JP3500972B2 - Logical File Management System in Disk Sharing Cluster System - Google Patents

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a plurality of computers,
In a disk-sharing cluster system consisting of a computer connection network connecting each computer, one or more disks storing logical files, and a disk connection network that allows each computer to directly access each disk, The present invention relates to a management system for logical files stored on each disk.

[0002]

2. Description of the Related Art Conventionally, in a configuration in which a plurality of computers are connected to a computer connection network and a disk is connected to each computer, as an access method for a logical file on a disk connected from one computer to another computer,
NFS (NetworkFileSystem) is used (reference: "NFS &NIS" ASCII publication, "NFS: Network FileSystem").
System Protocol specification "RFC 1094).

Here, the "logical file" represents the data stored on the disk virtualized in a certain unit. FIG. 15 is a block diagram showing the operation of the logical file access method in NFS. In NFS,
On the computer to which the disk is connected, an NFS daemon that manages the logical file stored in the disk (a daemon is a system resident type service process in a Unix OS etc.) is started. The client program running on other computer # 0 is
When writing the data of the logical file to the disk, the write request and the data to be written are transmitted to the NFS daemon on the computer # 1. When the NFS daemon receives the write request, it once writes the data to be written to computer # 1.
It is captured in the memory of and the data is written to the disk. On the other hand, when the client program of the computer reads the data of the logical file from the disk, the client program sends a read request to the NFS daemon on the computer # 1. When the NFS daemon receives the read request, it reads the data of the requested logical file from the disk and once captures it in the memory of computer # 1. After that, as a response to the client program, the data in the memory of the read computer is transmitted to the client program of the computer # 0.

[0004]

However, it is known that it is inefficient to use a logical file access method such as NFS in a disk sharing type cluster system. The details will be described below.

FIG. 16 is a block diagram schematically showing an operation in which a client accesses a logical file managed by NFS when NFS is operated in a disk sharing type cluster system. In the disk sharing type cluster system, each computer can directly read / write the disk. However, in the logical file management method in NFS, the data to be read / written must always pass through the computer in which the NFS daemon operates. That is, the overhead of temporarily relaying the read / write data occurs.

Further, in the disk sharing type cluster system, since a plurality of client programs can simultaneously access the same logical file, maintaining the consistency of the logical file becomes a problem. Here, maintaining consistency means preventing the processing of one client program and the processing of the other client program from having different results depending on the timing when a plurality of client programs access at the same time.

Particularly, regarding the management data of a logical file, if the file management data is updated at the same time by the request of the other client program while the file management data is updated by the request of one client program, the file management Data may be disturbed.

In NFS, the data of all files is once relayed by the NFS daemon to solve this problem.

On the other hand, in order to maintain the consistency of the file management data, an exclusive control system using a lock mechanism is known as a system that has been conventionally used (for example, Japanese Patent Laid-Open Nos. 9-34851 and 63-63). -943
See the description in Japanese Patent Publication No. 43).

For example, in a scalable mainframe having a cluster configuration, each computer acquires SCMP (System storage Cu) in order to acquire resources.
pled MultiProcessor System
m) is used as an exclusive control method (references:
"Overall picture of parallel general-purpose machine", Nikkei Watcher IBM edition separate volume,).

FIG. 17 is a block diagram showing an outline of this SCMP. The SSU (System Storage Unit) is a computer that controls exclusive control. First, when the computer # 0 accesses the data on the shared disk to change the management data of the logical file, it issues a lock request to the SSU and acquires the lock as a result of the response. When the computer # 0 acquires the lock, other computers cannot acquire the lock during this period. After that, when the computer # 0 issues a lock release request to the SSU, the lock is released. Finally, if a request for acquiring a lock arrives from another computer while the computer # 0 acquires the lock, the SSU notifies the computer that the lock is released.

As a result, the client program operating on each computer can directly read / write the disk and maintain the consistency of the management data of the logical file.

However, in the above-described exclusive control method by the lock mechanism, when the client program accesses the logical file, two means of locking and unlocking occur with respect to the means for controlling exclusive control of the logical file, There are problems such as increased network traffic and degraded network performance.

Further, in order to increase the capacity of the logical file stored on the disk and to speed up access to the logical file, it is effective to divide the data of the file and store the divided data on a plurality of disks. For example, in the field of disk devices, RAID (Redundant Arrays of Inexp
There is a disk array device represented by the “ensive disks” (reference: Gregory R. “Disk Arrays Hi
gh-Performance, High−Reliability Storage Subsy
stems ”, IEEE Computer, March 1994). By applying such a method even in a disk sharing type cluster system in which multiple disks are connected, the data of the logical file can be divided and stored in multiple disks. ,
It is possible to increase the capacity of the logical file and speed up access to the logical file.

However, in the disk sharing type cluster system, it is not enough to simply store data in a plurality of disk devices like a disk array device, and the data stored in a distributed manner is stored in one logical file. Logical file access management control to manage as data is required. In particular, in a disk sharing type cluster system in which logical files are distributedly stored in a plurality of disks, efficient management of logical files is an important issue.

Therefore, the present invention has been made in view of the above problems, and its main purpose is to allow a client program to read / write data of a logical file stored in a disk in a disk sharing type cluster system. In this case, it is to provide a system that reduces the overhead of relaying read / write data to another computer and reduces the number of communications between computers required to maintain the consistency of logical file management data. .

Another object of the present invention is to disperse the processing load in the address conversion means when the client program reads / writes the data of the logical file in the disk sharing type cluster system, and to improve the efficiency of the address conversion processing. It is to provide a system to improve the.

Still another object of the present invention is to perform read / write when a client program reads / writes data of a logical file in a disk sharing type cluster system in which data of one logical file is distributed and stored in a plurality of disks. An object of the present invention is to provide a system that reduces the overhead of relaying the created data to another computer and also reduces the number of communications between computers necessary to maintain the consistency of the logical file management data.

Further, according to the present invention, in the disk sharing type cluster system in which the data of one logical file is distributed and stored in a plurality of disks, the address conversion processing when the client program reads / writes the data of the logical file is speeded up. Providing a system is one of the purposes.

Further, according to the present invention, the I / O when the client program reads / writes the data of the logical file
One of the purposes is to provide a system that speeds up the O issuing process.

Another object of the present invention is to speed up the whole process by overlapping the address conversion process and the I / O issuing process when the client program reads / writes the data of the logical file. is there.

[0022]

According to a first aspect of the present invention which achieves the above object, a plurality of computers, a computer connection network for connecting the computers, one or a plurality of disks storing logical files, and A logical file management system of a disk-sharing cluster system that consists of a disk connection network that can directly access each disk from a computer.
Then , it runs on any one of the computers, and for the logical file stored on the disk, from the specified logical file identifier and offset address in the logical file,
An address conversion unit that determines the disk identifier and offset address on the disk and returns it, and a client program that runs on any computer reads / writes the logical file stored on the disk connected to the disk connection network. The address conversion request is issued to the address conversion means by using the specified logical file identifier and the offset address in the logical file as parameters, and the returned disk identifier and the offset address on the disk are used as parameters for the disk. And logical file access means for directly reading / writing via a connection network.

In the second invention of the present application, a plurality of computers, a computer connection network for connecting the computers, one or a plurality of disks in which logical files are stored, and a disk in which each computer can directly access each disk In a disk-sharing cluster system consisting of a connection network, it operates on multiple computers and is responsible for one or more disks that do not overlap in the system and has a designated logical file identifier and logical file offset address. From this, the address identifier that determines the disk identifier and offset address on the disk and returns it, and the client program that runs on any computer reads the logical file stored on the disk connected to the disk connection network. / Operates when writing and requests address translation from the logical file identifier Identifying the address conversion unit to row, and issues an address conversion request to the identified address translation means the specified logical file identifier and the logical file offset within address as a parameter,
And an adaptive logical file access means for directly reading / writing via the disk connection network using the returned disk identifier and the offset address on the disk as parameters.

