file management computer B in which the application program to which the additional writing has been performed operate are expanded, and the cache values of the other application execution computers A remain small.
This inconsistency is also corrected sequentially during the subsequent communication between the application execution computer A and the file management computer B.

The file size enlarging means 7 also plays a role of matching the shifted file size values. That is, the request source determines that the file size needs to be increased based on the old file size value, and the file size increasing means 7
If the file size has already been expanded when you call, you should return the larger of the requested size and the current file size without returning an error. By this control, the file size of the request source is updated, and the additional write to the area is processed as the overwrite write. Further, for this control, the parameter for the file size enlargement means 7 is not the increase size, but
It is important that the file size you want to grow is passed.

Referring to FIG. 20, in the address conversion table index number calculation step 1101, the extended address conversion means 21 divides the specified offset in the partial file by the block size to calculate the block number in the partial file. , Physical block number acquisition step 1
In 102, the block number in the partial file is used as an index to look up the address conversion table FA in the partial file management data F2 to obtain the physical block number recorded therein, and in the physical block allocation state determination step 1103, obtain it. It is determined whether the physical block number is a valid block number in the data block area D4 on the secondary storage device D.

If this block number is not a valid value, it means that no block is assigned to the offset position, and in the address translation mode decision step 1104, the designation of the address translation mode from the request source is decided. .

If the address translation mode is write, in the block securing step 1105, a new block is secured from the data block area D4 of the secondary storage device D to which the designated partial file belongs, and the address translation table entry registration step 1106. Then, the block number of the secured block is written in the position of the index calculated in the address conversion table index number calculation step 1101 in the address conversion table FA of the partial file management data F2, and the address conversion response transmission step 110
In step 7, the address conversion response is transmitted to the request source together with the block number of the secured block, and the process ends.

Address translation mode determination step 1104
If the read address translation mode is specified, a new block is not secured, the process jumps to the address translation response sending step 1107, and the address is sent to the request source together with a code indicating that the block is not yet assigned. The conversion response is transmitted, and the processing ends.

Physical block allocation state determination step 110
If a valid block number is detected in step 3, it indicates that the block of that block number is mapped to that position, and the process jumps to the address translation response transmission step 1107 to address the request source together with the block number. The conversion response is transmitted, and the processing ends.

The reason why a new block is allocated in the write mode is to expand a partial file when a file is written with automatic expansion. On the other hand, when trying to read the unallocated area, it is an error if it is larger than the file size, and if it is within the file size, it means an area that has never been written. It is enough to treat it as an all-zero block. Therefore, in the address conversion process in the read mode, a new block is not allocated and only the unallocated state is notified to the request source.

FIG. 21 shows a time chart of an embodiment of file writing with automatic extension. In the case of a write with automatic extension, in addition to the normal write processing, file size expansion and block allocation for the extension area are required. First, a file write request issued from an application program is passed to the file access control means 1 on the same application execution computer A as the application program, and further to the automatic extension type file read / write means 22 (SA01). Here, first, the file size FS is expanded by the file size expansion means 7 on the file management computer B that holds the file management data F1 of the specified file (S).
A02). The subsequent communication procedure is the same as that of a normal file write. However, in the process of the extended address translation means 21 on each secondary storage control computer C, a situation of write mode and block unallocation occurs, and at this time,
A process of assigning a block and mapping the block to the address conversion table FA is performed. That is, the partial file is expanded.

By this sequence, the file size of the designated file is expanded and the necessary partial file is expanded. That is, even when the information for managing a file is distributed and loaded in one file management computer B and a plurality of secondary storage control computers C, the file can be properly expanded.

Referring to FIG. 22, the third embodiment of the present invention is the same as the first embodiment except that the file reducing means 6 operating on the file management computer B and the secondary storage control computer are used. Partial file reduction means 10 operating on C
It has and.

Referring to FIG. 23, the file reducing means 6
In the file reference number determination step 601, the file reference number of the file management area of the specified identifier is determined.

If the number of file references is greater than 1, there is an application program that has this file open in addition to the application program that issued this reduction request, so a file reduction failure response transmission step 602 indicates a file reduction failure. The file reduction response is transmitted to the request source together with the code, and the processing is completed.

If the file reference number judging means 601 determines that the number of file references is 1 or less, the file open prohibiting step 603 prohibits other application programs from opening this file in the file management area of the specified file. Is added, and in the partial file selection step 604, partial files of sequentially designated files are selected.

When the partial file is selected, in the partial file reduced size calculation step 605, the reduced file size designated by the request source and the number of secondary storage devices D configured in the system, that is, the number of partial files From that, calculate the partial file size after reduction of the partial file as follows.

Block number in file = (reduced size)
÷ (Block size) Offset in block = (Reduced size)% (Block size) Boundary part file number = (Block number in file)%
(Number of secondary storage devices) Border offset in file = {(block number in file) / (number of secondary storage devices)} x (block size)
+ (Offset in block) If selected partial file number <boundary partial file number, partial file size after reduction = {(block number in file) / (number of secondary storage devices)} x (block size) +
(Block size) If the selected partial file number = boundary partial file number, the partial file size after reduction = (offset in the boundary partial file) If the selected partial file number> boundary partial file number, after reduction Partial file size = {(block number in file) / (number of secondary storage devices)} x (block size)

Next, in step 606 of transmitting a partial file reduction request, the partial file size after reduction and the partial file on the secondary storage control computer C that manages the secondary storage device D in which the selected partial file is stored Reduction means 1
0, a process request is sent to 0, and a response from the secondary storage control computer C is waited in the partial file reduction response reception waiting step 607, and when the response arrives, the process returns to the partial file selection step 604 to retrieve the next partial file. elect.

