JP2000207370A - Distributed file management device and distributed file management system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a distributed film management system which can make appropriate load distribution by means of plural server computers for generating, referring to and updating files. SOLUTION: A distributed file management system is provided with server computers A, B, and C, client computer groups 108-110, and a network 101. The server computer A 105 is constituted of a storage device 115 which records partial files, a network interface 113, a partial file management section 111 which controls the write and read of the partial files, a status management section 114 which holds load information, and a distributed file management section 112. Since the arrangement of the partial files is determined, based on the load information of each server computer A, B, and C, the concentration of loads to a specific server computer can be avoided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a distributed file management apparatus and a distributed file management system for distributing and managing files to a plurality of terminals in a computer network system. In particular, a server that connects a plurality of server computers and client computers via a network
In a client-type computer network system, the present invention relates to a distributed file management device and a distributed file management system for distributing and managing files by a plurality of server computers.

[0002]

2. Description of the Related Art Conventionally, as a distributed file management technique applied to a server-client type computer network system (hereinafter, also simply referred to as "network system") in which a plurality of server computers and client computers are connected via a network, for example, JP-A-8-
There is a distributed file system disclosed in 77054 public information.

FIG. 30 shows a conventional distributed file system disclosed in JP-A-8-77054. In FIG. 30, this conventional distributed file system includes a plurality of server computers 3002, 3003, 3004, 30
05 and a plurality of client computers 3006, 300
7, 3008, and the plurality of server computers 3002,
And a network 3001 for connecting a plurality of client computers 3006, 3007, 3008.

The server computer 3002 holds a partial file A-1 (3002-1) of the distributed file A and a partial file B-1 (3002-3) of the distributed file B. The server computer 3003 has a partial file A-2 (3003-1) of the distributed file A and a partial file B-2 (3003-2) of the distributed file B.
Is held. Also, the server computer 3004 has:
Partial file A-3 (3004-1) of the distributed file A
And a partial file B-3 (3004-
2) is held.

[0005] The server computer 3005 includes partial files A-1 to A-3 and B-1 to B- stored in the server computers 3002, 3003, and 3004, respectively.
3, a distributed file management unit 3005-2 that manages
Reference / update request distribution information 3005-1 for distributing a partial file reference request or update request (hereinafter also simply referred to as “reference / update request”) from the client computers 3006, 3007, and 3008.
Holding.

On the other hand, a client computer 3006 is provided with a distributed file creation unit 3006-1 for creating a partial file of a distributed file in response to a distributed file creation request, and a server computer 30 in response to an update request for the distributed file.
The update request distribution unit 3006-2 determines the location of the partial file of the distributed file based on the reference / update request distribution information 3005-1 of the server 05, and the reference / update of the server computer 3005 in response to the reference request for the distributed file. A reference request distribution unit 3006-3 for determining the location of the partial file of the distributed file based on the request distribution information 3005-1. The other client computers 3007 and 3008 have the same configuration.

According to the above-described conventional distributed file system, for example, when a user of the client computer 3006 requests creation of a distributed file, the distributed file creation unit 3006-1 of the client computer 3006 is determined in advance. A server computer that creates a partial file of the distributed file is determined based on the distribution condition, and a partial file of the distributed file is created in the server computer. Then, at the same time when the partial file is created, reference / update request distribution information 3005-1 indicating which server computer created the partial file is generated. The reference / update request distribution information 3005-1 is transmitted to the server computer 3005 via the network 3001, and is held by the server computer 3005.

Further, for example, the client computer 300
6 users refer to or update distributed files
When an update request is made, first, the client computer 3
006 requests the distributed file management unit 3005-2 of the server computer 3005 to open the corresponding distributed file. The distributed file management unit 3005-2 of the server computer 3005 responds to the request to open the distributed file from the client computer 3006, and sends the distributed / reference request distribution information 3 on the distributed file to the client computer 3006 via the network 3001.
005-1 is transmitted. The update request distribution unit 3006-2 or the reference request distribution unit 3006-3 of the client computer 3006 holds a partial file of the distributed file based on the reference / update request distribution information 3005-1 received from the server computer 3005. A reference / update request is transmitted to the server computer.

As described above, in the conventional distributed file system, a distributed file is divided into a plurality of partial files, and processing (creation, reference, and update) on the distributed file is divided into processing in units of partial files. Even when a plurality of processing requests are concentrated on one distributed file, the load can be distributed without concentrating the load on one server computer.

[0010]

However, FIG.
In the conventional distributed file system as shown in (1), when a partial file constituting a distributed file is created, a server computer that creates the partial file is determined based on a predetermined fixed distribution rule. Therefore, the distribution of files to the server computers does not take into account realistic load information of the server computers. For this reason, actually, a request for creating a partial file may be further issued to a server computer which has a high load due to concentrated access and processing. For this reason, there is a case where the load of only a specific server computer becomes large, and there is a problem that an appropriate load distribution is not performed by a plurality of server computers.

Further, in the conventional distributed file system, once the partial files are allocated to the server computer once in a fixed manner based on the above allocation rules, after the allocation, that is, in the allocated server computer, Since the created partial file is not moved or copied after the creation of the partial file, there is a problem that if access to a specific partial file is concentrated, the load cannot be distributed by the access. .

Accordingly, an object of the present invention is to provide a distributed file management device and a distributed file management system that can perform appropriate load distribution by a plurality of server computers when creating, referencing, and updating a file.

[0013]

According to a first aspect of the present invention, there is provided a distributed file management apparatus comprising: a plurality of server computers having storage means for storing data; and one or more clients. A distributed file management device connected to a network to which a computer is connected, a status management means for holding and managing load information of a plurality of server computers, and a distributed file processing device for responding to a distributed file processing request from a client computer. And a distributed file management unit that determines a server computer that processes the partial file based on the load information managed by the state management unit.

In the above-mentioned distributed file management apparatus according to the present invention, the status management means notifies the other distributed file management apparatuses of the load information, and the load information of the server computer notified from the other distributed file management apparatuses. May be provided as external load information.

Here, the external state management means may notify the load information to another distributed file management apparatus by multicast. In addition, the external state management means includes:
Among other distributed file management devices, the load information can be notified to another adjacent distributed file management device.

According to another aspect of the present invention, there is provided a distributed file management system comprising: a plurality of server computers; one or more client computers; a plurality of server computers; In a distributed file management system including a network connecting client computers, each of the plurality of server computers stores storage means for storing a partial file constituting a part or all of the distributed file, and holds load information. A state management means for managing, and a server for specifying a partial file of the distributed file in response to a distributed file processing request from a client computer, and processing the partial file based on the load information managed by the state management means Distributed file management means for determining a computer.

In the above-described distributed file management system according to the present invention, the status management means notifies the load information to the other server computer, and the load information of the other server computer notified from the other server computer. An external state management unit that stores the external load information may be provided. Here, the external state management means can also notify other server computers of the load information by multicast.

In the above-described distributed file management device and distributed file management system according to the present invention, the plurality of server computers are grouped into one or a plurality of server computer groups, and the external state management means is one or more. The load information may be notified to another server computer belonging to the predetermined server computer group among the plurality of server computer groups, or the external state management means may be configured to notify the adjacent server computer of the one or more server computer groups. The load information may be notified to another server computer belonging to the server computer group that performs the load.

Further, the distributed file management means determines a partial file to be moved and another server computer of the destination based on the access information, load information, and external load information for each partial file, and stores the partial file in the server. It is also possible to provide a distributed file moving means for moving to a computer.

Here, the distributed file moving means detects that the load of the storage means included in the load information is larger than a predetermined value, and detects the partial file to be moved based on the external load information and the access information. The other server computer may be determined, and the partial file may be moved to the other server computer. Alternatively, the distributed file moving means may determine that the remaining capacity of the storage means included in the load information is a predetermined value. It is also possible to detect that the file is smaller, determine the partial file to be moved and another server computer of the destination based on the external load information and the access information, and move the partial file to the other server computer. Good. Further, the distributed file moving unit detects that the load of the network included in the load information is larger than a predetermined value, and moves the partial file to be moved based on the external load information and the access information and another server of the moving destination. The computer may be determined and the partial file may be moved to the other server computer. Alternatively, the distributed file moving means may include load information, external load information, access information, and information between the client computer and the plurality of server computers. The communication cost between the server computer having the storage unit holding the partial file and the client computer that has issued the processing request is determined based on the connection information of A server computer may be determined, and the partial file may be moved to the other server computer.

Further, the distributed file moving means can confirm in advance whether or not the partial file can be moved with respect to another server computer to which the partial file is to be moved. When a partial file is moved to another server computer, another partial file can be moved from another server computer to the server computer. Further, the distributed file migration means may list other server computers that can move the partial file, and determine another server computer to which the partial file is to be migrated based on the list.
Further, the distributed file moving means may send information on the server computer that created the partial file to another server computer along with the movement of the partial file.

The above-mentioned distributed file management means determines a partial file to be copied and another server computer at the copy destination based on the access information, load information, and external load information for each partial file. A distributed file copying means for copying to another server computer may be provided.

At this time, the distributed file copying means detects that the load of the storage means included in the load information is larger than a predetermined value, and determines the partial file to be copied based on the external load information and the access information. The other server computer may be determined and the partial file may be copied to the other server computer. The distributed file copying means may determine that the network load included in the load information is larger than a predetermined value. It is also possible to detect that there is, determine the partial file to be copied and another server computer of the destination based on the external load information and the access information, and copy the partial file to the other server computer. Alternatively, the distributed file copy unit may include a server computer having a storage unit holding a partial file based on load information, external load information, access information, and connection information between the client computer and the plurality of server computers. Obtaining the communication cost with the client computer that made the processing request, determining another server computer having a communication cost smaller than the communication cost, and copying the partial file to the other server computer. Can also.

Here, the distributed file copying means sends the partial file to another server computer to which the partial file is copied.
It is preferable to check in advance whether the partial file can be copied. Further, when the partial file is copied to another server computer, the distributed file copying means may copy another partial file from the other server computer to the server computer. In addition, the distributed file copying means,
Another server computer candidate to which the partial file can be copied may be listed, and another server computer to which the partial file is to be copied may be determined based on the list. Further, the distributed file copying means can select a plurality of other copy destination server computers to which the partial file is to be copied, and can simultaneously copy the partial file to the selected plurality of other server computers by multicast.

The load information managed by the state management means can include the capacity and load of the storage means, and the communication load between the network and a plurality of server computers.

Further, in the above-described distributed file management apparatus and distributed file management system according to the present invention, the server computer includes a partial file management unit that writes the partial file into the storage unit and reads the partial file from the storage unit. Can also.

In the case where the processing request from the client computer is a request for creating a distributed file, the distributed file management means divides the distributed file into a plurality of partial files and stores a server computer that holds the divided partial files. Determined based on the load information managed by the status management means, and if the processing request from the client computer is a request for referencing or updating a distributed file, the partial file The existence may be determined, and the server computer that processes the processing request may be determined based on the load information managed by the state management unit.

Further, the distributed file management means may include a partial file size determining means for determining the size of a partial file constituting a part or all of the distributed file based on information from the client computer. Or, the distributed file management means is provided with a partial file size determining means for determining the size of a partial file constituting part or all of the distributed file based on the type of data recorded in the distributed file. Is also good.

In the above-described distributed file management apparatus and distributed file management system according to the present invention, the distributed file management means determines a server computer in which a partial file is to be allocated based on load information of the server computer. Concentration of load on the server computer can be avoided.

Further, since the distributed file management means determines the server computer on which the partial file is to be allocated based on the load information of the other server computers, it is possible to prevent the load from being concentrated on a specific server computer.

Further, by moving the partial file to another server computer, it is possible to avoid concentration of the load on the storage means of a specific server computer and imbalance in the capacity of the storage means. In addition, by copying the partial file to another server computer, it is possible to avoid concentration of a load on the storage device of a specific server computer.

Also, since the size of the partial files constituting the distributed file can be changed as appropriate, the logical
It is possible to avoid dividing data that is related in content, for example, data for one frame of an image into a plurality of partial files.

[0033]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a distributed file management apparatus and a distributed file management system according to the present invention will be described below with reference to FIGS.

(Embodiment 1) FIG. 1 is a block diagram showing an example of a first embodiment of a distributed file management system according to the present invention. In FIG. 1, the distributed file management system includes a plurality of computer sites A102 and B103, each including a server computer such as a personal computer or a workstation and a client computer group including a plurality of client computers such as a personal computer or a workstation. And computer site C104, computer site A102, computer site B
103 and a network 101 such as a local area network or a wide area network for interconnecting the computer sites C104.

Here, the computer site A102 is provided with a plurality of server computers such as a personal computer and a workstation (in FIG. 1, "server computer A10
5 ") and client computers 1 to n (108-
1 to 108-n)
8 is provided. This computer site A102 has a plurality of server computers (in FIG. 1, "server computer A1
05 is shown) and the client computer group A 108 are connected via an internal network 131 such as Ethernet, for example, in the Internet domain.