According to the third invention of the present application, a plurality of computers, a computer connection network connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and each computer directly accesses each disk. In a disk sharing type cluster system consisting of a disk connection network that can be used, a client program that runs on any computer can read the data of logical files distributed and stored in multiple disks connected to the disk connection network. / Operates when writing, and represents the set of data stored in the same disk among the distributed logical file data from the specified logical file identifier and logical file offset address. Determine the file identifier and offset address in the partial logical file, and return this A partial logical file address converting means, a partial logical file identifier obtained from the partial logical file address converting means, and an offset address in the partial logical file as a parameter to issue an address conversion request to the address converting means, and the returned disk And a distributed storage logical file access means for directly reading / writing through the disk connection network using the identifier and the offset address on the disk as parameters.

In the fourth invention of the present application, a plurality of computers, a computer connection network for connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and each computer directly accesses each disk. In a disk sharing type cluster system consisting of a disk connection network that can be used, a client program that runs on any computer can read the data of logical files distributed and stored in multiple disks connected to the disk connection network. It operates when writing / writing, divides the read / write range of the specified logical file into units of distributed storage, and sets a plurality of logical file offset addresses indicating the beginning of each divided range and the specified logical The partial theory returned by calling the partial logical file address conversion means with the file identifier as a parameter An address translation request is issued to the address translation means using the file identifier and the offset address in the partial logical file as parameters, and the returned multiple disc identifiers and offset addresses on the disks are used as parameters, and simultaneously for multiple disks. And a distributed storage logical file parallel access means for performing read / write.

According to the fifth aspect of the present invention, a plurality of computers, a computer connection network connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and each computer directly accessing each disk. In a disk sharing type cluster system comprising a disk connection network capable of performing the above, in the second aspect of the present invention, an address conversion unit operating on a plurality of computers and a client program operating on any computer are connected to the disk connection network. It operates when reading / writing data of a logical file that is distributed and stored on multiple disks, divides the read / write range of the specified logical file into distributed storage units, and With multiple logical file offset addresses indicating the beginning of the range and the specified logical file identifier as parameters Call the partial logical file address conversion means, specify each of the address conversion means to issue an address conversion request using the returned plurality of partial logical file identifiers as parameters, and obtain the partial logic obtained from the partial logical file address conversion means. Address conversion requests are issued simultaneously to a plurality of address conversion means using the file identifier and the offset address in the partial logical file as parameters, and the returned plurality of disk identifiers and offset addresses on the disks are used as parameters via the disk connection network. And an adaptive distributed storage logical file access means for directly reading / writing.

In the sixth invention of the present application, a plurality of computers, a computer connection network connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and each computer directly accesses each disk. In a disk sharing type cluster system comprising a disk connection network capable of performing the above, an address conversion unit operating on a plurality of computers of the second invention and a client program operating on any computer are connected to the disk connection network. It operates when reading / writing the data of the logical file that is distributed and stored in the specified multiple disks, and divides the read / write range of the specified logical file into units that are distributed and stored. With multiple logical file offset addresses indicating the beginning of each range and specified logical file identifiers as parameters The partial logical file address converting means is called, the address converting means that should issue the address conversion request using the returned plurality of partial logical file identifiers as parameters are respectively specified, and the partial logical file obtained from the partial logical file address converting means. An adaptive distributed storage logical file parallel access means for directly reading / writing a plurality of disks while issuing an address translation request to a plurality of address translation means using the identifier and the offset address in the partial logical file as parameters. , Are included.

[0028]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described. In the embodiment of the first invention of the present application, a plurality of computers, a computer connection network connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and a direct connection from each computer to each disk. In a disk sharing type cluster system consisting of a disk connection network that can access to a disk, an address conversion unit (7 in FIG. 1) operating on any computer is provided with a specified logical file identifier and a logical file offset address. From
The disc identifier and offset address on the disc are determined and returned.

The client programs (5-1 to 5-k in FIG. 1) operating on any computer read the logical file stored in the disk connected to the disk connection network (3 in FIG. 1). When writing / writing, the logical file identifier and the offset address in the logical file are designated to specify the logical file access means (6-1 to 6-in FIG. 1).
call k).

Logical file access means (6-1 in FIG. 1)
6-k) issues an address conversion request to the address conversion means (7 in FIG. 1) using the logical file identifier and the offset address in the logical file as parameters, and uses the returned disk identifier and offset address on the disk as parameters. Direct read / write via the disk connection network.

Thus, when the client program (5-1 to 5-k in FIG. 1) reads / writes the data of the logical file stored in the disk connected to the disk connection network, the client program operates. Directly read from the computer via the disk connection network
Since the data can be written, the overhead that occurs when the data to be read / written passes through another computer can be reduced.

Further, the logical file access means (6-1 to 6-k in FIG. 1) serializes the address conversion request arriving from each client program and, while processing the request for a certain logical file, the same. Wait for the processing of other requests for the logical file. As a result, even if one of the client programs makes an access to update the management data of the logical file, the consistency of the logical file management data is maintained.

That is, the communication that occurs between the computers when the client program accesses the data of the logical file is only one communication between the computer where the client program operates and the computer where the address conversion means operates at most. . As a result, the number of times of communication can be reduced as compared with the exclusive control method involving a lock request / unlock request that is performed in the conventional mainframe.

In the embodiment of the second invention of the present application, the address converting means (7-1 to 7-h in FIG. 3) for handling one or a plurality of disks which are not duplicated in the system are installed on a plurality of computers. The adaptive logical file access means (8-1 to 8-k in FIG. 3) operates to specify the address conversion means to issue the address conversion request from the logical file identifier, and specify the specified logical file identifier and logical file. An address translation request is issued to the specified address translation means by using the internal offset address as a parameter, and the returned disk identifier and the offset address on the disk are used as parameters for the disk connection network (FIG. 3).
Direct reading / writing is performed via 3).

As a result, the address translation processing is distributed and performed by the plurality of address translation means, so that the efficiency of the address translation processing is improved.

In the embodiment of the third invention of the present application, a client program operating on an arbitrary computer reads / writes data of a logical file distributedly stored in a plurality of disks connected to a disk connection network. The partial logical file identifier and the partial logical file partial logic that represents a partial set for each disk among the logical file data stored in a distributed manner from the specified logical file identifier and offset address in the logical file. Partial logical file address conversion means (11-1 to 11 in FIG. 6) that determines the offset address in the file and returns it
11-k), the partial logical file identifier obtained from the partial logical file address converting means and the offset address in the partial logical file are used as parameters to issue an address conversion request to the address converting means, and the returned disk identifier and disk are returned. Directly read / write via the disk connection network using the above offset address as a parameter
Distributed storage logical file access means (10-1 to 10-k in FIG. 6) for writing.

As a result, even with respect to the data of the logical files distributedly stored in a plurality of disks, similarly to the effect of the first invention, the data can be directly read from the computer on which the client program operates via the disk connection network. Since the data can be written / written, the overhead that occurs when the data to be read / written passes through another computer can be reduced. At the same time, the number of communications for maintaining the consistency of the logical file management data can be reduced.

In the embodiment of the fourth invention of the present application, the read / write range of the designated logical file is divided for each unit of distributed storage, and a plurality of logical file offsets indicating the heads of the respective divided ranges are set. The partial logical file address conversion means (11-1 to 11-k in FIG. 9) is called with the address and the specified logical file identifier as parameters, and the returned partial logical file identifier and a plurality of partial logical file offset addresses are used as parameters. Issue an address conversion request to the address conversion means, issue I / O to one disk using the returned set of disk identifiers and offset addresses on the disk as parameters, and then complete I / O Of the distributed storage logical file that simultaneously issues I / O to other disks without waiting for Parallel access means (12-1 in FIG. 9
.About.12-k).