In the partial file selection step 604, after all partial files have been selected once, in the file size update step 608, the file management area of the file for which the reduced file size designated by the request source is designated. And in the file management data write request transmission step 609, the I / O on the secondary storage control computer C to which the secondary storage device D which should store the file management data of the file designated by the request source is connected. A file management data write request is transmitted to the issuing means 12, a response from the secondary storage control computer C is waited in a file management data write response reception wait step 610, and a file open prohibition cancellation step 611 specifies Open permission by other application programs in the file management area of the created file. Mark the, in file reduces the response transmission step 612, and sends the file reduced response to the requesting, the process ends.

Referring to FIG. 24, the partial file reduction means 10 detects the block number of the block mapped to the largest offset from the address conversion table FA of the specified partial file in the final block detection step 1001. In the size comparison step 1002, the specified reduced size and the final block detection step 1
The offsets in the partial files of the blocks detected in 001 are compared.

If the detected block has an offset larger than the reduced size, in the address conversion table entry deletion step 1003, the block number of the last block is erased from the address conversion table FA of the specified partial file, and the block release step is executed. At 1004, the final block itself is registered as an empty block so that it can be reused, and the process returns to the final block detection step 1001 to detect the next final block.

In the size comparison step 1002, if no block is detected in the final block detection step 1001 and if the detected final block is within the specified reduction request size, write partial file management data. In the request sending step 1005, the I / O issuing means 12 is requested to write the changed partial file management data FA, and in the waiting step 1006 for receiving a partial file management data write response, a response is returned from the I / O issuing means 12. Waiting for arrival, the partial file reduction response transmission step 1007 transmits the partial file reduction response to the request source, and the process is completed.

FIG. 25 shows a time chart of an embodiment of file reduction. First, a file reduction request issued from an application program is passed to the file access control means 1 on the same application execution computer A as the application program (ST0).
1). Here, the file management computer B that manages the specified file is selected. Next, a reduction request is sent to the file reduction means 6 on the computer (ST02). Receiving this, the file reducing means 6 selects the secondary storage device D group in which the partial files of the specified file are stored, and controls all the secondary storage devices to which those secondary storage devices D are connected. Partial file reduction means 10 on computer C
A partial file reduction request to (ST03, ST0
5, ST07). Receiving this request, each partial file reducing means 10 reduces the designated partial file to the designated size. By this processing, the partial file management data F2 is changed, so that it is written in the secondary storage device D (ST04, ST06, ST08). Next, the file size FS of the designated file is reduced. Since the file management data F1 is changed, the secondary storage device D in which the file management data F1 was originally stored is selected,
Secondary storage control computer C to which the secondary storage device D is connected
File management data F via the above I / O issuing means 12
Write 1 (ST09 to ST10).

By this sequence, the partial file of the designated file is reduced, and the file management data F1 of this file and the partial file management data F2 of the reduced partial file are updated. That is, even when the information for managing a file is distributed and loaded in one file management computer B and a plurality of secondary storage control computers C, the file can be properly reduced.

Referring to FIG. 26, the fourth embodiment of the present invention is provided with a file parallel reading / writing means 13 instead of the file reading / writing means 2 on the application execution computer A in the first embodiment. ing.

Referring to FIG. 27, the process of the file parallel read / write means 13 is the read / write range limiting step 130.
In 6, the specified read / write range is limited to within the file size in the file access management area, and in the read / write block extraction step 1307, one block is extracted from the read / write range, and the read / write remaining determination step 1308.
Then, it is determined whether or not the read / write target block has been extracted.

When there is a read / write target block, the partial file determining unit 3 is called with the in-file offset of the first byte of the extraction block, the partial file to which the extracted block belongs and the offset in the partial file are calculated, and the address is determined. Conversion request transmission step 1309
In the partial file calculated by the partial file determination means 3, the address conversion means 11 on the secondary storage control computer C managing the secondary storage device D to which the partial file determined by the partial file determination means 3 belongs Along with the offset
Send processing request and extract read / write block Step 13
Returning to 07, the next read / write block is extracted.

If it is determined in the read / write remaining determination step 1308 that the blocks to be read / written have been exhausted, the process proceeds to a response reception waiting step 1310, waits for a response of some processing request to return, and in the response type determination step 1311, returns. Determine the response message that you received.

When the response message is a response to the address translation request issued in the address translation request sending step 1309, the physical block number included in the response is sent in the disk I / O issue request sending step 1315.
If the access type is write, write data is added to the I / O on the secondary storage control computer C that manages the secondary storage device D determined by the partial file determination means 3.
The issuing means 12 is requested to issue an I / O request, and the process returns to the response reception waiting step 1310 to wait for the next response.

In the response type determination step 1311, the response message is the disk I / O issue request transmission step 1
In the case of a response to the request issued at 315, the process proceeds to the access type determination step 1312, and the original access type when this means is called is determined.

When the original access type is read, in the read data copy step 1313, the data read from the secondary storage device D included in the response is copied to the buffer designated by the request source, and the read / write ends. In a decision step 1314, it is determined whether or not the input / output processing of all the read / write target blocks has been completed. If there are still blocks under processing, the process returns to the response reception waiting step 1310 to wait for the next response. When all the processing is completed, the processing is ended and the process returns to the calling source.

In the access type judgment step 1312, if the caller's request is a write, the read data copy step 1313 is skipped, and the read / write end judgment step 1314 is jumped to.

The file parallel read / write means 13 is characterized in that it simultaneously issues a plurality of address conversion requests and I / O issuance requests to other computers in the process of performing file read and file write. By this control, these plural processes are executed in parallel.

Next, the operation of the example of the fourth embodiment of the present invention shown in FIG. 26 will be described in detail.