Similarly to the computer site A102, the computer site B103 includes a plurality of server computers (only the "server computer B106" is shown in FIG. 1) and a client computer group B including a plurality of client computers.
The computer site C104 includes a plurality of server computers (in FIG. 1, "server computer C107").
Only) and a client computer group C110 composed of a plurality of client computers. Further, these computer sites B103 and C104
In the same manner as the computer site A102, a plurality of server computers (only the "server computer B106" and "server computer C107" are shown in FIG. 1) and a client computer group B109 and a client computer group C110 are respectively connected to the internal network. 132 and the internal network 133, for example, in the Internet domain.

The server computer A 105 includes a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 131 such as Ethernet.
Network interface 113 for connecting to
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
A state management unit 114 that monitors the load on the storage device 115, the remaining capacity of the storage device 115, and the load on the network interface 113, and retains information on these loads and capacity;
1, a state management unit 114, and a distributed file management unit 112 connected to the network interface 113.

The distributed file management unit 112 writes and reads a partial file in the partial file management unit 11.
Instruct 1 When creating a distributed file, the distributed file management unit 112 divides the distributed file into a plurality of partial files based on information obtained from the state management unit 114, and arranges (records) each partial file. Determine the server computer. When referencing or updating a previously created distributed file, the server computer in which a partial file of the relevant distributed file exists (is recorded) is determined.

Server computer B 106 and server computer C
107 has the same configuration as the server computer A105. In other words, the server computer B106 stores the storage device 1
20, a network interface 118, a partial file management unit 116, a state management unit 119, and a distributed file management unit 117. The server computer C107 includes a storage device 125, a network interface 123, and a partial file management unit 121.
, A state management unit 124 and a distributed file management unit 122.

FIG. 2 is a diagram showing an example of the configuration of a distributed file. In FIG. 2, the distributed file 201 includes a plurality of partial files 202-1 to 202-n.

In FIG. 1, each distributed file A, B,
This shows a state after C is created. That is, the storage device 115 of the server computer A105 records the partial file A1 (126a) of the distributed file A and the partial file B1 (126b) of the distributed file B. In the storage device 120 of the server computer B 106, a partial file C1 (126e) of the distributed file C and a partial file C2 (126f) of the distributed file C are recorded. In the storage device 125 of the server computer C107, a partial file A2 (126c) of the distributed file A and a partial file A3 (126d) of the distributed file A are recorded.

Next, the operation of the distributed file management system configured as described above will be described. In the following, a request to create a distributed file A is issued from the client computer 1 (108-1) of the client computer group A108 to the server computer A105, and partial files A1 to A3 as shown in FIG. 1 are created. A description will be given of the distributed processing in the case as an example. Here, each of the storage devices 115, 120, and 125 illustrated in FIG. 1 has a plurality of storage units or storage areas (hereinafter, also simply referred to as “storage units”). These plurality of storage units may be physically one recording medium, or may be a plurality of recording media.

In FIG. 1, first, a request to create a distributed file A is issued from the client computer 1 (108-1) to the server computer A105. The request for creating the distributed file A is sent to the distributed file management unit 112 of the server computer A 105 via the internal network 131 and the network interface 113 of the server computer A 105.
Received by

FIG. 3 is a flowchart showing an operation algorithm of the distributed file management unit when a request for creating a distributed file is received. Hereinafter, the detailed operation of the distributed file management unit 112 will be described with reference to FIGS.

Upon receiving the request for creating the distributed file A, the distributed file manager 112 first obtains the load information managed by the state manager 114 (step 301).

FIG. 4 is a diagram showing a load information table 401 managed by the state management unit 114. In FIG. 4, the load information table 401 includes a storage device load information table 4
02 and a network load information table 403.

This storage device load information table 402
Storage device 115 connected to server computer A105
"Storage device identifier" for identifying a plurality of storage units
, “Load” indicating load information [%] of each storage unit, and “remaining capacity” indicating remaining capacity [Mbytes] of each storage unit. Here, the “load” of each storage unit of the storage device 115 indicates what percentage of the maximum transfer rate of each storage unit of the storage device 115 is used.

The network load information table 40
No. 3 indicates to which computer site (destination site) data to be transmitted on the network 101 via the network interface 113 is transmitted, and how much bandwidth (used communication bandwidth [Mbps]) is used. Which computer site (sender site) the data is being received from, and how much bandwidth (communication bandwidth [Mbps]) is being used to receive it Is shown. The item of “transmission source site” indicates the computer site of the data transmission source, and the item of “transmission destination site” indicates the computer site of the data transmission destination. That is, SiteA stores the computer site A105 in SiteB
Indicates a computer site B106, and SiteC indicates a computer site C107. Here, the transmission source site and the transmission destination site are collectively called a network link (hereinafter, also simply referred to as a “link”). The item of “used communication bandwidth” indicates a communication bandwidth [Mbps] used between the transmission source site and the transmission destination site.

That is, when the distributed file management unit 112 receives the request to create the distributed file A, the distributed file management unit 112, as shown in FIG. 4, for example, stores the storage device load information table 402 acquired from the state management unit 114.
Based on this information, information is obtained that the load on the storage unit of the storage device identifier DiskID1 is 20 [%] and the remaining capacity is 10 [Mbytes] (step 301).

Next, the distributed file management unit 112 stores the storage device load information table 4 obtained from the state management unit 114.
02, a storage unit having sufficient remaining capacity and a load lower than a predetermined threshold is selected from the storage devices 115 connected to the server computer A105, and Partial files are sequentially allocated to them. Here, for example, 80 [%] may be used as the threshold. However, this threshold can be appropriately determined according to the configuration of the storage device 115 and the like. The size of the partial file is desirably a fixed length, and may be the same for all server computers. At this time, if all the partial files A1 to A3 have been allocated to the storage unit of the storage device 115, the process of step 304 is performed.
On the other hand, all the partial files A1 to A3 are stored in the storage device 115.
If not assigned to the storage unit of step 3,
03 is performed (step 302).

In the case of the present embodiment (FIG. 1), all the partial files A1 to A3 have not been allocated to the storage unit of the storage device 115 (step 302), so the partial file A1 (126a) of the distributed file A Is allocated to the server computer A105, and the remaining partial files A2 and A3 are allocated to another server computer C107.

The distributed file management unit 112 executes step 3
In order to allocate the partial files A2 and A3 that were not allocated in 02 to other server computers, the internal network 131 and the network 1
01, an inquiry is made as to whether the partial files A2 and A3 can be created. The other server computer that has received the inquiry checks the load information table 401 of its own state management unit and returns a response as to whether the partial files A2 and A3 can be created (step 303). The exchange of the inquiry and the response signal of the creation of the partial file is performed by the distributed file management unit of each server computer via the network interface and the network.

In the case of the present embodiment (FIG. 1), the distributed file management unit 112
13 and the partial file A via the network 101
It inquires of the distributed file management unit 122 of the server computer C107 of the computer site C104 whether or not the second file 126c (126c) and the partial file A3 (126d) can be created. Distributed file management unit 12 of server computer C107
2 makes a determination similar to that in step 302 based on the load information table 401 obtained from the state management unit 124, and returns a response as to whether the partial file A2 (126c) and the partial file A3 (126d) can be created. Is performed on the server computer A 105 (step 303). In the case of FIG. 1, the partial file A2 (126c) and the partial file A are stored in the storage device 125 of the server computer C107.
3 (126d) can be created.

Next, the distributed file management unit 112 of the server computer A stores the partial file A1 assigned to the storage device 115 of the server computer A, the partial file A2 assigned to the storage device 125 of the other server computer C, A3
(Step 30)
4).

FIG. 5 shows a distributed file management table 501.
FIG. FIG. 6 is a diagram showing the partial file management table 601. In FIG. 5, the distributed file management table 501 includes items of a “distributed file identifier” for identifying a distributed file and a “partial file identifier list” for identifying a partial file constituting the distributed file. . In FIG. 6,
The partial file management table 601 includes items of “partial file identifier” for identifying a partial file and “location” indicating the location of the partial file. Here, the “partial file identifier” shown in FIG. 6 corresponds to the “partial file identifier” constituting the “partial file identifier list” shown in FIG.

In the case of this embodiment (FIG. 1), for example, looking at distributed file A, in FIG. 5, distributed file A is divided into partial file A1 (126a), partial file A2 (126c), and partial file A3. (126
d). In FIG. 6, the location of the partial file A1 (126a) is "f".
ile: // siteA / serverA / DiskI
D1 / (server computer A105 of computer site A102
Storage device identifier DiskID1) ”, and the location of the partial file A2 (126c) is“ file: // s
itemC / serverC / DiskID2 / (storage device identifier DiskID2 of the server computer C107 of the computer site C104) "and the partial file A3 (12
The location of 6d) is "file: // siteC / se
rverC / DiskID2 / (computer site C104
Storage device DiskID2) of the server C107 ”).

Next, when creating the partial file A1 in the storage device 115 of the server computer A, the distributed file management unit 112 transmits data from the client computer 1 (108-1) via the partial file management unit 111. The partial file A1 of the distributed file A is written to the storage device 115.
Is created. When recording the partial files A2 and A3 in another server computer C107, the distributed file management unit 112 requests the distributed file management unit 122 of the recording server computer C107 to record the partial files A2 and A3. At the same time, an instruction is issued to the client computer 1 (108-1) that has issued the request to create the distributed file A, and the data is directly transmitted from the client computer 1 (108-1) to the server computer C107 that performs recording. . In the server computer C107 that has received the request, the distributed file management unit 122 registers the partial files A2 and A3 in the partial file management table 601.
In this way, partial files A2 and A3 are created on another server computer C107 (step 305).

As described above, when the partial file A1 (126a) of the distributed file A is created, the data from the client computer 1 (108-1) is transferred to the server computer A1.
This is done by damaging the storage device 115 of FIG. Also, the partial file A2 (126
The creation of c) and the partial file A3 (126d) is performed by sending data from the client computer 1 (108-1) to the direct server computer C107 and writing the data to the storage device 125 of the server computer C107.

If it is determined in step 303 that the creation of the partial file is not possible in all of the other server computers, the distributed file management unit 112 sends the distributed file creation request to the client computer 1 ( 108
Inform -1) that the creation of the distributed file has failed (step 306).

On the other hand, in step 302, if all the partial files of the distributed file can be created in the storage device of the server computer, the distributed file management unit 112 manages the distributed file allocated to the storage device 115. Information for the distribution file management table 501 shown in FIG. 5 and the partial file management table 601 shown in FIG.
(Step 307).

Next, the distributed file management unit 112 sends the client computer 1 via the partial file management unit 111.
The data from (108-1) is written to the storage device 115, and all partial files are created (step 30).
8).

As described above, according to the distributed file management system of the present invention, since the creation of the distributed file is performed in consideration of the load of each server computer, the load can be appropriately distributed.

Next, a case where a distributed file reference / update request is issued from the client computer 1 (108-1) in the client computer group A 108 to the server computer A 105 will be described.

First, the client computer 1 (108-
The reference / update request for the distributed file A issued from 1) is transmitted to the distributed file management unit 1 via the network interface 113 in the server computer A105.
12 received.

FIG. 7 is a flowchart showing an operation algorithm of the distributed file management unit when a request for referencing / updating a distributed file is received. Hereinafter, using FIG.
The detailed operation of the distributed file management unit 112 will be described. Further, in the following, a specific operation will be described using the example of the processing of the reference / update request for the distributed file A.

First, the distributed file management unit 112 references / updates the distributed file management table 501 and the partial file management table 601 based on the distributed file reference / update request from the client computer 1 (108-1).
The partial file to be updated is specified, and the location of the partial file is obtained (step 701).

Here, the client computer 1 (108-
In the case of the reference / update request for the distributed file A from 1), the distributed file management unit 112 determines that the distributed file A is the partial file A1 (126a) and the partial file A2 (126) based on the distributed file management table 501.
c), it is understood that the file is composed of the partial file A3 (126d). Also, the partial file management table 6
01, the location of the partial file A1 (126a) is “file: // siteA / serverA /
DiskID1 / (storage device identifier DiskID1 of server computer A105 of computer site A102) ", and the location of partial file A2 (126c) is" fi
le: // siteC / serverC / DiskID
2 / (storage device identifier DiskID2 of server computer C107 of computer site C104) "and the location of partial file A3 (126d) is" file: /// si
teC / serverC / DiskID2 / (storage device DiskID of server C107 of computer site C104
2) ".

The distributed file management unit 112 determines in step 7
01, all the partial files to be referenced / updated are stored in the server computer A1 from the location of the partial file obtained in step S01.
It is determined whether the file exists in the storage device 115 of the server computer 05 or whether some or all of the partial files exist in another server computer (step 702).