As a result, I /
It is possible to overlap the time to issue an O and wait for its response with the time to issue an I / O to another disk and wait for its response.
The entire processing time from the start of writing to the end of writing is shortened, and the read / write processing can be speeded up.

In the embodiment of the fifth invention of the present application, the read / write range of the specified logical file is divided for each unit of distributed storage, and a plurality of logical files indicating the beginning of each divided range are stored. Using the offset address and the designated logical file identifier as a parameter, the partial logical file address conversion means (11-1 to 11- in FIG. 11) is used.
k), and using the returned partial logical file identifier and the offset address in the partial logical file as parameters, a plurality of address conversion means (7-1 to 14 in FIG. 13).
-H) The address conversion means that should issue the address conversion request is specified, the address conversion request is issued to the address conversion means using the partial logical file identifier and the offset address in the partial logical file as parameters, and then the address conversion is performed. Without waiting for a response to the request, the processing is repeated for the offset address in the other logical file obtained as a result of the division, and the read / write operation is performed on the disk by using the set of the returned disk identifier and the offset address on the disk as a parameter. Adaptive distributed storage logical file access means for writing (13-1 in FIG. 11)
~ 13-k).

As a result, the time until one address translation request is issued and the response is waited for and the time until another address translation request is issued and the response is waited for can be overlapped, so read / write. When the address conversion is performed on all of the offset addresses in the plurality of logical files obtained by dividing the range, the entire processing time until the completion of all the address conversion processing is shortened, and the address conversion processing can be speeded up.

In the embodiment of the sixth invention of the present application, the read / write range of the designated logical file is divided for each unit of distributed storage, and a plurality of logical files indicating the beginning of each divided range are stored. Partial logical file address conversion means (11-1 to 11-in FIG. 13) using the offset address and the specified logical file identifier as parameters.
k), and using the returned partial logical file identifier and the offset address in the partial logical file as parameters, a plurality of address conversion means (7-1 to 14 in FIG. 13).
-H) specifies the address translation means that should issue the address translation request, issues the address translation request to the address translation means with the partial logical file identifier and the offset address in the partial logical file as parameters, and at the same time performs the address translation The process is repeated for all the offset addresses in the logical file obtained as a result of the division without waiting for the response of the request, and as soon as the response of the address translation request is returned, the disk identifier obtained from the response and the offset address on the disk are set as parameters. Adaptive distributed storage logical file parallel access means (14-1 to 14-k in FIG. 13) for issuing I / O to the disk as
Equipped with.

As a result, one address translation request is issued and the time until the response is awaited, and one I
Since it takes time to issue an I / O request and wait for its response, it is possible to shorten the time for all address conversion processing of offset addresses in a plurality of logical files that divide the read / write range, and Since the I / O issue processing time for the disk can be shortened, the entire address conversion processing and read / write processing can be speeded up.

As an embodiment of the present invention, in addition to the client program, address conversion means, logical file access means, adaptive logical file access means,
The partial logical file address conversion means, the distributed storage logical file access means, the distributed storage logical file parallel access means, the adaptive distributed storage logical file access means, the adaptive distributed storage logical file parallel access means, etc. function according to the program executed on the computer. In this case, the present invention provides a plurality of computers, a computer connection network that connects the computers, a plurality of disks in which data of one logical file is distributed and stored, and a disk from each computer. This can be implemented by loading the program from a recording medium recording these programs into a computer and executing the program in a hardware configuration provided with a disk connection network that can directly access the computer. Hereinafter, detailed description will be given with reference to examples.

[0045]

[Embodiment 1] FIG. 1 is a block diagram showing a configuration of a first embodiment of the present invention. Referring to FIG. 1, in this embodiment, m computers 1-1 to 1-m and each computer 1
Computer connection network 2 for mutually connecting -1 to 1-m, and n disks 4-1 to 4-1 for storing data of logical files
-N, each computer 1-1 to 1-m, and each disk 4-1 to
A disk connection network 3 for connecting 4-n and a plurality of client programs 5-1 to 5- that operate on any computer.
k, the logical file access means 6-1 to 6-k operating on the computer on which the client program operates, the specified logical file identifier and the offset address in the logical file, the disk identifier and the offset address on the disk are specified. Address conversion means 7 to be returned
And are provided.

In the configuration of this embodiment, for the sake of explanation, the computer connection network 2 through which communication data between computers flows and the disk connection network 3 through which read / write data between computers and disks flow are shown separately. If you use a connection network such as Fiber Channel, you can combine these.

FIG. 2 is a flow chart showing the processing flow of the address converting means 7 in the first embodiment of the present invention. A first embodiment of the present invention will be described with reference to FIGS. 1 and 2. The address conversion means 7 is a logical file access means 6-1 to 6- that operates on an arbitrary computer.
The logical file identifier and the offset address in the logical file are designated from k and called.

First, the address converting means 7 acquires the logical file identifier and the offset address within the logical file from the request received from the logical file accessing means 6-1 to 6-k (step 101).

Next, it is confirmed by comparing the logical file identifiers whether the address conversion of the same logical file as the designated logical file is currently being processed. If the same logical file is being processed, it waits until the previous processing is completed (step 102). For example,
When the address conversion request for the logical file identifier designated by the logical file access unit 6-1 is currently being processed, and the address conversion request for the same logical file identifier is received from the logical file access unit 6-2 at this time. In this case, the address conversion processing of the request from the logical file access means 6-2 is suspended until the response is returned to the logical file access means 6-1.

Next, the logical file management data corresponding to the designated logical file identifier is read from the disk in which the logical file management data is stored (step 103). As a method of identifying the disk storing the management data of the logical file and the address on the disk, the identifier of the stored disk and the offset address on the disk are calculated from the identifier of the logical file by a predetermined calculation formula. There is also a method or a method of storing it in a predetermined logical file management data storage area.

Next, referring to the management data of the logical file,
The offset address on the disk is specified from the offset address in the logical file (step 104). As management data of a logical file for address conversion, for example, there is an address conversion table of a set of an offset address in a logical file and a block number. The address conversion table is a table in which the offset address of the logical file is used as an index. A block represents a unit obtained by dividing a storage area on the disk by a certain fixed length (block size), and a block number is a value obtained by numbering the divided storage area on the disk from the beginning. The product of the block number and the block size is the offset address on the disk that indicates the beginning of the block.

The address conversion means 7 refers to the block number corresponding to the area in the address conversion table using the designated offset address in the logical file as an index, and calculates the offset address on the disk from the block number. In this way, the address conversion means 7 identifies the disc identifier and the offset address on the disc from the designated logical file identifier and offset address in the logical file, and returns them to the request source (step 105).

FIG. 3 is a flow chart showing the processing flow of the logical file access means 6-1 to 6-k in the first embodiment of the present invention. The logical file access means 6-1 to 6-k have client programs 5-1 to 5
-K is the disk 4-1 connected to the disk connection network 3
Read / write data of logical file stored in 4-n
When writing, the client programs 5-1-5
Called from -k.

First, the called logical file access means 6-1 to 6-k specify the logical file identifier and the offset address in the logical file and issue an address conversion request to the address conversion means 7 (step 11).
1).

Next, the response of the issued address translation request is waited for, and after the response arrives, the disk identifier in which the logical file specified from the response is stored and the offset address on the disk corresponding to the offset address in the logical file are acquired. (Step 112).

After that, the logical file access means 6-1
.. 6-k specify the offset address on the disk and issue I / O to the disk corresponding to the acquired disk identifier (step 113).

As a result, the client program 5-
Disks 1 to 5-k connected to the disk connection network 3
When reading / writing the data of the logical files stored in -1 to 4-n, the client program 5-
From a computer running 1-5-k to a disk,
It is possible to read / write directly. That is, in the first embodiment of the present invention, the conventional lead / read
It reduces the overhead that occurs when the data to write goes through another computer.