FIG. 28 shows a time chart of an embodiment of file parallel reading. First, the file read request issued from the application program is passed to the file access control means 1 on the same application execution computer A as the application program, and further to the file parallel reading / writing means 13 (PR01).
Here, the read range of the specified file is sliced in block units. Then, for each block, the partial file determining means 3 is used to calculate the partial file to which the block belongs and the offset within the partial file, and the address conversion on the secondary storage control computer C that manages the determined partial file. The means 11 calculates the position of the block on the secondary storage device D from the offset in the partial file (PR02, PR03, PR04), and the I / O on the secondary storage control computer C which also manages the partial file. The issuing unit 12 reads the target block from the secondary storage device D (PR05, PR06, PR).
07, PR08, PR09, PR10).

FIG. 29 shows a time chart of an embodiment of file parallel writing. The file write is different from the file read in the moving direction of the user data. First, a file write request issued from an application program is sent to the file access control means 1 on the same application execution computer A as the application program, and further, the file parallel reading / writing means 1
3 (PW01). Here, the writing range of the specified file is sliced in block units. Then, for each block, the partial file determining means 3 is used to calculate the partial file to which the block belongs and the offset within the partial file, and the address conversion on the secondary storage control computer C that manages the determined partial file. The means 11 determines the position of the block on the secondary storage device D from the offset in the partial file (PW02, PW02).
W03, PW04), the user data is written in the target block of the secondary storage device D by the I / O issuing means 12 on the secondary storage control computer C that also manages the partial file (PW05, PW06, PW07, PW
08, PW09, PW10).

FIG. 15 and FIG. 16 showing the read and write operations in the first embodiment of the present invention and FIG. 28 above.
Comparing FIG. 29, by introducing the file parallel reading / writing means 13, reading and writing of a plurality of blocks for files distributed and stored in the plurality of secondary storage devices D are processed in parallel, and the processing time is shortened. I understand.

Referring to FIG. 30, the fifth embodiment of the present invention is the same as the second embodiment except that instead of the automatic extension type file read / write means 22 on the application execution computer A, an automatic extension file parallel is used. The reading / writing means 23 is provided.

Referring to FIG. 31, in the process of the automatic extension type file parallel read / write means 23, in the access type judging step 1301, the kind of requested access is judged.

If the access type is write, in the file size comparison step 1302, the size relationship between the file size in the file access management area designated by the application program and the designated writing area is determined, and the writing area is the file. If it is within the size range, the process jumps to the read / write block extraction step 1307.

In the file size comparison step 1302, if the write area includes an area equal to or larger than the file size, a file size enlargement request transmission step 1303.
Then, the processing request is transmitted to the file size enlargement means 7 on the file management computer B recorded in the file access management area, and the file size enlargement response reception waiting step 13 is performed.
At 04, the response from the file management computer B is waited for, and at the file size update step 1305, the latest file size included in the file size expansion response is stored in the specified file access management area, and the process jumps to the read / write block extraction step 1307. To do.

In the access type judgment step 1301, if the access type is read, in the read range restriction step 1306, the specified read range is restricted within the file size in the file access management area,
The process proceeds to the read / write block extraction step 1307.

In the read / write block extraction step 1307, one block is extracted from the read / write range, and in the read / write remaining determination step 1308, it is determined whether or not the read / write target block has been extracted.

When there is a read / write target block, the partial file determining means 3 is called with the in-file offset of the first byte of the extraction block, the partial file to which the extracted block belongs and the offset in the partial file are calculated, and the address is determined. Conversion request transmission step 1309
Then, the partial file determination means 3 calculates in the extended address conversion means 21 on the secondary storage control computer C that manages the secondary storage device D to which the partial file determined by the partial file determination means 3 belongs. A processing request is transmitted together with the offset within the partial file, and the process returns to the read / write block extraction step 1307 to extract the next read / write block.

In the read / write remaining determination step 1308, when it is determined that there is no block to be read / written, the process proceeds to a response reception waiting step 1310, waits for a response of some processing request to return, and a response type determination step 13
At 11, the returned response message is determined.

If the response message is a response to the address translation request issued in the address translation request sending step 1309, the physical block number and access type included in the response are write in the disk I / O issue request sending step 1315. I / O on the secondary storage control computer C which manages the secondary storage device D determined by the partial file determining means 3 with the write data.
The issuing means 12 is requested to issue an I / O request, and the process returns to the response reception waiting step 1310 to wait for the next response.

In the response type determination step 1311, the response message is the disk I / O issue request transmission step 1
In the case of a response to the request issued at 315, the process proceeds to the access type determination step 1312, and the original access type when this means is called is determined.

If the original access type is read, in the read data copying step 1313, the data read from the secondary storage device D included in the response is copied to the buffer designated by the request source, and the reading / writing ends. In a decision step 1314, it is determined whether or not the input / output processing of all the read / write target blocks has been completed. If there are still blocks under processing, the process returns to the response reception waiting step 1310 to wait for the next response. When all the processing is completed, the processing is ended and the process returns to the calling source.

In the access type judgment step 1312, if the caller's request is a write, the read data copying step 1313 is skipped and the read / write end judgment step 1314 is performed.

The automatic extension type file parallel read / write means 23 is characterized in that it simultaneously issues a plurality of requests to other computers in the process of performing file write accompanied by automatic extension in addition to file read and file write. .
By this control, these plural processes are executed in parallel.

FIG. 32 shows a time chart of an embodiment of file writing with automatic extension. For additional lights,
In addition to normal write processing, file size expansion and block allocation for additional areas are required. First, the file write request issued from the application program is the file access control means 1 on the same application execution computer A as the application program, and further the automatic expansion type file parallel read / write means 2
3 (PA01). Here, first, the file size enlargement means 7 on the file management computer B holding the file management data F1 of the designated file.
The file size FS is expanded by (PA0
2). The subsequent communication procedure is the same as the file write in the second embodiment. However, when a write mode and block unallocation occur during the processing of the extended address translation unit 21 on each secondary storage control computer C, a block is assigned and the block is mapped to the address translation table FA. Processing is performed. That is, the file is automatically expanded.