In the case of the distributed file A, the partial file A1 (126a) exists in the storage device 115, and the partial file A2 (126c) and the partial file A3 (1
26d) is the server computer C1 of the computer site C104.
07 in the storage device 125.

Next, if all the partial files do not exist in the storage device 115 of the own server computer A 105 (step 702), the partial files to be referred / updated based on the location of the partial files obtained in step 701 The presence of the partial file is confirmed with another server computer in which is recorded (step 703).

Here, in the case of the distributed file A, the partial file A2 (126c) and the partial file A3 (12
The presence of 6d) is confirmed by the distributed file management unit 122 of the server computer C107 of the computer site C104.

When the existence of the partial file is confirmed in step 703, the distributed file management unit 112 determines that the partial file to be referred to / updated has its own server computer A105.
Of the storage device 115 via the partial file management unit 111 based on a reference / update request from the client computer 1 (108-1).
Read (reference) and write (update) the partial file existing in the file. If the partial file to be referenced / updated exists in the storage device of another server computer, the distributed file management unit 112
The server computer holding the partial file to be updated is requested to refer to / update the partial file. At the same time, the distributed file management unit 112 transmits the reference / update request to the client computer 1 (108-1).
The server computer holding the partial file to be updated is instructed to make a reference / update request directly (step 7).
04).

In the case of the distributed file A, the partial file A1 (126a) is stored in the server computer A105 of the computer site A102, and the partial file A2 (126c) and the partial file A3 (126d) are stored in the computer site C104.
Server computer C107. For the reference / update request for the partial file A1 (126a), the distributed file management unit 112 performs a reference / update process on the storage device 115 via the partial file management unit 111.
On the other hand, the reference / update request for the partial file A2 (126c) and the partial file A3 (126d) is made directly between the client computer 1 (108-1) and the server computer C107 that have issued the reference / update request. become.

If the existence of the partial file is not confirmed in step 703, the distributed file management unit 112
Notifies the client computer 1 (108-1) that made the request to refer to / change the distributed file that the reference / update to the distributed file has failed (step 705).

On the other hand, in step 702, if all the partial files exist in the storage device 115 of the server computer A 105, the distributed file management unit 112 responds to the reference / update request from the client computer 1 (108-1). Storage device 1 via the partial file management unit 111 based on the
15 is read (referenced) or written (updated) to the partial file (step 70).
6).

As described above, according to the above-described embodiment, when a request from a client computer to a server computer is to create a distributed file, the distributed file is divided into a plurality of partial files, and the load on the server computer is reduced. A server computer that creates each partial file is determined for each partial file based on the information, and a distributed file creation process is performed. When the request from the client computer is to refer to / update the distributed file, the server computer in which the partial file constituting the distributed file exists is specified and distributed on one or more server computers. The client computer handles the partial file as one distributed file. In this way, it is possible to eliminate the concentration of the load on a specific server computer when a request to create / reference / change a distributed file from a client computer to a server computer.

(Embodiment 2) FIG. 8 is a block diagram showing an example of a second embodiment of the distributed file management system according to the present invention. 8, the same components as those in FIG. 1 are denoted by the same reference numerals. The distributed file management system shown in FIG. 8 includes a plurality of computer sites A 802 and B 803 each including a server computer such as a personal computer and a workstation and a client computer group including a plurality of client computers such as a personal computer and a workstation. And a computer site C804, and a network 101 such as a local area network or a wide area network interconnecting the computer site A802, the computer site B803, and the computer site C804.

Here, the computer site A 802 is connected to a plurality of server computers such as a personal computer and a workstation (in FIG. 8, “server computer A 80
5 ") and client computers 1 to n (108-
1 to 108-n)
8 is provided. This computer site A802 includes a plurality of server computers (in FIG. 8, "server computer A8
05 is shown) and the client computer group A 108 are connected via an internal network 131 such as Ethernet, for example, in the Internet domain.

Further, similarly to the computer site A 802, the computer site B 803 includes a plurality of server computers (only the “server computer B 806” is shown in FIG. 8) and a client computer group B including a plurality of client computers.
The computer site C804 includes a plurality of server computers ("server computer C807" in FIG. 8).
Only) and a client computer group C110 composed of a plurality of client computers. Further, the computer site B 803 and the computer site C 804
, Like the computer site A 802, connects a plurality of server computers (only the “server computer B 806” and “server computer C 807” are shown in FIG. 8) to the client computer group B 109 and the client computer group C 110, respectively. 132 and the internal network 133, for example, in the Internet domain.

The server computer A 805 includes a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 131 such as Ethernet.
Network interface 113 for connecting to
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
A state management unit 814 that monitors the load on the storage device 115, the remaining capacity of the storage device 115, and the load on the network interface 113, and retains information on these loads and capacity;
1, a state management unit 814, and a distributed file management unit 112 connected to the network interface 113.

The state management unit 814 notifies the other server computers of the load information, and holds the external load information notified from the other server computers.
It has.

Server computer B 806 and server computer C
807 has the same configuration as the server computer A 805. That is, the server computer B 806 stores the storage device 1
20, a network interface 118, a partial file management unit 116, a state management unit 819 having an external state management unit 812, and a distributed file management unit 117. In addition, the server computer C807
Storage device 125 and network interface 123
, Partial file management unit 121, external state management unit 81
3 and the distributed file management unit 1
22.

The difference between the distributed file management system shown in FIG. 8 and the distributed file management system shown in FIG. 1 is that the state management units 814, 819, and 82 shown in FIG.
No. 4 is provided with external state management units 811, 812 and 813 for notifying the load information to the other server computers and holding the external load information notified from the other server computers.

FIG. 9 shows the external state management units 811, 812,
8 shows an example of an external load information table 901 managed by 813. In FIG. 9, the external load information table 90
1 is composed of items of “server computer location” indicating the location of the server computer and “storage device load information” indicating the load information of the storage device of the server computer indicated by the server computer location. The “storage device load information” includes items of “storage device identifier” for identifying the storage device, “load” indicating the load of the storage device, and “remaining capacity” indicating the remaining capacity of the storage device. Have been.

External state management units 811, 812, 813
Performs the following operation to notify the external load information from the external load information table 901 to another server computer.

First, the external state management units 811, 812, 8
Reference numeral 13 denotes the state management unit 8 periodically or at a predetermined timing.
14, 819, and 824, or inquires of the information shown in the storage device load information table 402, or notifies the state management units 814, 819, and 824 of a change in the state of the information shown in the storage device load information table 402 By receiving the information, information of the storage device load information table 402 managed by the state management units 814, 819, and 824 is obtained.

Next, the external state management units 811, 812, 8
13 is a network interface 11
The storage device load information is notified to each server computer via 3, 118, and 123. In the server computers that have received the notification, the network interfaces 113 and 11 respectively
8, 123, the external state management units 811, 812,
813 receives the storage device load information and records this information in each external load information table 901.

In the distributed file management system configured as described above, an example in which a request for creating a distributed file A is issued from the client computer 1 (108-1) in the client computer group A108 to the server computer A805. It will be described as.

First, the client computer 1 (108-
The request for creating the distributed file A issued from 1) is received by the distributed file management unit 112 via the internal network 131 and the network interface 113 of the server computer A 805.

FIG. 10 is a flowchart showing an operation algorithm of the distributed file management unit when a request for creating a distributed file is received. Hereinafter, the detailed operation of the distributed file management unit will be described with reference to FIG. In this embodiment, the processing of steps 302 and 303 shown in FIG. 3 described in the first embodiment can be integrated and processed in one step 1002.

First, the distributed file management section 112 loads the load information table 401 managed by the state management section 814.
And obtains each information from the external load information table 901 managed by the external state management unit 811 (step 100).
1).

The state management unit 114 manages a load information table 401 as shown in FIG. 4, a load information table 401 includes a storage device load information table 402 and a network load information table 40.
3.

This storage device load information table 402
Storage device 115 connected to server computer A105
"Storage device identifier" for identifying a plurality of storage units
, “Load” indicating load information [%] of each storage unit, and “remaining capacity” indicating remaining capacity [Mbytes] of each storage unit. Here, the “load” of each storage unit of the storage device 115 indicates what percentage of the maximum transfer rate of each storage unit of the storage device 115 is used.

The network load information table 40
No. 3 indicates to which computer site (destination site) data to be transmitted on the network 101 via the network interface 113 is transmitted, and how much bandwidth (used communication bandwidth [Mbps]) is used. Which computer site (sender site) the data being received is sent from, and how much bandwidth (used communication bandwidth [Mbps]) is being used to receive the data Is shown. The item of “transmission source site” indicates the computer site of the data transmission source, and the item of “transmission destination site” indicates the computer site of the data transmission destination. That is, SiteA stores the computer site A105 in SiteB
Indicates a computer site B106, and SiteC indicates a computer site C107. Here, the transmission source site and the transmission destination site are collectively called a network link (hereinafter, also simply referred to as a “link”). The item of “used communication bandwidth” indicates a communication bandwidth [Mbps] used between the transmission source site and the transmission destination site.

The external state management section 811 manages an external load information table 901 as shown in FIG.

Here, when creating the distributed file A,
The distributed file management unit 112 determines that the “storage device identifier” is “Dis” from the load information table 401 of the status management unit 814.
“Load” of the storage unit of the storage device 115 indicated by kID1
Is 20 [%] and the "remaining capacity" is 10 [Mbytes]. In addition, the distributed file management unit 112 determines from the external load information table 901 of the external state management unit 814 that the “storage device identifier” of the server computer B 806 of the computer site B 803 is the “load” Is 49 [%]
In this case, the “remaining capacity” is 1000 [Mbytes], and the “storage device identifier” of the server computer C 807 of the computer site C 804 is the storage device 125 indicated by the Disk ID 1.
"Load" of the storage unit is 30%, and "remaining capacity" is 30%.
It can be obtained that the information is 00 [Mbytes].

Next, based on the storage device load information table 402 obtained from the status management unit 814, the distributed file management unit 112 first selects one of the storage units of the storage device 115 connected to the server computer A 805. The storage unit that satisfies the condition that the “remaining capacity” is equal to or more than the predetermined capacity and the “load” is lower than the predetermined threshold is selected, and the partial files obtained by dividing the distributed file are sequentially stored in each storage unit that satisfies the condition. Assign. When all the partial files cannot be allocated to the storage device 115, the distributed file management unit 112 sets the “remaining capacity” of the storage unit based on the information in the external load information table 901 obtained from the external state management unit 811. Another server computer having a storage device having a storage unit having a storage capacity equal to or more than a predetermined capacity and having a “load” lower than a predetermined threshold is selected. Then, the partial files that have not been allocated are sequentially allocated to the storage unit of the storage device of the other server computer, and this allocation is notified to the other server computer. Then, it is determined whether or not all the partial files have been created by allocating them to the storage units of the respective storage devices of the server computer (step 1002).

Here, when creating the distributed file A,
The distributed file management unit 112 stores the partial file A1 (126a) in the storage device 1 of the server computer A805 based on the storage device load information table 401 of the state management unit 814.
Assigned to 15 predetermined storage units. Further, the distributed file management unit 112, based on the storage device load information of the external load information table 901 of the external state management unit 811, stores the partial file A2 (126c) and the partial file A3 (126
d) is assigned to each storage unit of the storage device 125 of the server computer C807.

As described above, the server computer A 805 obtains information on the load and the remaining capacity of the other server computer from the external state management unit 811 and inquires of the other server computer whether the partial file can be created. Without doing this, it is possible to determine the server computer on which the partial file is to be placed (stored) in consideration of the load and remaining capacity of other server computers.

Next, the distributed file management unit 112 registers the distributed file management information in the distributed file management table 501 as shown in FIG. 5 and the partial file management table 601 as shown in FIG. 6 ( Step 100
3). In FIG. 5, the distributed file management table 501
Is composed of items of a “distributed file identifier” for identifying a distributed file and a “partial file identifier list” for identifying a partial file constituting the distributed file. In FIG. 6, the partial file management table 601 includes items of “partial file identifier” for identifying the partial file and “location” indicating the location of the partial file. Here, the “partial file identifier” shown in FIG. 6 corresponds to the “partial file identifier” constituting the “partial file identifier list” shown in FIG.

Here, in the case of the distributed file A, in FIG. 5, the distributed file A is the partial file A1 (126
a), partial file A2 (126c), partial file A
3 (126d). In FIG. 6, the location of the partial file A1 (126a) is “file: // siteA / serverA”.
/ DiskID1 / (storage device identifier DiskID1 of server computer A105 of computer site A102) ", and the location of partial file A2 (126c) is" fi
le: // siteC / serverC / DiskID
2 / (storage device identifier DiskID2 of server computer C107 of computer site C104) "and the location of partial file A3 (126d) is" file: /// si
teC / serverC / DiskID2 / (storage device DiskID of server C107 of computer site C104
2) ".