Further, in the first embodiment of the present invention, the access exclusive control when a plurality of client programs read / write the same logical file is simply locked inside the address conversion means 7. It only manages the state, and like lock mechanisms in conventional mainframes, it is necessary to lock and unlock 2
No round trip communication is required, and at most one round trip communication with the address conversion means 7 is sufficient. Therefore, the number of communications required when the client program reads / writes the disk is reduced.

The address conversion means 7 is not limited to the processing method shown in FIG. 2, but only when the access is performed so that the logical file management data changes, the address conversion means 7 locks only during the processing. As described above, the lock period of the internal processing in the address conversion means 7 may be shortened and the address conversion means 7 may be configured to perform parallel processing. For example, the management data of the logical file is not changed by an address conversion request for reading the data of the logical file or an address conversion request (overwrite write) for writing the data on the area where the data storage area has already been determined. In this case, even if a plurality of address conversion requests arrive at the same time and the address conversion processing is performed, the logical file management data is not disturbed.

However, when a write (additional write) address conversion request while adding the data storage area to the logical file management data arrives, the data storage area on the disk is newly secured or during the processing. During the process of adding the updated area to the logical file management data and updating it, if the other address translation process is allowed, the logical file management data may be disturbed or the address translation process using wrong data may be performed. . Therefore, the address conversion process is exclusively controlled during the period of these processes.

As described above, in the address conversion means 7, the period during which the address conversion process is excluded can be set in fine units. In this case, if a lock mechanism that involves communication is used as in the conventional method, and such exclusive control in small units is performed, communication associated with locking and unlocking frequently occurs and the overhead increases. However, in the first embodiment of the present invention, since the lock control is closed by the internal processing of the address conversion means 7, communication does not occur.

[Second Embodiment] Next, a second embodiment of the present invention will be described. FIG. 4 is a diagram showing the configuration of the second exemplary embodiment of the present invention. Referring to FIG. 4, the second embodiment of the present invention is directed to a plurality of computers 1 on which client programs operate.
-1 to 1-k, a plurality of computers 1-i to 1-j on which the address conversion means operates, and each computer 1-1 to 1-k, 1-
A computer connection network 2 connecting i to 1-j, a disk connection network 3 connecting each computer to each disk, and a plurality of disk groups 9-1 to 9-h for storing data of logical files
And a plurality of disks 4-1-1 to constitute a disk group
4-h-b, a plurality of client programs 5-1 to 5-k operating on an arbitrary computer, and an adaptive logical file access means 8 operating on a computer running the client programs 5-1 to 5-k. -1 to 8-k, and a plurality of address conversion means 7 for identifying and returning the disc identifier and the offset address on the disc from the designated logical file identifier and offset address in the logical file for the assigned disc 1 to h.

A plurality of address conversion means 7-1 to 7-h
Is a plurality of disk groups 9 connected to the disk connection network 3.
Any one of -1 to 9-h is shared. In this embodiment, the address conversion means 7-x (x: 1 to h) is the disk group 9
-X shall be in charge. Further, the address conversion means 7-
The disks handled by 1 to 7-h are set so that the plurality of address conversion units 7-1 to 7-h do not overlap each other.

FIG. 5 is a flow chart showing a processing flow of the adaptive logical file access means 8-1 to 8-k in the second embodiment of the present invention. A second embodiment of the present invention will be described with reference to FIGS. 4 and 5. In the adaptive logical file access means 8-1 to 8-k, the client programs 5-1 to 5-k are stored in a plurality of disks 4-1-1 to 4-h-b connected to the disk connection network 3. It is called from the client programs 5-1 to 5-k when reading / writing the data of the logical file.

First, the called adaptive logical file access means 8-1 to 8-k specify the address conversion means for managing the logical file from the designated logical file identifier (step 121). As a specific method for specifying the address conversion means, for example, a method of internally having a conversion table of a logical file identifier and an identifier of the address conversion means for managing the logical file, or a logical file identifier stored in a specific disk is stored. There is a method of incorporating this information and using it.

Next, the adaptive logical file access means 8-
1 to 8-k issue an address translation request to any of the identified address translation units 7-1 to 7-h using the designated logical file identifier and offset address in the logical file as parameters (step 122). ).

After the address translation request is issued, the response is waited for, and after the response arrives, the disk identifier in which the logical file designated by the response is stored and the offset address on the disk are acquired (step 123).

Thereafter, the adaptive logical file access means 8
Each of -1 to 8-k designates an offset address on the disk and issues an I / O to the disk corresponding to the acquired disk identifier (step 124).

Thus, when a plurality of client programs read / write the data of the logical file stored in the disk connected to the disk connection network, the address conversion requests are concentrated on one address conversion means. Can be distributed to a plurality of address conversion means. As a result, the efficiency of the address conversion process is improved.

[Embodiment 3] Next, a third embodiment of the present invention will be described. FIG. 6 is a block diagram showing the configuration of the third exemplary embodiment of the present invention. Referring to FIG. 6, in this embodiment, a plurality of computers 1 on which client programs operate
-1 to 1-k and the computer 1 in which the address conversion means operates
-M, a computer connection network 2 that connects the computers 1-1 to 1-k, and 1-m, disks 4-1 to 4-n that store data of logical files, and computers 1-1 to 1-1 -K, 1-
disk connection network 3 that connects m with each of the disks 4-1 to 4-n, a plurality of client programs 5-1 to 5-k that operate on an arbitrary computer, and address translation that operates on any of the computers. Means 7, distributed storage logical file access means 10-1 to 0-k operating on the computer on which the client program operates, partial logical file address conversion means 11-1 to 11-operating on the computer on which the client program operates. and k.

FIG. 7 is a flow chart showing the processing flow of the partial logical file address conversion means 11-1 to 11-k in the third embodiment of the present invention. 6 and 7
The operation of the third embodiment of the present invention will be described with reference to FIG.

Partial logical file address conversion means 11-
1 to 11-k are distributed storage logical file access means 1
The logical file identifier and the offset address in the logical file are designated by 0-1 to 10-k and called.

First, the partial logical file address conversion means 11-1 to 11-k acquire the logical file identifier and the offset address in the logical file (step 13).
1).

Next, the offset address in the partial logical file and the partial logical file identifier corresponding to the logical file identifier and the offset address in the logical file are specified and returned (step 132).

Here, the partial logical file represents a set of data stored in the same disk among the data of the logical files distributedly stored in each disk. The offset address in the partial logical file represents the offset address in the logical file in the partial logical file. As a specific method of converting the offset address in the logical file into the offset address in the partial logical file, it can be obtained from the data division size and the order of the disks storing the data of the divided logical file.

For example, the number of disks for distributed storage is 4,
Assuming that the partition size is 4 KB and the data of the logical file of 32 KB is distributed and stored in order from the head disk, it becomes 4K.
It is divided into eight areas, each B. As a result, the offset address in the logical file is 0 to 0 on the first disk.
Data of 4 KB and 16 to 20 KB areas are stored, and these data form a partial logical file of the first disk. For example, the offset address in the logical file: 16 KB corresponds to the offset address in the partial logical file of the first disk: 4 KB.

The partial logical file identifier is a value that uniquely identifies on which disk the partial logical file is stored among a plurality of partial logical files forming the logical file.

FIG. 8 is a flow chart showing the operation of the distributed storage logical file access means 10-1 to 10-k in the third embodiment of the present invention. In the distributed storage logical file access means 10-1 to 10-k, the client programs 5-1 to 5-k read / write data of logical files distributed and stored in a plurality of disks connected to the disk connection network 3. Called when you do.

First, the called distributed storage logical file access means 10-1 to 10-k use the specified logical file identifier and the offset address in the logical file as parameters, and the partial logical file address conversion means 1
Call 1-1 to 11-k (step 141).

As a result, the offset address in the partial logical file and the partial logical file identifier are acquired (step 142).