Comparing FIG. 21 showing the operation of the file write with automatic extension in the second embodiment of the present invention with FIG. 32, by introducing the automatic extension type file parallel reading / writing means 23, the partial file It can be seen that the file write including the extension of is processed in parallel and the processing time is shortened.

Referring to FIG. 33, the sixth embodiment of the present invention comprises a file parallel reduction means 16 instead of the file reduction means 6 on the file management computer B in the third embodiment. ing.

Referring to FIG. 34, the file parallel reduction unit 16 determines the number of file references in the file management area of the designated identifier in the file reference number determination step 1601.

If the number of file references is greater than 1, there is an application program that has this file open in addition to the application program that issued this reduction request, so a file reduction failure response transmission step 1602 indicates a file reduction failure. Along with the code
The file reduction response is transmitted to the request source, and the processing ends.

In the file reference number determination step 1601,
If the number of file references is 1 or less, a file open prohibition step 1603 adds a mark to the file management area of the specified file to prohibit other application programs from opening this file, and a partial file selection step 1604 , Select the partial files of the files specified in order.

When the partial files are selected, in the partial file reduced size calculation step 1605, the reduced file size designated by the request source and the number of secondary storage devices D configured in the system, that is, the number of partial files From that, calculate the partial file size after reduction of the partial file as follows.

Block number in file = (reduced size)
÷ (Block size) Offset in block = (Reduced size)% (Block size) Boundary part file number = (Block number in file)%
(Number of secondary storage devices) Border offset in file = {(block number in file) / (number of secondary storage devices)} x (block size)
+ (Offset in block) If selected partial file number <boundary partial file number, partial file size after reduction = {(block number in file) / (number of secondary storage devices)} x (block size) +
(Block size) If the selected partial file number = boundary partial file number, the partial file size after reduction = (offset in the boundary partial file) If the selected partial file number> boundary partial file number, after reduction Partial file size = {(block number in file) / (number of secondary storage devices)} x (block size)

Next, at the partial file reduction request transmission step 1606, the partial file size after reduction and the partial file on the secondary storage control computer C that manages the secondary storage device D in which the selected partial file is stored A processing request is transmitted to the reducing means 10, the process returns to the partial file selection step 1604, and the next partial file is selected.

When all the partial files have been selected in the partial file selection step 1604, the process proceeds to the partial file reduction response reception waiting step 1607 and waits for the response of the request issued in the partial file reduction request transmission step 1606. When the response is returned, in the partial file end determination step 1608, it is determined whether or not the response to the reduction request for all the partial files selected in the partial file selection step 1604 is returned, and there is a response not yet returned. In this case, the process returns to the partial file reduction response reception waiting step 1607 to wait for the next response.

Partial file end determination step 1608
And if it is confirmed that all responses are returned,
In the file size update step 1609, the reduced file size specified by the request source is set in the file management area of the specified file, and in the file management data write request transmission step 1610, the file specified by the request source is changed. A file management data write request is transmitted and a file management data write response is received to the I / O issuing means 12 on the secondary storage control computer C to which the secondary storage device D which should store the file management data is connected. Waiting step 1
At 611, the response from the secondary storage control computer C is waited for, at the file open prohibition releasing step 1612, the file management area of the designated file is marked as open permission by another application program, and at the file reduction response transmitting step 1613, The file reduction response is transmitted to the request source, and the processing is completed.

Next, the operation of the example of the sixth embodiment of the present invention shown in FIG. 33 will be described in detail.

FIG. 35 shows a time chart of an embodiment of file parallel reduction. First, a file reduction request issued from an application program is passed to the file access control means 1 on the same application execution computer A as the application program (PT0.
1). Here, the file management computer B that manages the specified file is selected. Next, a reduction request is sent to the file parallel reduction means 16 on the computer (PT0
2). Upon receiving this, the file parallel reduction unit 16 selects the secondary storage device D group in which the partial files of the specified file are stored, and all the secondary storage devices to which those secondary storage devices D are connected. Partial file reduction requests are continuously sent to the partial file reduction means 10 on the control computer C (PT03, PT05, PT07). Upon receiving this request, the partial file reducing means 10 reduces the designated partial file to the designated size. By this processing, the partial file management data F2 is changed, so this is written to the secondary storage device D (PT04, PT06, PT0.
8). Next, the file size FS of the specified file
Shrink. Since the file management data F1 is changed, the secondary storage device D in which the file management data F1 was originally stored is selected, and the I / O on the secondary storage control computer C to which the secondary storage device D is connected is selected. Via the issuing means 12,
Write the file management data F1 (PT09 to PT1
0).

By this sequence, the partial file of the designated file is reduced, and the file management data F1 of this file and the partial file management data F2 of the reduced partial file are updated.

Comparing FIG. 25 showing the file reduction operation in the third embodiment of the present invention with FIG. 35 described above,
By introducing the file parallel reduction means 16, the change of the address conversion table FA in the partial file management data F2 distributedly loaded to the plurality of secondary storage control computers C of the specified file is processed in parallel, and this It can be seen that the processing time is shortened.

Referring to FIG. 36, the seventh embodiment of the present invention is provided with a file parallel opening means 14 in place of the file opening means 4 on the file management computer B in the first embodiment. ing.

Referring to FIG. 37, the file parallel opening means 14 executes the designated file search step 1401.
The file management area of the file specified by the request source is searched, and in open determination step 1402, it is judged whether the file management area of the specified file has already been registered.