Next, when recording the partial file in the storage unit of the storage device 115, the distributed file management unit 112 transmits the data from the client computer 1 (108-1) via the partial file management unit 111. Storage device 115
In a predetermined storage unit. When recording a partial file in the storage device of another server computer, the distributed file management unit 112 requests the distributed file management unit of the recording server computer to record the partial file. At the same time, the distributed file management unit 112 requests the client computer 1 (108-
Instruct 1) to send the data directly to the server computer that performs the recording. Distributed file management unit 11
The server computer receiving the request from the client computer 2 receives the data of the partial file from the client computer 1 (108-1) and records it in a predetermined storage unit of the storage device. Further, the distributed file management unit of the server computer registers information of the partial file in the partial file management table 601 of the partial file management unit. In this way, a partial file is created on another server computer (step 100).
4).

In the case of the distributed file A, the partial file A1
(126a) is created by the client computer 1 (108
This is performed by writing the data from -1) to a predetermined storage unit of the storage device 115. Partial file A2
(126c) and the partial file A3 (126d) are created by directly sending predetermined data from the client computer 1 (108-1) to the server computer C807 and writing the data to the predetermined storage unit of the storage device 125 of the server computer C807. Done by

On the other hand, if it is determined in step 1002 that the partial file cannot be created in any of the server computers, the distributed file management unit 112 sends a request to the client computer 1 (108-1) requesting the creation of the distributed file. It notifies that the creation of the distributed file has failed (step 1005).

The creation of the distributed file has been described above. The case where the client computer issues a request for referencing / updating the distributed file to the server computer will be described in the first embodiment (FIG. 7). The same is true.

As described above, in the second embodiment of the present invention, the status management units 814, 819, and 824 notify other server computers of the load information, and receive the notification from the other server computers. By providing the external state management units 811, 812, and 813 that hold the external load information, the distributed file management units 112, 117, and 122 allocate partial files of the distributed file based on the load information of other server computers. Server computer to be determined,
Concentration of load on a specific server computer can be avoided.

The external state management units 811, 812, 8
When the storage device load information is notified from 13 to each server computer, unicast or multicast may be used. In particular, when the multicast is used, the storage device load information can be notified to all the server computers all at once, and the communication amount for the notification can be reduced.

Also, the server computers to be notified can be divided into a plurality of groups in advance, and each of the server computers belonging to each group can be notified by unicast.
Also, each group can be notified by multicast. In this way, the amount of communication for notification can be reduced.

Furthermore, the server computer to be notified can be limited to the adjacent server computer, that is, the server computer directly connected to the network, and can be notified by unicast or multicast. As a result, the amount of communication for notification can be reduced.

(Embodiment 3) FIG. 11 is a block diagram showing an example of a third embodiment of the distributed file management system according to the present invention. In FIG. 11, the same components as those in FIG. 8 are denoted by the same reference numerals. The distributed file management system shown in FIG. 11 includes a plurality of computer sites A 1102 and a computer site B 110 each including a server computer such as a personal computer and a workstation and a client computer group including a plurality of client computers such as a personal computer and a workstation.
3, computer site C1104, and computer site A1
102, computer site B 1103, and computer site C
Networks 101 such as a local area network and a wide area network connecting
And

Here, the computer site A 1102 is provided with a plurality of server computers such as a personal computer and a workstation (in FIG. 11, “server computer A 1
105 ") and client computers 1 to n (10
8-1 to 108-n) Client computer group A
108. This computer site A1102
Connects a plurality of server computers (in FIG. 11, only "server computer A1105") and a group of client computers A108 to an internal network 13 such as Ethernet.
1 and is, for example, an Internet domain.

Further, similarly to the computer site A 1102,
The computer site B 1103 has a plurality of server computers (FIG. 1).
1, only "server computer B1106" is shown.)
And a client computer group B109 composed of a plurality of client computers.
It comprises a plurality of server computers (only the "server computer C1107" is shown in FIG. 11) and a client computer group C110 composed of a plurality of client computers. Further, the computer site B 1103 and the computer site C 1104 are, similarly to the computer site A 1102, a plurality of server computers (in FIG. 11, only “server computer B 1106” and “server computer C 1107”) and a client computer group B 109. And the client computer group C110 via an internal network 132 and an internal network 133, respectively.
For example, the Internet domain.

The server computer A 1105 includes a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 13 such as Ethernet.
Network interface 113 for connecting to 1
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
A state management unit 814 that monitors the load on the storage device 115, the remaining capacity of the storage device 115, and the load on the network interface 113, and retains information on these loads and capacity;
1, a state management unit 814, and a distributed file management unit 1112 connected to the network interface 113.

The state management unit 814 notifies the other server computers of the load information, and holds the external load information notified from the other server computers.
It has.

Further, the distributed file management unit 1112 stores the access information and the load information table 401 for each partial file.
And information of the external load information table 901 based on
Distributed file mover 1131 that determines a partial file to be moved and moves the partial file to another server computer
It has.

The server computer B 1106 and the server computer C 1107 have the same configuration as the server computer A 1105. That is, the server computer B 1106 is connected to the storage device 120 and the network interface 118
, Partial file management unit 116, external state management unit 81
State management unit 819 provided with distributed file transfer unit 1
132 and a distributed file management unit 1117 having the same. The server computer C1107 is connected to the storage device 125 and the network interface 123.
, Partial file management unit 121, external state management unit 81
3 and the distributed file transfer unit 1
133 and a distributed file management unit 1122 having

The difference between the distributed file management system shown in FIG. 11 and the distributed file management system shown in FIG. 8 is that the distributed file management unit 111 shown in FIG.
2, 1117 and 1122 determine the partial file to be moved from the access information for each partial file and the information in the load information table 401 and the external load information table 901 and move the distributed file to another server computer. It is provided with parts 1131, 1132, and 1133.

FIG. 12 shows an example of the access information table 1201 for each partial file managed by the status management unit 814. In FIG. 12, the access information table 1201 includes “partial file identifier” for identifying a partial file and “access information per unit time”.
It consists of items. The “access information per unit time” includes “access source site identifier”, which is information of a site where a client computer accessing the partial file exists, and “access count” indicating the number of accesses to the partial file. Consists of items. The access information table 1201 is stored in the state management unit 814.
, Is continuously updated every unit time.

FIG. 13 shows an example of the partial file management table 1301 managed by the distributed file management unit 1112. In FIG. 13, the partial file management table 1301 includes a “partial file identifier” for identifying the partial file, a “location” indicating the location of the partial file, and an “original location” indicating the location where the partial file was first created. ". FIG.
In step 3, when the partial file is created, the information indicating the "location" and the "original location" are the same, but as the partial file moves to another server computer, the information of the "location" becomes Change. The partial file management table 1301 shown in FIG. 13 is different from the partial file management table 601 shown in FIG.
The items are added.

In the distributed file management system configured as described above, distributed file A, distributed file B,
After the distribution file C is created by the server computer A 1105 as described above, a process of moving a partial file of each distribution file will be described in detail.

FIG. 14 shows an example of the contents of the partial file management table 1301 of the distributed file A, the distributed file B, and the distributed file C created by the server computer A1105. In FIG. 14, FIG.
Shows the partial file management table 1401 of the server computer A1105, (B) shows the partial file management table 1402 of the server computer B1106, and (C) shows the partial file management table 14 of the server computer C1107.
03 is shown. The partial file management table 1401 of FIG.
The location of each partial file indicated by 2, A3, B1, C1, C2 and the original location are shown. In the partial file management table 1402, the partial file identifiers indicate the location of each partial file indicated by C1 and C2,
The original location is shown. In the partial file management table 1403, the partial file identifiers are A2, A
The location of each partial file indicated by 3 and the original location are shown. Here, FIG. 14 shows a state before the movement of each partial file. Therefore, in all the partial files, the location and the original location match.

An operation algorithm of the distributed file mover 1131 of the server computer A 1105 when a partial file is moved in the state shown in FIG. 14 will be described.

FIG. 15 shows an operation algorithm of the distributed file transfer unit 1131. First, the distributed file transfer unit 1
131 monitors the information in the load information table 401 (FIG. 4) managed by the state management unit 814 at regular time intervals (step 1501).

The distributed file mover 1131 sets a predetermined threshold value (for example, a value such as 80%, etc., which is determined arbitrarily according to the system configuration) for the “load” of a certain storage device. When it is detected that the number exceeds the limit (step 1501), the "partial file identifier" of the partial file included in the detected storage device is searched for by referring to the partial file management table 1301. The “access information per unit time” of the found “partial file identifier” is obtained from the access information table 1201. "Access information per unit time" obtained here
Is compared for each "partial file identifier", and the "partial file identifier" having the largest "number of accesses" is selected (step 1502). That is, the source file is selected. For example, here, the partial file A1 (12
It is assumed that 6a) is selected.

Next, based on the external load information table 901 (FIG. 9), the distributed file migration unit 1131 identifies a server computer having a storage device with a sufficient “remaining capacity” and a “load” lower than a predetermined value. select. Then, the distributed file moving unit 1131 checks whether the partial file can be moved to the selected server computer, and determines a server computer that can be moved (Step 1502). That is, the destination server computer is selected. For example, here
Storage device identifier DiskID of server computer C1107
It is assumed that the storage unit of the storage device 125 indicated by 2 is selected.

Next, the partial file is moved based on the information obtained by the selection of the source partial file and the selection of the destination server computer (step 150).
3).

In the above example, in step 1502, the partial file A1 (126a)
Since the server computer C1107 has been selected as the destination server computer, the distributed file migration unit 1131 of the source server computer A1105 performs the partial file management unit 11
1, the partial file A1 (1
Read 26a). This read partial file A
1 (126a) together with information on the "original location" of the partial file A1 (126a) (FIG. 13)
The data is transmitted to the network 101 via the network interface 113 and the internal network 131.

On the other hand, in the transfer destination server computer C1107, the distributed file transfer unit 1133 checks the partial file A1 transmitted from the transfer source server computer A1105.
(126 a) and its “original location” information are received from the network 101 via the internal network 133 and the network interface 123. The partial file management unit 121 receives the received partial file A
1 (126a) is written to the storage device 125. The “original location” of the partial file A1 (126a)
Is registered in the partial file management table (FIG. 14C) of the partial file management unit 121.

Thereafter, the destination server computer C1107
Corresponds to the server computer A1105 of the transfer source and the server computer indicated in the "original location" (FIG. 13) (in this example, the server computer A1105, that is, the server computer A1105 in which the transfer source and the original are the same). , The transfer of the partial file A1 (126a) is completed. In the source server computer A1105 and the server computer indicated by the "original location", the partial file A1 registered in the partial file management table is used.
The information of (126a) is rewritten.

FIG. 16 shows the state of the partial file management tables 1401, 1402, 1403 shown in FIG. 14 after the partial file A1 (126a) has been moved as in the above example. 16A shows a partial file management table 1601 of the server computer A 1105, FIG. 16B shows a partial file management table 1602 of the server computer B 1106, and FIG.
1107 shows a partial file management table 1603 of FIG. That is, each of the partial file management tables 1601, 1602, and 1603 in FIGS. 16A to 16C correspond to each of the partial file management tables 1401, 1402, and 1403 in FIGS. . Here, in the partial file management table 1601, “partial file identifiers” are A1, A2, A3, B1, C1, C2.
The "location" and "original location" of each partial file indicated by "." In the partial file management table 1602, the "location" and "original location" of each partial file whose "partial file identifier" is indicated by C1 and C2 are shown. Partial file management table 1
In 603, “partial file identifiers” are A1, A2, A
The “location” and “original location” of each partial file indicated by 3 are shown. Here, the partial file management tables 1601, 1602, 160 shown in FIG.
3 and the partial file management table 1 shown in FIG.
The difference between the states 401, 1402, and 1403 is due to the movement of the partial file A1 (126a) from the server computer A1105 to the server computer C1107. That is, the difference between FIG. 16 and FIG.
In the partial file management table 1401 of FIG.
"Location" of partial file A1 (126a) is "fil"
e: // siteA / serverA / DisiID1
In the partial file management table 1601 of FIG. 16A, the information registered as “/” indicates that the “location” of the partial file A1 (126a) is “file: //”.
siteC / serverC / DiskID2 / ”and that the item of the partial file A1 is added in the partial file management table 1603 of FIG. 16C.

FIG. 17 further shows the state of the server computer B1 from the state shown in FIG.
The distributed file mover 1132 of the 106 moves the partial file C1 (126e) to the server computer C1107, and the distributed file mover 11 of the server computer C1107.
Reference numeral 33 denotes a partial file management table when the partial file A1 (126a) is moved to the server computer B1106.

17A shows a partial file management table 1701 of the server computer A 1105, FIG. 17B shows a partial file management table 1702 of the server computer B 1106, and FIG.
1107 shows a partial file management table 1703 of FIG. In the partial file management table 1701, "partial file identifiers" are A1, A2, A3, B1, C1, C
The “location” and “original location” of each partial file indicated by 2 are shown. In the partial file management table 1702, “partial file identifiers” are C1 and C
2. "Location" of each partial file indicated by A1,
"Original location" is shown. In the partial file management table 1703, the “location” and “original location” of each partial file whose “partial file identifier” is indicated by A2, A3, and C1 are shown.