Next, an address translation request is issued to the address translation means 7 by using the partial logical file identifier and the offset address within the partial logical file as parameters (step 143).

After issuing the address translation request, the response is waited for, and after the response arrives, the identifier of the disk corresponding to the partial logical file identifier and the offset address in the partial logical file and the offset address on the disk are acquired (step 144).

After that, using the disc identifier and the offset address on the disc as parameters, I / O is issued via the disc connection network 3 (step 145).

As a result, the data of the logical files distributed and stored in a plurality of disks can be directly read / written from the computer on which the client program operates via the disk connection network, and therefore read / write. The overhead when data passes through another computer can be reduced. Further, it is possible to reduce the number of communications for maintaining consistency when a plurality of client programs access the same file.

[Embodiment 4] Next, a fourth embodiment of the present invention will be described. FIG. 9 is a block diagram showing the configuration of the fourth exemplary embodiment of the present invention. Referring to FIG. 9, the present embodiment is directed to a plurality of computers 1 on which client programs operate.
-1 to 1-k and the computer 1 in which the address conversion means operates
-M, a computer connection network 2 that connects the computers 1-1 to 1-k, and 1-m, disks 4-1 to 4-n that store data of logical files, and computers 1-1 to 1-1 -K, 1-
disk connection network 3 that connects m with each of the disks 4-1 to 4-n, a plurality of client programs 5-1 to 5-k that operate on an arbitrary computer, and address translation that operates on any of the computers. Means 7, partial logical file address conversion means 11-1 to 11-k operating on the computer running the client program, distributed storage logical file parallel access means 12-1 to 12 operating on the computer running the client program. -K and.

FIG. 10 shows distributed storage logical file parallel access means 12-1 to 12 in the fourth embodiment of the present invention.
It is a flowchart which shows operation | movement of -k. 9 and 1
The operation of the fourth exemplary embodiment of the present invention will be described with reference to FIG. The following description will focus on the distributed storage logical file parallel access means 12-1.

Distributed storage logical file parallel access means 1
2-1 is called when the client program 5-1 reads / writes the data of the logical file distributed and stored in the plurality of disks connected to the disk connection network 3.

First, the called distributed storage logical file parallel access means 12-1 divides the read / write range of the designated logical file into units of the distributed storage, and the logical file indicating the beginning of each range. The inner offset address is specified (step 151). For example, when the designated logical file offset address is 0, the call size is 16 KB, and the unit for distributed storage is 4 KB, the range is divided into four ranges of 0, 4 KB, 8 KB, and 12 KB.

Next, the partial logical file address conversion means 11-1 is called with the logical file offset address indicating the head of each divided range and the logical file identifier as parameters (step 152). as a result,
An offset address in the partial logical file and a partial logical file identifier corresponding to the offset address in the logical file indicating the beginning of each divided range are acquired (step 1
53).

Next, an address translation request is issued to the address translation means by using the acquired offset address in partial logical file and partial logical file identifier as parameters (step 154).

After issuing the address translation request, the response is waited for. After the response arrives, the offset address in the partial logical file, the disk identifier corresponding to the partial logical file identifier, and the offset address on the disk are acquired (step 155).

Here, the above processing is repeated for all the read / write ranges divided in step 151 (step 156).

Next, using the acquired disc identifier and offset address on the disc as parameters, I / O is issued to the disc (step 157). However,
Step 15 asynchronously without waiting for I / O completion
I / O issuance is repeated for all read / write ranges divided in step 1 (step 158).

After all I / O issuing processing is completed, I
The completion of the / O issuance is waited for (step 159).

After that, the process ends when all I / O issuance is completed (step 160).

As described above, when the client program reads / writes the data of the logical file stored in a distributed manner, the read / write range is divided into a plurality of areas, and the I / O is issued simultaneously for each of the read / write areas. The entire processing time from the start of writing to the end of writing is shortened, and the read / write processing can be speeded up.

[Fifth Embodiment] Next, a fifth embodiment of the present invention will be described. FIG. 11 is a block diagram showing the configuration of the fifth exemplary embodiment of the present invention. Referring to FIG. 11, in this embodiment, a plurality of computers 1-1 to 1-k on which a client program operates, a plurality of computers 1-i to 1-j on which an address converting unit operates, and each computer 1- 1-1-k,
A computer connection network 2 connecting 1-i to 1-j, a disk connection network 3 connecting each computer to each disk, and a plurality of disk groups 9-1 to 9-for storing data of logical files.
h and a plurality of disks 4-1-1 that form a disk group
~ 4-h-b, a plurality of client programs 5-1 to 5-k operating on an arbitrary computer, a plurality of address converting means 7-1 to 7-h, and a computer on which the client program operates. It comprises partial logical file address conversion means 11-1 to 11-k that operate and adaptive distributed storage logical file access means 13-1 to 13-k that operates on a computer on which a client program operates.

FIG. 12 shows adaptive distributed storage logical file access means 13-1 to 13 in the fifth embodiment of the present invention.
It is a flowchart which shows operation | movement of -k. The operation of the fifth exemplary embodiment of the present invention will be described with reference to FIGS. Hereinafter, description will be made focusing on the adaptive distributed storage logical file access unit 13-1.

Adaptive distributed storage logical file access means 1
3-1 is called when the client program 5-1 reads / writes data of a logical file distributed and stored in a plurality of disks connected to the disk connection network 3.

First, the called adaptive distributed logical file access means 12-1 divides the read / write range of the data of the specified logical file into units of distributed storage, and sets the logic indicating the beginning of each range. The in-file offset address is specified (step 171).

Next, the partial logical file address conversion means 11-1 is used with the logical file offset address indicating each divided range and the logical file identifier as parameters.
Is called (step 172).

As a result, the partial logical file offset address and the partial logical file identifier corresponding to the logical file offset address and the logical file identifier are acquired (step 173).

Next, the address conversion means 7-1 to h in which the partial logical file is managed are specified (step 17).
4). In this example, the address conversion means 7-1 is specified.

Next, an address conversion request is issued to the specified address conversion means 7-1 using the offset address in the partial logical file and the partial logical file identifier as parameters (step 175). Here, without waiting for the response of the address translation request, for all of the read / write range divided in step 171,
The above processing is repeated until the address translation request is issued (step 176).

After issuing the address translation request for all of the divided read / write ranges, the response is awaited. As soon as the response arrives, the offset address on the disk and the disk identifier corresponding to the specified offset address in partial logical file and partial logical file identifier are acquired (step 177).

This is repeated until the responses to all issued address translation requests are returned (step 178).

Next, I / O is issued respectively using the obtained offset address on the disk and the disk identifier as parameters (step 179).

As a result, when reading / writing the data of the logical file distributedly stored in a plurality of disks, the read / write range is divided into a plurality of areas, and the address conversion request is issued simultaneously for each of them. Since the entire processing time from the start of the translation process to the completion thereof is shortened, the address translation process can be sped up.

[Sixth Embodiment] Next, a sixth embodiment of the present invention will be described. FIG. 13 is a block diagram showing the configuration of the sixth exemplary embodiment of the present invention. Referring to FIG. 13, in the present embodiment, a plurality of computers 1-1 to 1-k on which a client program operates, a plurality of computers 1-i to 1-j on which an address converting unit operates, and each computer 1- 1-1-k,
A computer connection network 2 connecting 1-i to 1-j, a disk connection network 3 connecting each computer to each disk, and a plurality of disk groups 9-1 to 9-for storing data of logical files.
h and a plurality of disks 4-1-1 that form a disk group
~ 4-h-b, a plurality of client programs 5-1 to 5-k operating on an arbitrary computer, a plurality of address converting means 7-1 to 7-h, and a computer on which the client program operates. It comprises partial logical file address conversion means 11-1 to 11-k that operate and adaptive distributed storage logical file parallel access means 14-1 to 14-k that operates on a computer on which a client program operates.