If the file management area of the specified file is not registered, it means that no other application program has opened the corresponding file, and in the file management area securing step 1403, a new file management area is created. And the file reference count (the number of application programs that have opened the file) is set to 1 in the file reference count initialization step 1404.
The secured file management area is initialized, and the secondary storage managing the secondary storage device D in which the file management data F1 of the designated file is stored in the file management data read request transmission step 1405. A request for reading the corresponding file management data F1 is transmitted to the I / O issuing means 12 on the control computer C, and in step 1406 waiting for reception of a file management data read response, the response from the secondary storage control computer C is awaited. In the file management data storage step 1407, the read file management data F1 is stored in the file management area secured in the file management area securing step 1403, and in the partial file selection step 1408, the partial files of the designated file are sequentially ordered. Select one by one.

When a partial file is selected, in a partial file open request transmission step 1409, a partial file open means on the secondary storage control computer C that manages the secondary storage device D in which the selected partial file is stored. 8, the processing request is transmitted to step 8, and the process immediately returns to the partial file selection step 1408 to select the next partial file.

In the partial file selection step 1408, if all the partial files have already been selected once and the partial files have been selected, the process proceeds to the partial file open response reception waiting step 1410 and the partial file open request transmission step. When the response of the processing request requested in 1409 is waited for and even one response is returned, in the next partial file end determination step 1411, all open requests for all the partial files selected in the partial file selection step 1408 are received. Determine if you're back. If there is a response that has not been returned yet, the process returns to the partial file open response reception waiting 1410 and waits for the next response. When all the responses are returned, a file open response transmission step 1415
Then, the response is returned to the request source, and the processing is completed.

If the file management area of the specified file is found in the open status judgment step 1402, it means that another application program has already opened the file, and in the open prohibition status judgment step 1412, By referring to the detected file management area, it is determined whether opening from other clients is prohibited.

If the open is prohibited, the file open failure response transmission step 1413 transmits the file open response to the request source together with the code indicating the open failure, and the processing is terminated. In the open prohibition state determination step 1412, if the specified file is permitted to open, in the file reference number increment step 1414, 1 is added to the detected file reference number of the file management area, and the file open response transmission step 1415, the designated file search step 1401
A file open response is transmitted to the request source together with part of the file management data such as the identifier of the file management area detected in step 1 and the file size recorded in that area, and the processing ends.

The file parallel open means 14 is characterized in that it simultaneously issues a plurality of processes for requesting another computer in the process of executing a file open. By this control, these plural processes are executed in parallel.

Next, the operation of the example of the seventh embodiment of the present invention shown in FIG. 36 will be described in detail.

FIG. 38 shows a time chart of an embodiment of file opening. First, a file open request issued from an application program is passed to the file access control means 1 on the same application execution computer A as the application program (P
O01). Here, the file management computer B that manages the specified file is selected. Next, an open request is sent to the file parallel open means 14 on the computer (PO02). Upon receiving this, the file parallel opening unit 14 receives the file management data F of the designated file.
1 is stored in the secondary storage device D, and the I / O on the secondary storage control computer C to which the secondary storage device D is connected is selected.
The file management data F1 is read through the O issuing means 12 path (PO03 to PO04). Next, the secondary storage device D group in which the partial files of the specified file are stored is selected, and the partial file opening means on all the secondary storage control computers C to which those secondary storage devices D are connected. 8
A partial file open request continuously to (PO0
5, PO07, PO09). Upon receiving this request, the partial file opening means 8 reads the partial file management data F2 of the specified partial file to the secondary storage control computer C. By this sequence, the file management data F1 of the specified file is stored in one file management computer B.
Then, the partial file management data F2 of the partial files forming the specified file are read to the entire secondary storage control computer C, and the file access is ready.

Comparing FIG. 13 showing the file opening operation in the first embodiment of the present invention with FIG. 38, by introducing the file parallel opening means 14, a plurality of secondary storages of designated files can be carried out. It can be seen that the reading of the partial file management data F2 distributedly stored in the device D is processed in parallel, and this processing time is shortened.

Referring to FIG. 39, the eighth embodiment of the present invention comprises a file parallel closing means 15 instead of the file closing means 5 on the file management computer B in the first embodiment. ing.

Referring to FIG. 40, the file parallel close means 15 determines the file reference count decrement step 1
In step 501, 1 is subtracted from the number of file references in the file management area of the specified identifier, and in the file reference number determination step 1502, the number of file references after subtraction is determined.

If the number of file references is zero, it means that there is no other application program that has the file open, and the partial file selection step 1
At 503, partial files of the designated file are selected in order.

While there is a target partial file, a partial file close request transmission step 1504 closes a partial file on the secondary storage control computer C that manages the secondary storage device D in which the selected partial file is stored. A processing request is sent to the means 9, and the process immediately returns to the partial file selection step 1503 to select the next partial file.

In the partial file selection step 1503, when all the partial files have been selected once, the process proceeds to the partial file close response reception waiting step 1505 and the partial file close request transmission step 150.
Wait for the response to the request issued in 4. If even one response is returned, the partial file end judgment step 150
In step 6, it is determined whether or not the close responses of all the partial files selected in the partial file selection step 1503 have been returned.
Returning to the step 1505 of waiting for reception of a partial file close response, the next response is waited for.

Partial file end determination step 1506
Then, when it is confirmed that all the responses have been returned, the process proceeds to the file management data write request transmission step 1507, and the secondary storage control computer C that manages the secondary storage device D in which the file management data F1 of the file should be stored. A file management data write request is sent to the above I / O issuing means 12, and a file management data write response reception waiting step 1508 waits for a response from the secondary storage control computer C and releases the file management area. In step 1509, the file management area of the designated file is released, and in file close response transmission step 1510, a file close response is transmitted to the request source, and the processing ends.