Here, the states of the partial file management tables 1701, 1702, and 1703 shown in FIG.
Partial file management tables 1601 and 160 shown in FIG.
The difference between the statuses 2 and 1603 is that the partial file A1 (12
6a) from server computer C 1107 to server computer B
1106 and the partial file C1 (12
6e) from the server computer B1106 to the server computer C1
This is due to the movement to 107.

That is, the partial file A1 (126a)
File management table 1702
Is added with the item of the partial file A1 (126a). In the partial file management table 1703, the item of the partial file A1 (126a) (FIG. 16)
See) has been removed. Further, the server computer B11
06 refers to the "original location" of the partial file A1 (126a) and informs the server computer A1105 of the movement. With this notification, the server computer A1105 indicated by the "original location" registers in the partial file management table 1701. Partial file A
1 (126a) to "file: // site"
eB / serverB / DiskID2 / ".

In response to the movement of the partial file C1 (126e), the “location” of the partial file C1 (126e) registered in the partial file management table 1702 is changed to “file: // siteB / serverB”.
/ DisiID3 / ”(FIG. 16 (B)) to“ fil
e: // siteC / serverC / DiskID3
/ "Has been changed. In the partial file management table 1703, the partial file C1 (126e)
Items have been added.

Next, in the state shown in FIG. 17, when the client computer 1 (108-1) refers to the distributed file C, the operation when the reference content is included in the partial file C1 (126e) will be described. I do.

(1) Client computer 1 (108-
1) The server computer A11 that created the distributed file C
05 is requested to refer to the distributed file C. In the server computer A 1105, the distributed file management table 1
Referring to 701, it is checked which of the partial files C1 and C2 constituting the distributed file C is being referred to. Here, the partial file C1 (126
e). The “location” of the partial file C1 (126e) is “file: // siteB / serverB”.
/ DiskID3 / ”, the server computer A1105
Sends the partial file C to the server computer B1106.
1 (126e) is confirmed.

(2) The server computer B 1106 checks the distributed file management table 1702 and finds the partial file C
The "location" of 1 (126e) is checked. Partial file C
1 (126e) is “file: /// si
teC / serverC / DiskID3 / "
The server computer B1106 checks with the server computer C1107 whether or not the partial file C1 (126e) exists.

(3) The server computer C 1107 checks the distributed file management table 1703 and finds the partial file C
The "location" of 1 (126e) is checked. Partial file C
1 (126e) is “file: /// si
teC / serverC / DiskID3 / "
The partial file C1 (126e) is stored in the server computer C11.
07.

(4) The server computer C1107 sends the partial file C1 (126e) to the server computer B1106.
Is "file: // siteC / serverC / D
The notification that the file exists in "iskID3 /".

(5) Upon receiving this notification, the server computer B
1106, the partial file C1 (126e) is stored in the server computer A1105 as "file: // siteC / se
rverC / DiskID3 / ".

(6) The server computer A 1105 requests the server computer C 1107 to refer to the partial file C 1 (126 e). Simultaneously with this request, the client computer 1 (108-1) requesting the reference is instructed to make a reference request for the partial file C1 (126e) directly to the server computer C1107. In the server computer A1105, the “location” of the partial file C1 (126e) is changed to “file: // siteC / server”.
C / DiskID3 / ".

As described above, in the present embodiment, the distributed file management units 1112, 1117, and 1122 perform operations based on the information in the access information table 1201, load information table 401, and external load information table 901 for each partial file. To determine the partial file to be moved. In addition, by providing the distributed file moving units 1131, 1132, and 1133 for moving a partial file to another server computer, by moving the partial file to another server computer, the load on the storage device of a specific server computer can be reduced. Concentration can be avoided.

In the third embodiment, the operation algorithm of the distributed file transfer unit 1131 (see FIG. 1)
In step 1501 of 5), instead of detecting that the “load” of each storage unit of the storage device has exceeded a predetermined value, the “remaining capacity” of each storage unit of the storage device has a predetermined value, for example, 10 It may be detected that the value falls below a value such as [Mbytes] (however, this value is determined according to the configuration of the device or system). This can avoid imbalance in the capacity of each storage unit of the storage device.

In step 1501 described above,
Instead of detecting that the “load” of each storage unit of the storage device has exceeded a predetermined value, the “use communication bandwidth” is changed from the “load information” of the network 101 to a predetermined value, for example, the available communication bandwidth. Of 80% (however, this value is
(Determined according to the configuration of the device or system). In step 1502, by selecting a partial file that increases the load on the network 101 and a computer site that increases the load on the network from the access information table 1201, concentration of the load on the network can be avoided. For example, computer site A1102
When the “used communication bandwidth” (FIG. 4) of the link between the (source site) and the computer site B 1103 (destination site) exceeds a predetermined value, it exists in the server computer A 1105 and causes an increase in network load. Move the partial file to the server computer B 1103. This allows
The “use communication bandwidth” between the computer site A 1102 and the computer site B 1103 can be reduced.

In the third embodiment, in step 1502, it is confirmed whether or not the partial file of the destination server computer can be moved. In step 1503, the partial file is moved. By moving the partial file without confirming whether the partial file can be moved in step 1502, the confirmation processing in step 1502 can be omitted. At this time, if the server computer at the destination of the partial file cannot accept the transfer of the partial file, the server computer at the destination searches for a destination for further moving the partial file, and this partial file is What is necessary is just to move.

In step 1503 described above,
The partial file has been moved from the source server computer to the destination server computer.
Another partial file that can be moved to the source server computer may be selected from the partial files in the destination server computer, and the other partial file may be moved to the source server computer. As a result, partial files can be prevented from being concentrated on one server computer, and the load on file access can be further reduced.

In step 1502 described above,
When selecting a server computer to which a partial file is to be moved, a server computer list is set in advance, and among the server computers in the list, the remaining capacity of each storage unit of the storage device is sufficient, and the load is a predetermined value. It is preferable to select a server computer having a lower storage device. As a result, the time spent for selecting the server computer to which the partial file is to be moved can be reduced.

(Embodiment 4) FIG. 18 is a block diagram showing a distributed file transfer unit 11 of the distributed file management system shown in FIG.
It is a flowchart which shows another operation algorithm of 31, 1132, 1133.

In FIG. 18, first, the distributed file movers 1131, 1132, and 1133 monitor the communication cost to each partial file at predetermined intervals (step 180).
1). Here, the communication cost can be, for example, a communication time between a client computer that is referencing the partial file and a server computer that holds the partial file. In FIG. 11, for example, the client computer 1 (108-1) and the partial file A2 (126)
The communication cost of c) is different from the client computer 1 (108-1) referring to the partial file A2 (126c),
The communication time is assumed to be the communication time with the server computer C1107 holding the partial file A2 (126c).

Here, the distributed file movers 1131, 1
132 and 1133 detect that the communication cost to the partial file exceeds a predetermined value, for example, 1 second (however, this value is determined according to the configuration of the device or system) (step 1801). ), A partial file whose communication cost exceeds a predetermined value is selected as a source partial file (step 1802). When a plurality of client computers are accessing a partial file whose communication cost exceeds a predetermined value, the communication cost for each access is obtained, and these are added to obtain the total communication cost.

When selecting the destination server computer,
Based on the external load information table 901, a server computer having a sufficient amount of “remaining capacity” of each storage unit of the storage device and having a storage device whose “load” is lower than a predetermined value is selected. Then, the above-mentioned total communication cost is transmitted to the selected server computer, and as a result of moving the partial file, how the communication cost changes is sequentially inquired, and the server computer having the minimum communication cost is selected. (Step 1802). Or, distributed file transfer unit 1
131, 1132, and 1133 may have connection information between the sites, estimate the communication cost from the connection information, and select a server computer having the minimum communication cost (step 1802).

FIG. 19 is a diagram showing an example of the connection information table. In FIG. 19, a connection information table 1901
Has items of “transmission source site” for transmitting the partial file, “destination site” for transmitting the partial file, and “communication time” indicating a communication cost from the transmission source site to the transmission destination site. From this connection information table 1901, obtain "communication time" between the server computer and the client computer according to which site the server computer holding the partial file belongs to and which site the client computer referring to the partial file belongs to. Can be. Based on the “communication time” and the total communication cost described above, the server computer with the “communication time” which is the minimum after the partial file is moved is selected as the destination of the partial file (step 1802). Here, when a plurality of client computers are accessing a partial file whose communication cost exceeds a predetermined value, a server computer with which the “communication time” after moving the partial file has the minimum communication cost is selected. It is good to do so.

For example, the client computer 1 (108-
When the communication cost (hereinafter, also referred to as "cost A2") between the first file and the partial file A2 (126c) exceeds a predetermined value, the partial file A2 (126c) is transferred to the server computer A1.
When moving to 105, the server computer A1105 is inquired about the communication cost, or the communication cost is obtained from the connection information table 1901. If the result is lower than the cost A2, the server computer A1105 is set as a candidate for the destination server computer.
This processing is also performed for other server computers, and a server computer with the minimum communication cost is searched for (step 180).
2).

Finally, the partial file is moved (step 1803) based on the information obtained by the selection of the source partial file and the selection of the destination server computer (step 1802).

As described above, in the fourth embodiment, the distributed file moving units 1131, 1132, and 1133 perform the operations of the load information table 401, the external load information table 901, the access information table 1201, and the connection information table. Based on the connection information between the server computer and the client computer obtained from 1901, the communication cost between the server computer in which the partial file being processed is present and the client computer requesting the processing is determined, and the communication cost is determined to be a predetermined value. When the value exceeds the value, the partial file is moved to another server computer in which the communication cost is reduced. Therefore, in addition to the effects obtained in the third embodiment described above,
The average value of the access time from the client computer to the partial file can be reduced.

In the above-described fourth embodiment, the communication time is taken as an example of the communication cost. However, the communication time may be “delay” or “fluctuation (fluctuation width)”.

In the fourth embodiment, when a plurality of client computers are accessing the same partial file, the partial file is moved so as to minimize the total communication cost. The partial file may be moved so as to minimize the communication cost.

(Embodiment 5) FIG. 20 is a block diagram showing an example of a fifth embodiment of the distributed file management system according to the present invention. 20, the same components as those in FIG. 8 are denoted by the same reference numerals. The distributed file management system shown in FIG. 20 includes a plurality of computer sites A2002 and B200 each including a server computer such as a personal computer and a workstation and a client computer group including a plurality of client computers such as a personal computer and a workstation.
3, computer site C2004, and computer site A2
002, computer site B2003, and computer site C
A network 101 such as a local area network or a wide area network that interconnects the networks 2004
And

Here, the computer site A2002 is connected to a plurality of server computers such as a personal computer and a workstation (in FIG. 20, "server computer A2
005 ") and client computers 1 to n (10
8-1 to 108-n) Client computer group A
108. This computer site A2002
Connects a plurality of server computers (only “server computer A2005” is shown in FIG. 20) and a group of client computers A108 to an internal network 13 such as Ethernet.
1 and is, for example, an Internet domain.

Further, similarly to the computer site A2002,
The computer site B2003 has a plurality of server computers (FIG. 2).
0 indicates only "server computer B2006")
And a client computer group B109 composed of a plurality of client computers.
It includes a plurality of server computers (only “server computer C2007” is shown in FIG. 20) and a client computer group C110 including a plurality of client computers. Further, similarly to the computer site A2002, the computer site B2003 and the computer site C2004 include a plurality of server computers (in FIG. 20, only "server computer B2006" and "server computer C2007") and a client computer group B109. And the client computer group C110 via an internal network 132 and an internal network 133, respectively.
For example, the Internet domain.

The server computer A 2005 includes a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 13 such as Ethernet.
Network interface 113 for connecting to 1
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
A state management unit 814 that monitors the load on the storage device 115, the remaining capacity of the storage device 115, and the load on the network interface 113, and retains information on these loads and capacity;
1, a state management unit 814, and a distributed file management unit 2012 connected to the network interface 113.

The state management unit 814 notifies the other server computers of the load information, and holds the external load information notified from the other server computers.
It has.

The distributed file management unit 2012 includes an access information table 1201 (FIG. 12), a load information table 401 (FIG. 4), and an external load information table 901.
A distributed file copy unit 2001 is provided for determining a partial file to be copied based on each piece of information for each partial file obtained from FIG. 9 and copying the partial file to another server computer.

The server computers B2006 and C2007 have the same configuration as the server computer A2005. That is, the server computer B 2006 is connected to the storage device 120 and the network interface 118.
, Partial file management unit 116, external state management unit 81
2 and a distributed file management unit 2017 including a distributed file copy unit 2032. In addition, the server computer C2007
Storage device 125 and network interface 123
, Partial file management unit 121, external state management unit 81
3 and a distributed file management unit 2022 provided with a distributed file copy unit 2033.