FIG. 14 shows the adaptive distributed storage logical file parallel access means 14-1 to 14-1 according to the sixth embodiment of the present invention.
It is a flow chart which shows operation of 14-k. The operation of the sixth exemplary embodiment of the present invention will be described with reference to FIGS. 13 and 14. Hereinafter, description will be made focusing on the adaptive distributed storage logical file parallel access unit 14-1.

The adaptive distributed storage logical file parallel access means 14-1 is called when the client program 5-1 reads / writes the data of the logical file distributedly stored in a plurality of disks connected to the disk connection network. .

First, the called adaptive distributed storage logical file parallel access means 14-1 divides the range for reading / writing the data of the specified logical file into units for distributed storage, and sets the beginning of each range. The offset address in the indicated logical file is specified (step 181).

Next, the partial logical file address conversion means 11-1 is used with the logical file offset address and the logical file identifier indicating each divided range as parameters.
Is called (step 182).

As a result, the offset address in the partial logical file and the partial logical file identifier corresponding to the offset address in the logical file and the logical file identifier are acquired (step 183).

Next, the address conversion means 7-1 to 7-h whose partial logical file is managed are specified (step 184). Here, it is assumed that the address conversion means 7-1 is specified.

Next, an address conversion request is issued to the specified address conversion means 7-1 using the offset address in the partial logical file and the partial logical file identifier as parameters (step 185).

Here, without waiting for the response of the address translation request, the above processing is repeated for all the read / write ranges divided in step 181 until the address translation request is issued (step 186).

After issuing the address conversion request for all of the divided read / write ranges, the response is awaited (step 187).

Here, it is detected that some kind of response has arrived (step 188).

If the response is the response to the address translation request, the offset address on the disk and the disk identifier corresponding to the specified offset address in the partial logical file and the specified partial logical file identifier are acquired (step 189).

Next, an I / O is issued to the disk using the offset address on the disk and the disk identifier (step 190).

However, the I / O is issued asynchronously and the process returns to step 187 to wait for either the I / O completion response or the address translation request response. In step 188, if the response is I / O completion, step 1
At 81, it is confirmed whether or not I / O has been performed for all the divided ranges, and if not completed, the process returns to step 187 to wait for either an I / O completion response or an address translation request response.

If all have been completed, the process ends (step 191).

As a result, it is possible to shorten the address conversion processing time for all the offset addresses in a plurality of logical files obtained by dividing the read / write range, and at the same time, shorten the I / O issue processing time for a plurality of disks. It is possible to speed up the entire conversion process and read / write process.

[0125]

As described above, the present invention has the following effects.

The effect of the first invention of the present application is that when a client program reads / writes data of a logical file stored in a plurality of disks connected to a disk connection network, the disk is accessed from a computer on which the client program operates. Since it is possible to read / write directly via the connection network, it is possible to reduce the overhead that occurs when the data to be read / written passes through another computer.

At the same time, when the client program makes an access accompanied by updating of the logical file management data, the logical file access means serializes the address conversion request arrived from each client program, and requests the certain logical file. While processing, waiting for the processing of other requests, the consistency of the logical file management data is maintained.

As a result, the communication that occurs between the computers when the client program accesses the data of the logical file is only one communication between the computer in which the client program operates and the computer in which the address conversion means operates. Become. As a result, it is possible to reduce the number of communications as compared with the conventional exclusive control method for maintaining the consistency of the logical file management data accompanied by the lock request / unlock request which is performed in the mainframe.

The effect of the second invention of the present application is that the load of address translation requests concentrated on one address translation unit from a plurality of client programs can be distributed to the plurality of address translation units, and the address translation processing efficiency is improved. ,
That's what it means.

The effects of the third invention of the present application are as follows.
Similar to the invention, since it is possible to directly read / write from the computer on which the client program operates via the disk connection network, it is possible to reduce the overhead that occurs when the data to be read / written passes through another computer. That is.

Further, when a plurality of client programs access the same logical file, it is possible to reduce the number of communications for maintaining the consistency of the logical file management data.

The effect of the fourth invention of the present application is that the entire processing time from the start to the end of read / write is shortened,
This means that the read / write processing can be speeded up.

The effect of the fifth invention of the present application is that when the address conversion is performed for all the offset addresses in a plurality of logical files obtained by dividing the read / write range, the entire processing time until the completion of all the address conversion processing is shortened. That is, the address conversion process can be speeded up.

The effect of the sixth invention of the present application is that the entire time of the address conversion processing for all offset addresses in a plurality of logical files obtained by dividing the read / write range is shortened, and at the same time, the I / O issuing processing for a plurality of disks That is, the entire time of the address conversion process and the read / write process can be speeded up.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a flow chart showing the operation of the address conversion means in the first embodiment of the present invention.

FIG. 3 is a flow chart showing an operation of a logical file access means in the first exemplary embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 5 is a flow chart showing the operation of the adaptive logical file access means in the second exemplary embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 7 is a flow chart showing an operation of a distributed storage logical file access means in the third exemplary embodiment of the present invention.

FIG. 8 is a flowchart showing an operation of a partial logical file address conversion means in the third exemplary embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of a fourth exemplary embodiment of the present invention.

FIG. 10 is a flow chart showing the operation of the distributed storage logical file parallel access means in the fourth exemplary embodiment of the present invention.

FIG. 11 is a block diagram showing a configuration of a fifth exemplary embodiment of the present invention.

FIG. 12 is a flow chart showing the operation of the adaptive distributed storage logical file access means in the fifth exemplary embodiment of the present invention.

FIG. 13 is a block diagram showing a configuration of a sixth exemplary embodiment of the present invention.

FIG. 14 is a flow chart showing the operation of the adaptive distributed storage logical file parallel access means in the sixth exemplary embodiment of the present invention.

FIG. 15 is a diagram schematically showing a conventional logical file access method.

FIG. 16 is a block diagram schematically showing a conventional logical file access method.

FIG. 17 is a block diagram showing a conventional exclusive control method for logical file access.

[Explanation of symbols]

1-1 to 1-k, 1-i to 1-j Computer 2 Computer connection network 3 Disk connection network 4-1 to 4-n, 4-1-1 to 4-1-a, 4-h-1
~ 4-h-b Disks 5-1 to 5-k Client programs 6-1 to 6-k Logical file access means 7, 7-1 to 7-h Address conversion means 8-1 to 8-k Adaptive logical file access Means 9-1 to 9-h Disk groups 10-1 to 10-k Distributed storage logical file access means 11-1 to 11-k Partial logical file address conversion means 12-1 to 12-k Distributed storage logical file parallel access means 13-1 to 13-k Adaptive distributed storage logical file access means 14-1 to 14-k Adaptive distributed storage logical file parallel access means

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Junichi Yamato 5-7-1 Shiba, Minato-ku, Tokyo Within NEC Corporation (56) Reference JP-A 63-155255 (JP, A) JP-A 62 -276620 (JP, A) JP-A-7-93199 (JP, A) JP-A-3-164827 (JP, A) Soltis S. et al. , The Design and Performance of a Shared Disk File System System for IRIX, the Sixth NASA Gode s ede e s ea, Spence, Center of Consence, USA 41-55, URL, http: // romulus. gsfc. nas a. gov / msst / conf 1998. html O'Keefe M. et al. T. , Shared File Systems and Fiber Channel, the Six NASA Goddard Space Flight Center Conference on SA, USA, USA, Essence, USA, October, Ess., USA, October, Ess. 1-16, URL, http p /// romulus. gsfc. n asa. gov / msst / conf 1998. html (58) Fields investigated (Int.Cl. ⁷ , DB name) G06F 12 / 00,15 / 16-15/177 G06F 3/06

Claims (12)