At the file reference number determination step 1502,
If the number of file references is not zero, there is an application program that has the file still open, and therefore the close processing is not performed, and the process jumps to the file close response transmission step 1510.

The file parallel close means 15 is characterized in that, in the process of executing the file close, a plurality of processes for requesting another computer are issued at the same time. By this control, these plural processes are executed in parallel.

Next, the operation of the example of the eighth embodiment of the present invention shown in FIG. 39 will be described in detail.

FIG. 41 shows a time chart of an embodiment of file parallel close. First, the file close request issued from the application program is passed to the file access control means 1 on the same application execution computer A as the application program (PC01). Here, the file management computer B that manages the specified file is selected. Next, a close request is sent to the file parallel close means 15 on the computer (PC02). Receiving this, the file parallel close means 15 continuously issues a partial file close request to the partial file close means 9 on the secondary storage control computer C which holds the partial file management data F2 of the partial file of the designated file. Send to (PC03, PC05,
PC07). Partial file closing means 9 receiving this
Is the partial file management data F that may have been updated.
2 is written in the secondary storage device D (PC04, PC06,
PC08). Finally, the file parallel closing means 15
Is the file management data F stored in your computer.
1 is the secondary storage device D in which the data was originally stored.
Is written to the secondary storage device D via the I / O issuing means 12 on the secondary storage control computer C to which is connected (PC09
~ PC10). By this sequence, the file management data F1 of the specified file and the partial file management data F2 distributed and loaded in the plurality of secondary storage control computers C are surely written in the predetermined location of the secondary storage device D. Will be returned.

Comparing FIG. 14 showing the file closing operation in the first embodiment of the present invention with FIG. 41 described above, by introducing the file parallel closing means 15, a plurality of partial files of the designated file can be obtained. It can be seen that the writing of the management data F2 to the secondary storage device D is processed in parallel, and this processing time is shortened.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments and various other additions and modifications can be made. For example, the second to second aspects of the present invention
The eighth inventions can be used in combination with each other.

Although the computers for executing the respective means are classified into the application execution computer A, the file management computer B, and the secondary storage control computer C for explanation, essentially,
The means that operates on the application execution computer A operates on the same computer as the application program, and the means that operates on the secondary storage control computer C operates on the computer to which the secondary storage device D to be managed is connected. It is important to operate, and any computer that operates other means may be used. Therefore, other means may operate on the application execution computer A or the secondary storage control computer C. However, in order to execute these arrangements, a communication function between the application executing computers A and a communication function between the secondary storage control computers C are required. That is, communication between arbitrary secondary storage control computers is enabled, and means for operating some or all of the file management computers is configured to operate on the secondary storage control computers, or between any application execution computers. A means that enables communication and operates on part or all of the file management computer can be configured to operate on the application execution computer. Furthermore, it is possible that the application program runs on one computer and the secondary storage device D is connected. In this case as well, the present invention can be used by operating both means on the computer according to the above-mentioned principle. That is,
Communication between arbitrary application execution computers can be performed, and a part or all of the means for operating on the file management computer can be configured to operate on the application execution computer. That is, the three types of computer classification in the present invention are not so definite, and merely represent the division of each means having the same character.

[0190]

As described above, according to the first file data distributed storage system of the present invention, the data storage areas are distributed and arranged in the plurality of secondary storage devices connected to the plurality of computers. It is possible to define a logical file and perform general operations on the logical file such as opening, closing, reading, and writing on the file. As a result, first, it becomes possible to provide a file having a larger capacity than the conventional file system.

Further, according to the present invention, not only the secondary storage device but also the CPU processing capability for file access control, the data transfer capability by the main storage bus, the I / O bus, etc. are at appropriate positions. Accurate file access is realized as a total of these by using hardware resources in pieces. Therefore, even if many application programs in the system access files almost at the same time, fragmentary processing of each is alternately processed on each hardware resource, and macroscopically, these file accesses proceed simultaneously. Is processed in. This means that in addition to the secondary storage device, the load of all hardware resources such as the CPU, main storage bus, and I / O bus is distributed. In addition, each file is distributed and stored in the secondary storage device connected to a plurality of computers, and the entire system operates even if access is concentrated on one file. This makes it possible to obtain file access throughput according to the scale of the platform used. In other words, there is no bias in resource allocation, and by increasing the size of the platform, it is possible to obtain a large file access throughput.

Further, by the file automatic extension function at the time of writing and the file reduction function included in the second and third file data distributed storage systems of the present invention, the files are distributed to a plurality of secondary storage devices connected to a plurality of computers. By doing so, it is possible to expand and contract a file whose storage area is secured. With these functions, it is possible to secure a secondary storage area as needed, and to set the capacity distribution to multiple files configured in the system in various ways. Higher file management becomes possible.

Furthermore, by the parallel read / write, parallel automatic expansion, and parallel reduction functions of the fourth, fifth, and sixth file data distribution storage methods of the present invention, one read or write with a size larger than a certain size, and even a file. The process of expanding and contracting is faster. In particular, the more widely distributed the files are stored, the higher the parallelism of these processes can be and the higher the speed becomes. That is, depending on the scale of the platform, individual file access to some extent becomes faster.

Furthermore, the parallel open and parallel close functions of the seventh and eighth file data storage methods of the present invention also speed up preparation work for file access.
Generally, in a situation where the file management data is distributed to a plurality of secondary storage devices, the opening and closing operations take time in proportion to the distribution range. By utilizing the parallel open and parallel close of the present invention, it is possible to prevent the processing time from increasing so much even if the distribution range is expanded.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a file data distributed storage system of the present invention.