The difference between the distributed file management system shown in FIG. 20 and the distributed file management system shown in FIG. 8 is that the distributed file management unit 201 shown in FIG.
2, 2017 and 2022 correspond to the access information table 12
01, a distributed file copy unit 2031 that determines a partial file to be copied based on each piece of information for each partial file obtained from the load information table 401 and the external load information table 901, and copies the partial file to another server computer. 2032 and 2033.

As to the operation of the distributed file management system configured as described above, after the distributed file A, the distributed file B, and the distributed file C are created by the server computer A2005 as shown in FIG. Will be described in detail by taking as an example the case of performing.

FIG. 21 is a flowchart showing an operation algorithm of the distributed file copy unit 2031 of the server computer A2005. 21, first, the distributed file copy unit 2031 monitors the load information table 401 (FIG. 4) managed by the state management unit 814 at predetermined time intervals (step 2101).

When the distributed file copy unit 2031 detects that the “load” of an arbitrary storage unit of the storage device 115 has exceeded a predetermined value, for example, a value such as 80 [%] (step 2101), The partial file included in the storage unit of the storage device 115 is searched for by referring to the partial file management table 1301 (FIG. 13). Then, the access information of the found partial file is obtained from the access information table 1201. The “access count” of the obtained access information is compared, and the partial file having the largest “access count” is selected as the copy source partial file. Here, for example, the partial file A1
It is assumed that (126a) is selected. Next, based on the external load information table 901 (FIG. 9), a server computer having sufficient storage capacity in the storage unit of the storage device and having a storage device whose “load” is lower than a predetermined value is selected. Then, it is confirmed whether or not the partial file can be copied to the selected server computer, and the copyable server computer is selected as the copy destination server computer (step 210).
2).

Here, for example, the server computer C2007
Of the storage device 125 (storage device identifier: Disk)
It is assumed that ID2) is selected.

The partial file is copied based on the information obtained by selecting the copy source partial file and the copy destination server computer (step 210).
3), the monitoring processing in step 2101 is continued again.

Here, in the case of the above example, step 210
2, the partial file A1 as the copy source partial file
(126a) Since the server computer C2007 has been selected as the copy destination server computer, the distributed file copy unit 2031 of the server computer A2005 of the copy source transfers the partial file A1 (126a) from the storage device 115 via the partial file management unit 111. ) Is read, and this partial file A1 (126a) is
13 to the network 101. At the same time, information on the “original location” (FIG. 13) of the partial file A1 (126a) is also sent.

On the other hand, the copy destination server computer C2007
Then, the distributed file copy unit 2033 executes the partial file A1 sent from the copy source server computer A2005.
(126a) is received via the network interface 123. Then, the partial file management unit 121
Is written to a predetermined storage unit of the storage device 125 via
Also, the “original location” of the partial file A1 (126a) is received, and the partial file management table 1403 is received.
Register with. Thereafter, the copy destination server computer C200
Reference numeral 7 denotes the server computer A2005 of the copy source and the server computer indicated in the “original location” (in this example, the “original location” is also the server computer A2005)
Is notified that the copying of the partial file A1 (126a) is completed. The server computer of the copy source and the server computer indicated in the "original location" (both server computer A200
In 5), the information of the partial file A1 (126a) in the partial file management table 1401 is rewritten.

FIG. 22 is a diagram showing a partial file management table of the server computer after the copy processing. As a result of the copy of the partial file A1 (126a), the partial file management tables 1401, 1402, and 1 shown in FIG.
Reference numeral 403 denotes the partial file management table 22 shown in FIG.
01, 2202, and 2203. That is, FIG. 22A shows the partial file management table 2201 of the server computer A2005, and FIG. 22B shows the partial file management table 2202 of the server computer B2006.
(C) shows the partial file management table 2203 of the server computer C2007. In the partial file management table 2201, the "location" and "original location" of each partial file in which the "partial file identifier" is indicated by A1, A2, A3, B1, C1, and C2 are shown. In the partial file management table 2002, “partial file identifiers” are C1 and C2.
The "location" and "original location" of each partial file indicated by "." In the partial file management table 2203, “partial file identifier” is A1,
"Location" of each partial file indicated by A2 and A3
And "original location". The difference between the state of the partial file management table shown in FIG. 14 and the state of the partial file management table shown in FIG. 22 is that the partial file A1 (126a) is copied from the server computer A2005 to the server computer C2007. is there. That is, in the partial file management table 1401 of FIG. 14A, the partial file A1 (126a)
Is "file: // siteA / ser"
verA / DisiID1 / ”, whereas FIG.
In the partial file management table 2201 of FIG. 2A, the “location” of the partial file A1 (126a) is “file: // siteA / serverA / Dis”.
kID1 / "and" file: // siteC / ser
verC / Disk ID2 / (location of copy) ". Further, in the partial file management table 2103 of FIG. 22C, an item of the partial file A1 (126a) is added.

In addition, from the state of FIG.
The distributed file copy unit 2032 of 2006 copies the partial file C1 (126e) to the server computer C2007, and the distributed file copy unit 2033 of the server computer C2007 copies the partial file A1 (126a) to the server computer B2006. Can also.

FIG. 23 is a diagram showing a partial file management table in a state where a partial file is further copied from the state shown in FIG. FIG. 23A illustrates the server computer A20.
05 shows the partial file management table 2301 of FIG.
(B) shows the partial file management table 2302 of the server computer B2006, and (C) shows the server computer C2.
007 shows a partial file management table 2303. The partial file management table 2301 in FIG.
Has "partial file identifiers" of A1, A2, A3, B
"Location" of each partial file of C1, C1 and C2;
"Original location" is shown. In the partial file management table 2302 of FIG. 23B, the “location” and “original location” of each partial file whose “partial file identifier” is C1, C2, and A1 are shown.
Also, the partial file management table 230 shown in FIG.
3 shows the “location” and “original location” of each partial file whose “partial file identifier” is A1, A2, A3, and C1. The difference between the state of the partial file management table in FIG. 22 and the state of the partial file management table in FIG. 23 is that the partial file A1 (126a) is copied from the server computer C2007 to the server computer B2006, and the partial file C1 (126e). Is copied from the server computer B2006 to the server computer C2007.

That is, the partial file A1 (126a)
Corresponding to the copy of the partial file management table 230
2, the item of the partial file A1 (126a) is added. In addition, the server computer B 2006 refers to the “original location” of the partial file A1 (126a) and notifies the server computer A 2005 of the copy, so that the server computer A 2005 at the “original location”
However, in the partial file management table 2301, "fil" is added to the "location" of the partial file A1 (126a).
e: // siteB / serverB / DiskID2
/ ”Has been added. Also, the partial file C1 (126
The partial file management table 2 corresponding to the copy of e)
In 303, the item of the partial file C1 (126e) is added. Further, in the partial file management table 2302, the “location” of the partial file C1 (126e) is changed to “file: // siteB / serv”.
erB / DiskID3 / ”and“ file: /// s
itemC / serverC / DiskID3 / (location of copy) ".

In the state shown in FIG. 23, when the client computer 1 (108-1) refers to the distributed file C,
The operation when the reference content is included in the partial file C1 (126e) will be described.

(1) Client computer 1 (108-
1) The server computer A20 that created the distributed file C
05 is requested to refer to the distributed file C. In the server computer A2005, the distributed file management table 2
Referring to 301, it is checked which of the partial files C1 and C2 constituting the distributed file C is being referred to, and it is recognized that the reference is to the partial file C1 (126e). The “location” of the partial file C1 (126e) is “file: // siteB / serverB / Dis” in the distributed file management table 2301.
kID3 / ”, the server computer A2005 sends the partial file C1 (12
Confirm whether 6e) exists.

(2) The server computer B 2006 checks the distributed file management table 2302 and finds the partial file C
Check the location of 1 (126e). Partial file C1
The “location” of (126 e) is “file: // siteB / serv” in the distributed file management table 2302.
erB / DiskID3 / ”and“ file: // si
teC / serverC / DiskID3 / "
The server computer B2006 obtains the server computer B200 from the load information table 401 and the external load information table 901.
6 and the server computer C2007 with a low load. Here, when the server computer C2007 is selected, the server computer B2006 checks with the server computer C2007 whether or not the partial file C1 (126e) exists.

(3) The server computer C 2007 checks the distributed file management table 2303 and finds the partial file C
1 (126e) "location" is confirmed. The “location” of the partial file C1 (126e) is “file: // siteC / se” in the distributed file management table 2303.
rverC / DiskID3 / ", it is understood that the partial file C1 (126e) exists in the server computer C2007.

(4) The server computer C2007 sends the partial file C1 (126e) to the server computer B2006.
Is "file: // siteC / serverC / D
The notification that the file exists in "iskID3 /".

(5) The server computer B2006 sends the partial file C1 (126e) to the server computer A2005.
Is "file: // siteC / serverC / D
The notification that the file exists in "iskID3 /".

(6) The server computer A2005 requests the server computer C2007 to refer to the partial file C1 (126e). At the same time as this request, the client computer 1 (108-1) requests the client computer 1 (108-1) to refer to the partial file C1 (126e) directly to the server computer C2007. To instruct.

As described above, in the present embodiment, the distributed file managers 112, 117, and 122 store information for each partial file obtained from the access information table 1201, the load information table 401, and the external load information table 901. The distributed file copying units 2031, 2032, and 2033 determine the partial files to be copied based on the
By copying the partial file to another server computer, it is possible to avoid concentration of a load on a storage device of a specific server computer.

In the fifth embodiment, the operation algorithm of the distributed file copy unit 2031 (see FIG. 2)
In step 2101 of 1), instead of detecting that the “load” of each storage unit of the storage device has exceeded a predetermined value, the “used communication bandwidth” of the network load information table 403 has a predetermined value. The detected link may be detected. In step 2102, a partial file having a high network load and a computer site having a high network load may be selected from the access information table 1201. by this,
Concentration of network load can be avoided. For example, the computer site A2002 (transmission source site) and the computer site B2
"Used communication bandwidth" of link 003 (destination site)
Exceeds a predetermined value, a partial file existing in the server computer A2005 and causing an increase in network load is copied to the server computer B2003. Thereby, the computer site A 2002 and the computer site B 200
The "used communication bandwidth" between the three is reduced.

FIG. 24 is a flowchart showing another operation algorithm of the distributed file copy unit 2031 of the server computer A2005. In FIG. 24, first, the distributed file copy unit 2031 monitors the communication cost for the partial file at predetermined intervals (step 2401).

Here, the communication cost can be, for example, the communication time between a client computer that refers to a partial file and a server computer that holds the partial file. For example, the client computer 1 (10
8-1) and the communication cost of the partial file A2 (126c), the client computer 1 (108-1) referring to the partial file A2 (126c) and the server computer holding the partial file A2 (126c) C2007
The communication time may be set between the two.

Next, the distributed file copy unit 2031
When it is detected that the communication cost to the partial file exceeds a predetermined value (step 2401), the partial file whose communication cost exceeds the predetermined value is selected as a copy source partial file. When a plurality of client computers are accessing a partial file whose communication cost exceeds a predetermined value, the communication cost is obtained for each access, and these are added to obtain the total communication cost. on the other hand,
When selecting the server computer of the copy destination, based on the external load information table 901, select a server computer having a storage device with sufficient “remaining capacity” in the storage unit of the storage device and a low “load”, The total communication cost is transmitted to the selected server computer, and as a result of copying the partial file, an inquiry is sequentially made as to how the communication cost changes. Then, the server computer with the minimum communication cost is selected as the copy destination server computer. Alternatively, the distributed file copy units 2031, 2032, and 2033 have, for example, a connection information table 1901 shown in FIG. 19 as connection information between sites (between a server computer and a client computer), and determine communication cost (communication time)
And a server computer with the minimum communication cost may be selected. Connection information table 1 in FIG.
901 and information indicating which site the server computer belongs to and which site the client computer belongs to,
"Communication time" between server computer and client computer
Can be obtained. Then, the server computer with the shortest “communication time” may be selected as the copy destination of the partial file. If a plurality of client computers are accessing a partial file whose communication cost exceeds a predetermined value, the server computer having the minimum total communication cost is selected as the copy destination server computer (step 24).
02).

For example, the client computer 1 (108-
When the communication cost (cost A2) of 1) and the partial file A2 exceeds a predetermined value, the server computer A105 is inquired about the communication cost when the partial file A2 is copied to the server computer A105. Alternatively, the cost is obtained from the connection information 1801. If the result is lower than the cost A2, the server computer A1
05 is set as a copy destination candidate. Repeat this process,
Find a server computer that minimizes communication costs. Based on the information obtained by the selection of the copy source partial file and the selection of the copy destination server computer, the partial file is copied (step 2403), and step 240 is performed again.
The monitoring process of No. 1 is continued.