(57) [Claims] 1. A plurality of computers, a computer connection network for connecting the respective computers, one or a plurality of disks storing logical files, and a disk connection capable of directly accessing the respective disks from the respective computers. A logical file management system in a disk sharing cluster system including a network
A disk identifier that operates on any one of the plurality of computers and that stores a logical file stored in the disk from a specified logical file identifier and a logical file offset address. And an address conversion unit that determines offset addresses on the disk and returns them to the request source, and a client program that operates on any computer of the plurality of computers, in the disk connected to the disk connection network. It is activated when reading / writing the stored logical file, issues an address conversion request to the address converting means using the specified logical file identifier and offset address in the logical file as parameters, and the address converting means Disc identifier returned from and off on disc Ttoadoresu as a parameter, logical file management system including, be characterized by the a logical file access means directly perform read / write via the disk connection network. 2. A plurality of computers, a computer connection network for connecting the computers, one or a plurality of disks in which logical files are stored, and a disk connection that allows the computers to directly access the disks. A logical file management system in a disk sharing cluster system including a network
A system that operates on at least two computers of the plurality of computers and is in charge of one or more of the disks that do not overlap with each other, and uses the specified logical file identifier and logical file offset address An address conversion unit that determines an identifier and an offset address on the disk and returns them to a request source, and a disk in which a client program that operates on any computer of the plurality of computers is connected to the disk connection network Is started at the time of reading / writing the logical file stored in, and the address conversion means that should issue the address conversion request is specified from the logical file identifier, and the specified logical file identifier and offset address in the logical file are used as parameters. , For the specified address translation means An adaptive logical file access means for issuing an address conversion request and directly reading / writing via the disk connection network using the disk identifier returned from the address conversion means and the offset address on the disk as parameters. A logical file management system characterized by including. 3. A plurality of computers, a computer connection network connecting the respective computers, a plurality of disks in which data of one logical file is distributed and stored, and the respective computers can directly access the respective disks. A logical file in a disk sharing type cluster system including a disk connection network.
Is a file management system, operating on any one of the plurality of computers, for the logical file stored in the disk, from the specified logical file identifier and offset address in the logical file, An address conversion unit that determines a disk identifier and an offset address on the disk, and returns these to the request source, and a plurality of client programs that operate on any computer of the plurality of computers and that are connected to the disk connection network. The same disk among the data of the logical files stored in a distributed manner is started from the specified logical file identifier and the offset address in the logical file, which is activated when reading / writing the data of the logical files stored in the disk in a distributed manner. Of a partial logical file that represents a set of data stored in Parameter file identifier and offset address in partial logical file, and returns these partial logical file address conversion means, partial logical file identifier and offset address in partial logical file obtained from the partial logical file address conversion means As a distribution, the address translation request is issued to the address translation means, and the disk identifier returned from the address translation means and the offset address on the disk are used as parameters to directly read / write through the disk connection network. A logical file management system comprising: a storage logical file access means. 4. A plurality of computers, a computer connection network for connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and a disk in which each computer can directly access each disk. Logical file in the disk sharing type cluster system including the connection network
A management system, which operates on any one of the plurality of computers, and for a logical file stored on the disk, from a specified logical file identifier and a logical file offset address, a disk An address conversion unit that determines an identifier and an offset address on the disk, and returns these to the request source, and a plurality of the client programs that operate on any computer of the plurality of computers connected to the disk connection network. It is activated when reading / writing to the data of the logical file that is distributed and stored in the disk, and from the specified logical file identifier and offset address in the logical file, it is written to the same disk in the data of the logical file that is distributed and stored. Partial logic of a file that represents a set of stored data Partial logic of a file A partial logical file address conversion means for determining a file identifier and an offset address in the partial logical file and returning them, and a client program operating on any computer of the plurality of computers are connected to the disk connection network. It is activated when reading / writing the data of the logical file distributedly stored in the plurality of disks, and divides the read / write range of the specified logical file into units of the distributed storage, and The partial logical file address conversion means is called with a plurality of logical file offset addresses indicating the beginning of the range and the specified logical file identifier as parameters, and the returned partial logical file identifier and partial logical file offset address are used as parameters. Add to address translation means Distributed storage logical file parallel access means for issuing a read / write command to a plurality of disks at the same time by issuing a data conversion request and using a set of a plurality of disk identifiers and offset addresses on the disks returned from the address conversion means as parameters. And a logical file management system characterized by including. 5. A plurality of computers, a computer connection network for connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and a disk in which each computer can directly access each disk. A logical file management system for a disk sharing cluster system including a connection network
In the system , it operates on at least two computers of the plurality of computers and is in charge of one or more of the disks that do not overlap each other, and a disk identifier is assigned from a specified logical file identifier and offset address in the logical file. Address conversion means for determining offset addresses on the disk and returning them to the request source, and a plurality of the disks in which a client program operating on any computer among the plurality of computers is connected to the disk connection network It is started when reading / writing the data of the logical file distributed and stored in the disk, and it is stored in the same disk among the data of the logical file distributed and stored from the specified logical file identifier and offset address in the logical file. Of a partial logical file that represents a set of stored data A partial logical file address conversion means for determining a file identifier and an offset address in the partial logical file and returning them to a request source, and a client program operating on any computer of the plurality of computers is connected to the disk connection network. It is activated when reading / writing the data of the logical file that is distributed and stored in the specified multiple disks, and divides the read / write range of the specified logical file into units that are distributed and stored. A partial logical file address conversion means is called with a plurality of logical file offset addresses indicating the beginning of each range and the specified logical file identifier as parameters, and an address conversion request is issued with the returned plurality of partial logical file identifiers as parameters. Specify each address translation method that should be A plurality of disk identifiers returned by issuing an address conversion request to a plurality of the address converting means at the same time using the partial logical file identifier obtained from the partial logical file address converting means and the offset address in the partial logical file as parameters. And an adaptive distributed storage logical file access means for directly reading / writing via the disk connection network with the offset address on the disk as a parameter, and a logical file management system. 6. A plurality of computers, a computer connection network connecting the respective computers, a plurality of disks in which data of one logical file is distributed and stored, and each of the computers can directly access the respective disks. Disk connection network and logical file management of disk sharing type cluster system including
In the management system , the disk identifier operates on at least two of the plurality of computers and is in charge of one or more of the disks that do not overlap each other, and the disk identifier is determined from the specified logical file identifier and offset address in the logical file. And an address translation unit that determines offset addresses on the disk and returns them to the request source, and a plurality of client programs that operate on any computer of the plurality of computers and that are connected to the disk connection network. It is activated when reading / writing to the data of the logical file that is distributed and stored in the disk, and from the specified logical file identifier and offset address in the logical file, the same disk is stored in the data of the logical file that is distributed and stored. Of a partial logical file that represents a set of stored data A partial logical file address conversion means for determining the logical file identifier and the offset address in the partial logical file and returning them to the request source, and the client program operating on any computer of the plurality of computers is a disk connection network. It is activated when reading / writing the data of the logical file distributed and stored in multiple disks connected to, and divides the read / write range of the specified logical file for each unit of distributed storage. A plurality of partial logics returned from the partial logical file address conversion means by calling the partial logical file address conversion means with a plurality of logical file offset addresses indicating the beginning of each range and a specified logical file identifier as parameters Address conversion request with file identifier as parameter Each of the address conversion means to be issued is specified, and the address conversion is performed for a plurality of the address conversion means by using the partial logical file identifier and the offset address in the partial logical file obtained from the partial logical file address conversion means as parameters. A logical file management system comprising: an adaptive distributed storage logical file parallel access means for directly reading / writing a plurality of disks at the same time when a request is issued. 7. A plurality of computers, a computer connection network for connecting the respective computers, one or a plurality of disks storing logical files, and a disk connection capable of directly accessing the respective disks from the respective computers. Logical file management system of disk sharing type cluster system including network