FIG. 2 is a data layout diagram on a secondary storage device.

FIG. 3 is a structural diagram of file management data.

FIG. 4 is a flow chart of processing of a file access control means.

FIG. 5 is a flow chart of processing of a file reading / writing means.

FIG. 6 is a flow chart of processing of a partial file determining unit.

FIG. 7 is a flow chart of processing of a file opening unit.

FIG. 8 is a flow chart of processing of file closing means.

FIG. 9 is a flowchart of a process of a partial file opening unit.

FIG. 10 is a flow chart of processing of a partial file closing means.

FIG. 11 is a flow chart of processing of an address conversion unit.

FIG. 12 is a flowchart of the processing of the I / O issuing means.

FIG. 13 is a communication sequence diagram of file open processing.

FIG. 14 is a communication sequence diagram of file close processing.

FIG. 15 is a communication sequence diagram of file read processing.

FIG. 16 is a communication sequence diagram of file write processing.

FIG. 17 is a configuration diagram of a second file data distributed storage system of the present invention.

FIG. 18 is a flow chart of processing of an automatic extension type file read / write means.

FIG. 19 is a flowchart of the process of a file size expanding unit.

FIG. 20 is a flowchart of the processing of the extended address conversion unit.

FIG. 21 is a communication sequence diagram of a file write process with automatic extension.

FIG. 22 is a configuration diagram of a third file data distributed storage system of the present invention.

FIG. 23 is a flowchart of the process of a file reducing unit.

FIG. 24 is a flowchart of the process of the partial file reduction unit.

FIG. 25 is a communication sequence diagram of file reduction processing.

FIG. 26 is a configuration diagram of a fourth file data distributed storage system of the present invention.

FIG. 27 is a flowchart of the processing of the file parallel reading / writing means.

FIG. 28 is a communication sequence diagram of file parallel read processing. ,

FIG. 29 is a communication sequence diagram of file parallel write processing.

FIG. 30 is a configuration diagram of a fifth file data distributed storage system of the present invention.

FIG. 31 is a flow chart of processing of an automatic extension type file parallel reading / writing means.

FIG. 32 is a communication sequence diagram of file parallel write processing with automatic extension.

FIG. 33 is a configuration diagram of a sixth file data distributed storage system of the present invention.

FIG. 34 is a flow chart of processing of file parallel reduction means.

FIG. 35 is a communication sequence diagram of file parallel reduction processing.

[Fig. 36] Fig. 36 is a configuration diagram of a seventh file data distributed storage system of the present invention.

FIG. 37 is a flow chart of processing of file parallel opening means.

FIG. 38 is a communication sequence diagram of file parallel open processing.

FIG. 39 is a configuration diagram of an eighth file data distributed storage system of the present invention.

FIG. 40 is a flow chart of processing of file parallel closing means.

FIG. 41 is a communication sequence diagram of file parallel close processing.

[Explanation of symbols]

A ... Application execution computer B ... File management computer C ... Secondary storage control computer D ... Secondary storage device N ... Communication mechanism 1 ... File access control means 2 ... File reading / writing means 3 ... Partial file determination means 4 ... File open means 5 ... File closing means 6 ... File reduction means 7 ... File size expansion means 8 ... Partial file opening means 9 ... Partial file closing means 10 ... Partial file reduction means 11 ... Address conversion means 12 ... I / O issuing means 13 ... File parallel reading / writing means 14 ... Automatic extension type file read / write means 15 ... Automatic expansion type file parallel reading / writing means 21 ... Extended type address translation means 22 ... Automatic extension type file read / write means 23 ... Automatic extension type file parallel reading / writing means

Continuation of the front page (56) References JP-A-3-92942 (JP, A) Masuda et al., "RAID file system realized in distributed environment", 51st Annual Conference of Information Processing Society of Japan (second half of 1995) Proceedings Collection (4), September 22, 1995, p. 4-25 to 4-26 (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06F 12/00

Claims (11)