As described above, by changing steps 2101 and 2102 in FIG. 21 to steps 2401 and 2402 in FIG. 24, the average value of the access time from the client computer to the partial file can be reduced. In the above description, “communication time” is taken as an example of the communication cost. However, “delay” or “fluctuation (fluctuation width)” of the communication time may be used.

In addition, step 2102 in FIG.
In step 2402 of step 4, it is confirmed whether or not the partial file can be copied to the copy destination server computer. In step 2103 and step 2403, the partial file is copied. By copying the partial file without confirming whether the file can be copied, steps 2102 and 2402 can be performed.
Can be omitted. At this time, in step 2103 and step 2403, if the copy destination server computer cannot accept the copy of the partial file, the copy destination server computer further searches for a copy destination for copying the partial file. The partial file may be copied, or the partial file transmitted for the copy may be discarded, and the copy source server computer may be notified that the copy partial file has been discarded.

In the fifth embodiment described above, FIG.
In step 2103 of FIG. 1 and step 2403 of FIG. 24, the partial file is copied from the server computer of the copy source.
Although a copy is simply made to the copy destination server computer, in addition to the copy processing, a partial file that may be moved to the copy source server computer is selected from the partial files in the copy destination server computer, and the The partial file may be moved to the source server computer. As a result, partial files are not concentrated on one server computer, and concentration of a load on a storage device of a specific server computer can be avoided.

In step 2102 of FIG. 21, a server computer list is set in advance when selecting a server computer of a copy destination, and a storage device of the storage device is selected from the server computers in the server computer list. It is also possible to select a server computer having a storage device with a sufficient remaining capacity of the unit and a low load. As a result, the time for selecting a server can be reduced.

(Embodiment 6) FIG. 25 shows a server computer A2 in the distributed file management system shown in FIG.
15 is a flowchart illustrating another operation algorithm of the distributed file copy unit 2031 of FIG. This embodiment has the same configuration as the above-described fifth embodiment, and performs the same operation except for the operation of the distributed file copy unit. Hereinafter, the operation algorithm of the distributed file copy unit 2031 will be described.

In FIG. 25, first, the distributed file copy unit 2031 monitors the load information table 401 (FIG. 4) managed by the state management unit 814 at predetermined time intervals (step 2501).

When the distributed file copy unit 2031 detects that the “load” of an arbitrary storage unit of the storage device 115 has exceeded a predetermined value, for example, a value such as 80 [%] (step 2501), The partial file included in the storage device is stored in the partial file management table 1301.
(FIG. 13). Then, “access information per unit time” of the found partial file is obtained from the access information table 1201. "Access count" of the acquired "Access information per unit time"
Are compared, and the partial file having the largest “access count” is selected as the copy source partial file (step 2502). Here, it is assumed that the partial file A1 is selected, for example.

Next, based on the external load information table 901 (FIG. 9), a plurality of server computers having sufficient storage capacity of the storage unit of the storage device and having a storage device whose “load” is lower than a predetermined value. Select Then, it is confirmed whether or not the partial file can be copied to the selected server computer, and the copyable server computer is determined as the copy destination server computer (step 2502).

Here, for example, it is assumed that the storage device 120 having the storage device indicated by the storage device identifier DiskID1 of the server computer B 106 and the storage device 125 having the storage device indicated by the storage device identifier DiskID2 of the server computer C107 are selected. I do.

Next, the partial file is copied based on the information of the copy source partial file and the copy destination server computer obtained in step 2502 (step 2).
503), the monitoring processing in step 2501 is continued.

In the above example, in step 2502,
Partial file A1 (126
a) Since the server computers B2006 and C2007 have been selected as the copy destination computers, the distributed file copy unit 203 of the server computer A2005 of the copy source has been selected.
1 is the storage device 11 via the partial file management unit 111
5, the partial file A1 (126a) is read out, and the
01 is transmitted by multicast. At the same time, the "original location" of the partial file A1 (126a) (FIG. 1)
The information about 3) is also transmitted.

On the other hand, the copy destination server computer B 2006
And the computer C2007, the distributed file copy unit 203
2 and 2033 receive the partial file A1 (126a) transmitted from the server computer A2005 of the copy source via the network interfaces 118 and 123, and the storage device 1 via the partial file management units 116 and 121.
20 and 125 are written. Also, the partial file A1 (1
26a) “Original location” (FIG. 13) is also received, and the respective partial file management tables 1402, 1
403 (FIG. 14). Thereafter, the copy destination server computer B2006 and the server computer C2007 respectively become the copy source server computer and the "original location".
Since the server computer shown in FIG. 13 (that is, in this case, both are server computers A2005), the copying of the partial file A1 (126a) to both server computers A2005 has been completed. Notify. In the copy source server computer and the server computer indicated in the "original location" (FIG. 13), that is, in the server computer A2005, the partial file management table 1
The information of the partial file A1 (126a) of 401 (FIG. 14) is rewritten.

As described above, in the present embodiment, when the distributed file copying units 2031, 2032, and 2033 copy a partial file to another server computer, a plurality of copy destination server computer candidates are selected. Since the partial file is simultaneously copied to a plurality of selected server computers by multicast communication, it is possible to reduce the communication amount when copying the partial file.

(Embodiment 7) FIG. 26 is a block diagram showing an example of another embodiment of the distributed file management system according to the present invention. Here, in FIG. 26, the same components as those in FIG. 1 are denoted by the same reference numerals. In FIG. 26, this distributed file management system includes a plurality of computer sites A2602, B2603 provided with a server computer such as a personal computer and a workstation and a client computer group including a plurality of client computers such as a personal computer and a workstation. And a computer site C2604, and a network 101 such as a local area network or a wide area network interconnecting the computer site A2602, the computer site B2603, and the computer site C2604.

Here, the computer site A2602 is provided with a plurality of server computers such as a personal computer and a workstation (in FIG. 26, "server computer A2
605) and client computers 1 to n (10
8-1 to 108-n) Client computer group A
108. This computer site A2602
Connects a plurality of server computers (in FIG. 26, only "server computer A2605") and a client computer group A108 to an internal network 13 such as Ethernet.
1 and is, for example, an Internet domain.

Also, similar to the computer site A2602,
The computer site B 2603 has a plurality of server computers (FIG. 2).
6, only "server computer B2606" is shown.)
And a client computer group B2609 including a plurality of client computers, and a computer site C2604.
Has a plurality of server computers (in FIG. 26, only "server computer C2607" is shown) and a client computer group C2610 composed of a plurality of client computers. Further, the computer site B 2603 and the computer site C 2604 are connected to the computer site A 2602
26, a plurality of server computers (in FIG. 26,
“Server computer B2606” and “server computer C26”
07) is connected to the client computer group B109 and the client computer group C110 via the internal network 132 and the internal network 133, respectively, for example, in the Internet domain.

The server computer A 2605 includes a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 13 such as Ethernet.
Network interface 113 for connecting to 1
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
A state management unit 114 that monitors the load on the storage device 115, the remaining capacity of the storage device 115, and the load on the network interface 113, and retains information on these loads and capacity;
1, a state management unit 114, and a distributed file management unit 2612 connected to the network interface 113.

The distributed file management unit 2612 writes and reads a partial file to the partial file management unit 1.
Instruct 11. Also, the distributed file management unit 2612
When creating a distributed file, the state management unit 11
Based on the information obtained from step 4, the distributed file is divided into a plurality of partial files, and each partial file is arranged (recorded).
Server computer to be used is determined. When referencing or updating a previously created distributed file, the server computer in which a partial file of the corresponding distributed file exists (is recorded) is detected. Here, the distributed file management unit 2612 determines the partial file size for dividing the distributed file into the partial files according to the information from the client computer or the type of data recorded in the distributed file. A portion 2631 is provided.

The server computer B 2606 and the server computer C 2607 have the same configuration as the server computer A 2605. That is, the server computer B 2606 is connected to the storage device 120 and the network interface 118.
, Partial file management unit 116, and state management unit 119
And a distributed file management unit 2617. Further, the server computer C2607 has a storage device 12
5, a network interface 123, a partial file management unit 121, a state management unit 124, and a distributed file management unit 2622. The distributed file management units 2617 and 2622 include partial file size determination units 2632 and 2633, respectively.

Here, the difference between the distributed file management system shown in FIG. 26 and the distributed file management system shown in the figure is that the distributed file management system shown in FIG.
In the distributed file management units 2612, 2617, 2622,
Partial file size determination units 2631, 2632, and 263 that determine the size of the partial file when dividing the distributed file into partial files according to the information from the client computer or the type of data recorded in the distributed file.
3 is provided.

[0211] The partial file size determination units 2631 and 263 of the distributed file management system configured as described above.
The operation of 2,2633 will be described below.

In FIG. 3, for example, when the client computer 1 (108-1) issues a request to create a distributed file A to the server computer A 2605, the partial file The size determining unit 2631 determines the size when allocating the partial fill. Partial file size determination unit 2
When the 631 determines the size of the partial file,
The type of data recorded in the distributed file (for example, MJ
PEG, MPEG1, MPEG2, etc.) or an instruction from the client computer 1 (108-1).

As described above, in the distributed file management system of this embodiment, the distributed file management unit 261
2, 2617, 2622 partial file size determination unit 2
631, 2632, and 2633 configure the distributed file by determining the size of the partial file when the distributed file is divided into the partial files according to the information from the client computer and the type of data recorded in the distributed file. The size of the partial file can be changed as appropriate. As a result, it is possible to prevent logically and contentsally related data, for example, data for one frame of an image from being divided into a plurality of partial files.

(Embodiment 8) Next, a case will be described in which the distributed file management unit and the state management unit of the plurality of server computers are combined into one server computer and are centrally managed by the server computers.

FIG. 27 shows an example of the configuration of the distributed file management system according to the first embodiment of the present invention in which the distributed file management unit and the status management unit are combined into one server computer. Is shown. 27, the same components as those in FIG. 1 are denoted by the same reference numerals.

The distributed file management system shown in FIG. 27 includes a plurality of computer sites A27 each including a server computer such as a personal computer or a workstation and a client computer group including a plurality of client computers such as a personal computer or a workstation.
02, computer site B2703, and computer site C2
704 and a management server computer X2711 for centrally managing distributed files located on other server computers
Computer site X2710 provided with
702, computer site B2703, computer site C27
04 and a network 101 such as a local area network or a wide area network for interconnecting the computer sites X2710.

Here, the computer site A2702 is provided with a plurality of server computers such as a personal computer and a workstation (in FIG. 27, "server computer A2
705 ") and client computers 1 to n (10
8-1 to 108-n) Client computer group A
108. This computer site A2702
Connects a plurality of server computers (in FIG. 27, only "server computer A2705") and a group of client computers A108 to an internal network 13 such as Ethernet.
1 and is, for example, an Internet domain.

Also, similar to the computer site A2702,
The computer site B2703 has a plurality of server computers (FIG. 2).
7, only "server computer B2706" is shown.)
And a client computer group B109 composed of a plurality of client computers.
It includes a plurality of server computers (only the “server computer C2707” is shown in FIG. 27) and a client computer group C110 including a plurality of client computers. Further, these computer site B2703 and computer site C2704 are, like computer site A2702, provided with a plurality of server computers (in FIG. 27, only "server computer B2706" and "server computer C2707") and a client computer group B109. And the client computer group C110 via an internal network 132 and an internal network 133, respectively.
For example, the Internet domain.

The server computer A 2705 has a storage device 115 such as a hard disk for recording a partial file of the distributed file, and an internal network 13 such as Ethernet.
Network interface 113 for connecting to 1
And a partial file management unit 111 that controls writing and reading to and from the storage device 115 storing the partial file.
And is constituted by.

The server computer B 2706 and the server computer C 2707 have the same configuration as the server computer A 2705. That is, the server computer B 2706 is connected to the storage device 120 and the network interface 118
And a partial file management unit 116. Further, the server computer C2707 has the storage device 125
, A network interface 123, and a partial file management unit 121.

The management server computer X 2711 monitors the network interface 2713 for connection to the internal network 134 such as Ethernet, the load on the storage device of each server computer, the remaining capacity, and the load on the network interface, and holds information on the load. State management unit 2714 to perform writing and reading of partial files and the partial file management units 111 and 11 of each server computer.
6, 121, or when creating a distributed file, the distributed file is divided into a plurality of partial files based on information from the state management unit 2714, and a server computer on which each partial file is arranged is determined. When a partial file is created and a distributed file is referred to or updated, a distributed computer that refers to or updates the partial file by detecting the server computer where the partial file of the distributed file to be referenced or updated is located It is configured by a file management unit 2712.

In FIG. 27, each distributed file A,
This shows a state after B and C have been created. That is,
In the storage device 115 of the server computer A2705, a partial file A1 (126a) of the distributed file A and a partial file B1 (126b) of the distributed file B are recorded. In the storage device 120 of the server computer B 2706, a partial file C1 (126e) of the distributed file C and a partial file C2 (126f) of the distributed file C are recorded. The storage device 125 of the server computer C2707 stores the partial file A2 (1
26c) and the partial file A3 of the distributed file A (126
d) are recorded.