A recording medium that records a program that realizes a program , (a) a logical file stored on the disk that is executed on any one of the plurality of computers and that has a specified logical file identifier Address conversion means for determining the disk identifier and the offset address on the disk from the offset address in the logical file, and returning these to the request source, and (b) a client operating on any computer of the plurality of computers. The program is started when the logical file stored in the disk connected to the disk connection network is read / written, and the specified logical file identifier and the offset address in the logical file are used as parameters for the address conversion means. Issue an address translation request, and A logical file access means for directly reading / writing via the disk connection network using the disk identifier and the offset address on the disk returned from the above, and (a) and (b) A recording medium recording a program for causing the computer to function. 8. A plurality of computers, a computer connection network for connecting the computers, one or a plurality of disks storing logical files, and a disk connection capable of directly accessing the disks from the computers. Logical file management system of disk-sharing cluster system including network
A recording medium for recording a program for realizing, (a) at least two operating on a computer, the logical file identifier charge one or more of the disc do not overlap each other, the designated one of said plurality of computers And an address conversion means for determining a disk identifier and an offset address on the disk from the offset address in the logical file and returning them to the request source, (b) a client operating on any computer of the plurality of computers The program is started when the logical file stored in the disk connected to the disk connection network is read / written, and the address conversion means that should issue the address conversion request is specified from the logical file identifier and designated. Parameter of logical file identifier and offset in logical file As an issue, an address translation request is issued to the identified address translation means, and the disk identifier returned from the address translation means and the offset address on the disk are used as parameters to directly read / write via the disk connection network. A recording medium recording an adaptive logical file access means for writing, and a program for causing each of the means (a) and (b) to function on the computer. 9. A plurality of computers, a computer connection network for connecting the respective computers, a plurality of disks in which data of one logical file is distributed and stored, and each of the computers can directly access the respective disks. Disk connection network and logical file management of disk sharing type cluster system including
In a recording medium in which a program for realizing a physical system is recorded , (a) a logical file stored on the disk, which operates on any one of the plurality of computers, Address conversion means for determining the disk identifier and the offset address on the disk from the specified logical file identifier and the offset address in the logical file, and returning these to the request source, (b) any of the plurality of computers A client program operating on a computer is started when reading / writing data of a logical file distributedly stored in the plurality of disks connected to the disk connection network,
From the specified logical file identifier and logical file offset address, the partial logical file identifier of the partial logical file that represents the set of data stored in the same disk among the distributed logical file data and the partial logical file offset Partial logical file address conversion means for determining addresses and returning them, (c) the partial logical file identifier and the partial logical file offset address obtained from the partial logical file address converting means as parameters, and the address converting means And a distributed storage logical file access means for directly reading / writing via the disk connection network by issuing an address conversion request to the address conversion means, and using the disk identifier returned from the address conversion means and the offset address on the disk as parameters. , (A) to a recording medium recording a program for operating each means in the computing machine of (c). 10. A plurality of computers, a computer connection network for connecting the respective computers, a plurality of disks in which data of one logical file is distributed and stored, and a disk which can directly access the respective disks from the respective computers. Logical file management of disk sharing type cluster system including connection network
A recording medium having a program for realizing the system recorded thereon (a) operating on any one of the plurality of computers and designating a logical file stored on the disk. Address conversion means for determining the disk identifier and the offset address on the disk from the logical file identifier and the offset address in the logical file, and returning these to the request source; and (b) an arbitrary computer among the plurality of computers. The client program operating on is started when reading / writing the data of the logical file distributedly stored in the plurality of disks connected to the disk connection network,
From the specified logical file identifier and logical file offset address, the partial logical file identifier of the partial logical file that represents the set of data stored in the same disk among the distributed logical file data and the partial logical file offset Partial logical file address conversion means for deciding addresses and returning them, and (c) a client program operating on an arbitrary computer of the plurality of computers is stored in the plurality of disks connected to the disk connection network. It is activated when reading / writing the data of the distributed logical file and divides the read / write range of the specified logical file for each unit of distributed storage, and indicates the beginning of each divided range. The offset address in multiple logical files and the specified logical file identifier are The partial logical file address conversion means as a parameter, issue an address conversion request to the address conversion means using the returned partial logical file identifier and the offset address in the partial logical file as parameters, and return a plurality of disk identifiers And the offset address on the disk as a parameter, the distributed storage logical file parallel access means for reading / writing a plurality of disks at the same time, and the respective means (a) to (c) functioning on the computer. A recording medium on which a program for recording is recorded. 11. A plurality of computers, a computer connection network for connecting the respective computers, a plurality of disks in which data of one logical file is distributed and stored, and a disk which can directly access each disk from each of the computers. Logical file management of disk sharing type cluster system including connection network
A recording medium having a program for realizing the system recorded thereon: (a) is in charge of one or a plurality of disks that operate on at least two computers of the plurality of computers and do not overlap each other. Address conversion means for determining a disk identifier and an offset address on the disk from the specified logical file identifier and offset address in the logical file and returning these to the request source, (b) any of the plurality of computers The client program that operates on the computer is started when reading / writing the data of the logical file distributedly stored in the plurality of disks connected to the disk connection network,
From the specified logical file identifier and logical file offset address, the partial logical file identifier of the partial logical file that represents the set of data stored in the same disk among the distributed logical file data and the partial logical file offset Partial logical file address conversion means for deciding addresses and returning them, and (c) distributed storage of client programs operating on any computer among the plurality of computers in a plurality of disks connected to the disk connection network. It is activated when reading / writing the data of the specified logical file, and divides the read / write range of the specified logical file into units that are distributed and stored, and shows a plurality of heads of the divided ranges. Parameter of the offset address in the logical file and the specified logical file identifier The partial logical file address conversion means as a parameter, and specifies the address conversion means to issue an address conversion request using a plurality of partial logical file identifiers returned from the partial logical file address conversion means as parameters. Using the partial logical file identifier and the offset address in the partial logical file obtained from the logical file address conversion means as parameters, at the same time, an address conversion request is issued to a plurality of the address conversion means and returned from the address conversion means. An adaptive distributed storage logical file access means for directly reading / writing via a disk connection network using a plurality of disk identifiers and an offset address on the disk as parameters, and each means of (a) to (c) above. Make it work on a computer Recording medium recording the eye of the program. 12. A plurality of computers, a computer connection network for connecting the computers, a plurality of disks in which data of one logical file is distributed and stored, and the computers can directly access the disks. Disk connection network,
File of the disk-shared cluster system including the
In a recording medium having a program for realizing a management system recorded therein , (a) one or a plurality of disks that operate on at least two of the plurality of computers and do not overlap each other, Address conversion means for determining a disk identifier and an offset address on the disk from the designated logical file identifier and offset address in the logical file and returning these to the request source; and (b) any one of the plurality of computers. A client program operating on a computer is started when reading / writing data of a logical file distributedly stored in the plurality of disks connected to the disk connection network,
From the specified logical file identifier and logical file offset address, the partial logical file identifier of the partial logical file that represents the set of data stored in the same disk among the distributed logical file data and the partial logical file offset Partial logical file address conversion means for determining addresses and returning them to the request source; and (c) a plurality of client programs operating on any computer of the plurality of computers connected to the disk connection network. It is activated when reading / writing the data of the logical file distributedly stored in the disk, and divides the read / write range of the specified logical file into units of the distributed storage, and Multiple logical file offset addresses that indicate the beginning and the specified logical file identification Invoking the partial logical file address converting means with the identifier as a parameter, and specifying the address converting means for issuing the address conversion request using the partial logical file identifiers returned from the partial logical file address converting means as parameters, Using the partial logical file identifier and the offset address in the partial logical file obtained from the partial logical file address conversion means as parameters, an address conversion request is issued to the plurality of address conversion means and at the same time, to the plurality of disks. A recording medium in which a program for causing each of the means (a) to (c) to function on the computer is recorded.