(57) [Claims] 1. A plurality of secondary storage control computer which one or more secondary storage devices are connected for recording the file data and file management data, can communicate with any of the secondary storage control computer One or more file management computers, one or more application execution computers that can communicate with any of the above secondary storage control computers and any of the above file management computers, and communication between these computers is possible. A method for distributed storage of file data in a multi-node computer system having a communication mechanism group for controlling a partial file configured as a set of storage areas in one secondary storage device and managing the partial file. Partial file management data that is data and it
A set of corresponding storage areas is stored in the same secondary storage device, and one logical file is configured as a set of the partial files distributed and stored in the plurality of secondary storage devices, In addition, the file that is the management data for each file
File management data is stored in any of the secondary storage devices, operates on the secondary storage control computer, and issues an I / O request to the secondary storage device connected to the secondary storage control computer. I
/ A O issuing unit, operating in the secondary storage control calculation machine reads the specified part file management data on the attached secondary storage device to the secondary storage control computer, access to the partial file a partial file open means for allowing the secondary storage control operates on a computer, changed parts file management of the designated partial file on the attached secondary storage device to the secondary storage control computer Partial file closing means for writing data and ending access to the partial file, and a designated partial file on the secondary storage device operating on the secondary storage control computer and connected to the secondary storage control computer address conversion means for converting the partial file offset within the physical address on the secondary storage device, and operates in the file management computer machine, before by the I / O issuing unit File management data
Read from the secondary storage device for storing the file tube <br/> the by the I / O issuing unit using physical data portions off <br/> § i le management data the partial file management read from the secondary storage device for storing data, and a file open means to open all the partial file of the file by the partial file open means with the partial file management data, and operates in the file management computer machine, wherein the partial file closing means closes all partial file of the file, by the I / O issuing unit, a file close means to write the modified the file <br/> Lee le management data in the secondary storage device, the application A part that runs on the execution computer and calculates the corresponding partial file and the offset within the partial file from the offset of the specified file File conversion means and an application execution computer, which selects a partial file by the partial file determination means, and the address conversion on the secondary storage control computer to which the secondary storage device to which the selected partial file belongs is connected File read / write means for calculating a physical address by means, and reading / writing data by the I / O issuing means on the secondary storage control computer in charge of the partial file, and operating on the application execution computer and designated by the application program The file access request is opened, if the file access request is open, send a file open processing request to the file opening means of the file management computer, if the file access request is closed, the file management computer of the File black Over's means sends the file close processing request, if the file access request is a read or write, the feed to the file access request to the file reading and writing means, before Symbol file access request is opened, the claw's, except
In the case of rie de or write in the file access request to the file management computer sends to the application execution computers, characterized in that it comprises a said file access request file access control means to return to the application program waits for a response File data distributed storage method. 2. An offset within a partial file of a specified partial file on a secondary storage device operating on a secondary storage control computer instead of the address conversion means and connected to the secondary storage control computer. A means for converting to a physical address on the secondary storage device, which comprises an extended type address conversion means having a function of allocating an area at the time of address conversion of an unallocated storage area, operates on the file management computer, and specifies A file size expanding means for changing the specified file to a specified file size, and operating on an application execution computer instead of the file reading / writing means, selecting a partial file by the partial file determining means, and selecting The secondary storage control computer to which the secondary storage device to which the selected partial file belongs is connected. Means for calculating a physical address by the address conversion means, and for reading and writing data by the I / O issuing means on the secondary storage control computer in charge of the partial file, wherein the file size expanding means in the case of additional writing 2. The file data distributed storage system according to claim 1, further comprising an automatic expansion type file read / write means for expanding the file size in advance and calculating a physical address by the expansion type address conversion means. 3. A partial file reducing means which operates on a secondary storage control computer and reduces a specified partial file to a specified size, and a partial file reducing means which operates on a file management computer and is specified by the partial file reducing means. A file reducing means for reducing each partial file of the file, the file access control means,
Access requests other than open, close and file shrink.
File or write request, the file access request
The file management computer, application execution calculation
Sent to the machine, the file if the access request is a file reduction request
The file management means of the file management computer
The file data distribution case according to claim 1 , further comprising an operation of sending a file access request.
Delivery method. 4. A secondary storage operating on an application execution computer instead of the file reading / writing means, selecting a partial file by the partial file determining means, and connected to a secondary storage device to which the selected partial file belongs. A means for calculating a physical address by the address conversion means on the storage control computer, and reading and writing data by the I / O issuing means on the secondary storage control computer that is in charge of the partial file. A request is divided into a plurality of small read / write requests, and a file parallel read / write means for simultaneously performing physical address calculation by the address conversion means and read / write by the I / O issue means for a plurality of read / write fragments is included. The file data distributed storage method according to claim 1. 5. A partial file of a specified partial file on a secondary storage device that operates on a secondary storage control computer and is connected to the secondary storage control computer, instead of the automatic extension type file read / write means. It has the function of converting the internal offset to a physical address on the secondary storage device and allocating the area when converting the address of the unallocated storage area,
Further, the designated read / write request is divided into a plurality of small read / write requests, and physical addresses are calculated by the extended address conversion means for a plurality of read / write fragments, new secondary storage area allocation and secondary processing by the I / O issuing means are performed. 3. The file data distributed storage system according to claim 2, further comprising an automatic expansion type file parallel reading / writing means for performing reading / writing of the storage device simultaneously. 6. A means for operating on a file management computer, instead of the file reducing means, for reducing each partial file of the file designated by the partial file reducing means, wherein a plurality of the partial file reducing means are provided. 4. The file data distributed storage system according to claim 3, further comprising file parallel reduction means for simultaneously activating. 7. Instead of the file open means operates in the file management computer machine reads the file management data from the secondary storage device for storing the file management data by the I / O issuing unit, wherein the Phi It said read the partial file <br/> Lee le management data by I / O issuing unit from the secondary storage device for storing the partial file management data by using the Le management <br/> data, the partial file tube 2. A means for opening all partial files of a file by means of said partial file opening means using logical data, comprising a file parallel opening means for simultaneously activating a plurality of said partial file opening means. Distributed file data storage method described. 8. Instead of the file close means operates in the file management computer machine, said portion to close all partial file of the file by the file closing means, the file that has been changed by said I / O issuing unit
And means to write the management data in the secondary storage device, the file data distributed storage system according to claim 1, characterized in that it comprises a file parallel closing means for activating a plurality of the partial file closing means simultaneously. 9. Communication between arbitrary secondary storage control computers is enabled, said file opening means is operated on said secondary storage control computer, and said file closing means is operated on said file management computer, or the file closing means is operated at the secondary storage control calculation machine, the file claim 1 and 2 open means is configured to operate in said file management computer machine,
The file data distributed storage method described in 3, 4, 5, 6, 7 or 8. 10. Communication between arbitrary application execution computers is made possible, said file open means is operated on said application execution computer, said file close means is operated on said file management computer, or said file close means to operate in the application execution calculation machine, according to claim 1, the file open means is configured to operate in said file management computer machine,
File data distributed storage method described in 2, 3, 4, 5, 6, 7 or 8. 11. A dual-purpose computer capable of executing an application program and connected to a secondary storage device, enables communication between an arbitrary dual-purpose computer and any other file management computer, and on the dual-purpose computer, 7. A configuration in which the partial file determination means, the file read / write means, the file access control means, the partial file open means, the partial file close means and the address conversion means are operated. , 7 or 8 file data distributed storage method.