Next, the operation of the distributed file management system configured as described above will be described. In the following, a request to create a distributed file A is issued from the client computer 1 (108-1) of the client computer group A108 to the server computer A2705, and partial files A1 to A3 as shown in FIG. 27 are created. A description will be given of the distributed processing in the case as an example. Here, the storage devices 115, 120, and 125 shown in FIG. 27 each have a plurality of storage units or storage areas (hereinafter, also simply referred to as “storage units”). These plurality of storage units may be physically one recording medium, or may be a plurality of recording media.

In FIG. 27, first, a request to create a distributed file A is issued from the client computer 1 (108-1) to the management server computer X2711 of the computer site X2710. The request to create the distributed file A is received by the distributed file management unit 2712 via the internal network 131 and the network 101 of the computer site A 2702, the internal network 134 of the computer site X 2710, and the network interface 2713 of the management server computer A 2711.

FIG. 28 is a flowchart showing an operation algorithm of the distributed file management unit 2712 when a request for creating a distributed file is received. Hereinafter, the detailed operation of the distributed file management unit 2712 will be described with reference to FIGS. 27 and 28.

In FIG. 28, first, the distributed file management unit 2712 refers to the load information managed by the state management unit 2714 (step 2801).

The state management unit 2714 manages, for example, a load information table 401 as shown in FIG. Although FIG. 4 shows information regarding the server computer A 2705, the state management unit 2714 prepares and manages the load information table 401 as shown in FIG. 4 for each server computer. In FIG. 4, the load information table 401
Is composed of a storage device load information table 402 and a network load information table 403. The storage device load information table 402 includes a “storage device identifier” for identifying each storage unit of the storage device connected to the server computer, a “load” of each storage unit of the storage device, and each storage device of the storage device. It is composed of information on the “remaining capacity” of the section. The “load” of each storage unit of the storage device is indicated by what [%] of the maximum transfer rate of the storage unit is used. The network load information table 403 indicates to which site data to be transmitted via the network interface of each server computer is transmitted, how much bandwidth is used, and how much data is received. It indicates from which site the data is sent, and how much bandwidth is being used to receive it. Also,
The “source site” indicates the computer site of the data transmission source, the “destination site” indicates the computer site of the data transmission destination, and the “bandwidth used” indicates the computer site of the transmission source and the computer site of the transmission destination. It shows the communication bandwidth used between them.

For example, when creating a distributed file A,
The distributed file management unit 2712 is a server computer A270
The storage device identifier of the storage device 115 of No. 5 is DiskID1
Is “20%” for the storage unit indicated by
Thus, information that the “remaining capacity” is 10 [Mbytes] can be obtained.

Next, based on the “storage device load information” obtained from the status management unit 2714, the distributed file management unit 2712 selects “storage device information” from among the storage units of the storage device connected to each server computer. Storage where the value of “remaining capacity” is larger than a predetermined value and the value of “load” is lower than a predetermined value, for example, 80 [%] (this value is determined according to the configuration of the system and other devices) The storage device having the unit is selected. Then, the partial files are sequentially allocated to the storage unit of the storage device. At this time, the size of the partial file may be the same fixed length in all server computers. In this partial file allocation process, it is detected whether all the partial files have been allocated (step 28).
02).

Here, when creating the distributed file A,
In FIG. 27, the partial file A1 (126a) is allocated to the server computer A2705, and the partial file A2 (126a) is assigned.
(126c) and the partial file A3 (126d) are allocated to the server computer C2704.

When all the partial files have been allocated, the distributed file management unit 2712 stores the information for managing the distributed files in the distributed file management table 501 of FIG. 6 is registered in the partial file management table 601 (step 28).
03).

In the above example, as shown in FIG.
The distributed file A includes a partial file A1 (126a), a partial file A2 (126c), and a partial file A3.
(126d).

In FIG. 6, the partial file A1
The location of (126a) is "file: // siteA
/ ServerA / DiskID1 / (computer site A
Storage device identifier Dis of the server computer A105 of 102
kID1) ”and the location of the partial file A2 (126c) is“ file: // siteC / server ”.
C / DiskID2 / (storage device identifier DiskID2 of the server computer C107 of the computer site C104) "and the location of the partial file A3 (126d) is" f
ile: // siteC / serverC / DiskI
D2 / (storage device DiskID2 of server C107 of computer site C104) ".

Next, in order to create the partial file registered in step 2803 on the corresponding server computer, the distributed file management unit 2712 issues a partial file creation request to the partial file management unit of the server computer that creates the partial file. . At the same time as this creation request, data is transmitted from the client computer 1 (108-1) to the server computer that creates a partial file directly from the client computer 1 (108-1) that has issued the request for creating a distributed file. To instruct. In the server computer requested to create the partial file, the partial file management unit writes data from the client computer 1 (108-1) to each storage unit of the storage device. Distributed file management unit 27
Step 12 repeats this process until all partial files have been created (step 2804).

Here, partial file A of distributed file A
1 (126a), the server computer A2705
Of the client computer 1
The data from (108-1) is written to a predetermined storage unit of the storage device 115. In addition, the partial file management unit 121 of the server computer C2707 transmits the client computer 1 (1
08-1) is written into a predetermined storage unit of the storage device 125 to create a partial file A2 (126c) and a partial file A3 (126d).

On the other hand, if it is determined in step 2802 that all the partial files cannot be allocated, the distributed file management unit 2712 sends a request to the client computer 1 (108-1) that has issued the distributed file creation request. Then, it notifies that the creation processing of the distributed file has failed (step 2805).

Next, a case where a request for referencing or updating a distributed file (hereinafter also referred to as a “reference / update request”) is issued from a client computer to a management server computer will be described. In the following description, the client computer 1 in the client computer group A 108
An example in which a request for referencing / updating the distributed file A is issued from (108-1) to the management server computer X2711 will be described.

First, the client computer 1 (108-
Reference to distributed file A issued by 1) /
The update request is received by the distributed file management unit 2712 via the network interface 2713 in the management server computer X2711.

FIG. 29 is a flowchart showing an operation algorithm when the distributed file management unit receives a request for referencing / updating a distributed file. Hereinafter, the operation of the distributed file management unit 2712 will be described with reference to FIG.

First, the distributed file management unit 2712 responds to the request for referencing / updating the distributed file A from the client computer 1 (108-1), and the distributed file management table 501 (FIG. 5) and the partial file management table 601.
From FIG. 6, the partial file to be updated or referred to and the location of the partial file are obtained (step 290).
1).

Here, in the case of a request for referencing / updating the distributed file A, the distributed file A is read from the distributed file management table 501 (FIG. 5) as a partial file A1 (126).
a), a partial file A2 (126c) and a partial file A3 (126d).
Further, according to the partial file management table 601 (FIG. 6), the partial file A1 (126a) is set to “fi1e: //
/ SiteA / serverA / DiskID1 / ”exists in the storage unit of the storage device 115, and the partial file A2 (126c) is“ f11e: /// siteC /
serverC / DiskID2 / "exists in the storage unit of the storage device 125, and the partial file A3 (126
d) is “f1le: // siteC / serverC”
/ DiskID2 / ”in the storage unit of the storage device 125.

The distributed file management unit 2712 sends a request to the server computer that holds the partial file corresponding to the reference / update request from the client computer 1 (108-1).
A reference / update request for the partial file is made. Simultaneously with this request, the distributed file management unit 2712 refers to the client computer 1 (108-1) directly to the server computer where the partial file to be referenced or updated by the client computer 1 (108-1) exists. Instruct to make an update request. In response to a request from the distributed file management unit 2712, the partial file management unit of each server computer reads (references) a partial file existing in the storage device based on a reference / update request from the client computer 1 (108-1). ) Or write (update) a partial file to the storage device (step 2902).

Here, in the case of the distributed file A, the partial file A1 (126a) is transferred to the server computer A2705 of the computer site A2702 by the partial file A2 (126a).
c) is the server computer C27 of the computer site C2704
07, the partial file A3 (126d) exists in the server computer C2707 of the computer site C2704. Therefore, processing of the reference / update request for the partial file A1 (126a) is performed by the distributed file management unit 271.
2 in response to a request from client computer 1 (108
-1) and directly between the partial file management unit 111. On the other hand, the processing of the reference / update request for the partial file A2 (126c) and the partial file A3 (126d) is performed in response to a request from the distributed file management unit 2712.
It is performed directly between the client computer 1 (108-1) and the server computer C2707.

As described above, according to the distributed file management system of the present embodiment, in addition to the effects shown in the first embodiment, the management of the distributed file and the management of the system state are centralized. In addition, there is no need to have a plurality of overlapping management units, so that the system configuration can be simplified and the cost can be reduced.

In the distributed file management system shown in FIG. 27, the management unit of the distributed file management system shown in the first embodiment has been described as being configured to be concentrated on one management server computer. The present invention can be applied to the distributed file management systems according to the second to seventh embodiments.

In the distributed file management system of FIG. 27 described above, the management unit of the distributed file management system is described as being concentrated on one management server computer. A management server computer may be provided for each computer site. In this way, it is possible to prevent the load on the management server computer from being concentrated in a large-scale system.

In the above-described embodiment, when a partial file is present on a plurality of servers, the selection of the partial file is not limited to the selection in the order of small "load" of the server computer. The server computers may be used in order based on a predetermined rule, or a server computer having a “load” equal to or less than a predetermined threshold may be selected at random.

[0248]

As described above, according to the distributed file management apparatus and the distributed file management system of the present invention, a request is made in response to a request for creating, referencing, or updating a distributed file from a client computer to a server computer. In the management unit of each server computer or management server computer,
Since the arrangement of the partial files is determined based on the load information of each server computer, it is possible to avoid concentration of the load on a specific server computer.

Further, the load information is notified to another server computer, and the external load information notified from the other server computer is held, and the allocation of the partial file is determined based on the load information of the other server computer. For the determination, it is possible to avoid concentration of the load on a specific server computer.

Also, based on the access information, load information, and external load information for each partial file, a partial file to be moved is determined, and the partial file is moved to another server computer. It is now possible to avoid concentration of the load on the server and imbalance in the storage device capacity of each server computer.

Further, a partial file to be copied is determined based on the access status, load information, and external load information for each partial file, and the partial file is copied to another server computer. It is now possible to avoid the concentration of the load.

Also, the size of the partial file that is created by dividing the distributed file is determined according to the information from the client computer and the type of data recorded in the distributed file. This makes it possible to prevent data that can be changed and is related in terms of content and logic, for example, data for one frame of an image, from being divided into a plurality of partial files and recorded.

Also, by centralizing the management units for managing the partial files of the distributed file on one or a plurality of management server computers, resource duplication can be minimized, thereby suppressing an increase in cost. Now you can do it.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a distributed file management system according to the present invention.

FIG. 2 is a diagram showing a configuration of a distributed file according to the present invention.

FIG. 3 is a flowchart illustrating a distributed file creation algorithm of a distributed file management unit according to the present invention.

FIG. 4 is a diagram showing an example of a load information table according to the present invention.

FIG. 5 is a diagram showing an example of a distributed file management table according to the present invention.

FIG. 6 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 7 is a flowchart showing a distributed file reference / update algorithm of the distributed file management unit of the present invention.

FIG. 8 is a block diagram showing a distributed file management system of the present invention.

FIG. 9 is a diagram illustrating an example of an external load information table according to the present invention.

FIG. 10 is a flowchart illustrating a distributed file creation algorithm of a distributed file management unit according to the present invention.

FIG. 11 is a block diagram showing a distributed file management system according to the present invention.

FIG. 12 is a diagram showing an example of a partial file access information table according to the present invention.

FIG. 13 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 14 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 15 is a flowchart illustrating an operation algorithm of the distributed file moving unit according to the present invention.

FIG. 16 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 17 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 18 is a flowchart illustrating an operation algorithm of the distributed file moving unit according to the present invention.

FIG. 19 is a diagram showing an example of a connection information table according to the present invention.

FIG. 20 is a block diagram showing a distributed file management system according to the present invention.

FIG. 21 is a flowchart illustrating an operation algorithm of a distributed file copy unit according to the present invention.

FIG. 22 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 23 is a diagram showing an example of a partial file management table according to the present invention.

FIG. 24 is a flowchart illustrating an operation algorithm of a distributed file copy unit according to the present invention.

FIG. 25 is a flowchart illustrating an operation algorithm of a distributed file copy unit according to the present invention.

FIG. 26 is a block diagram showing a distributed file management system of the present invention.

FIG. 27 is a block diagram showing a distributed file management system according to the present invention.

FIG. 28 is a flowchart showing a distributed file creation algorithm of the distributed file management unit of the present invention.

FIG. 29 is a flowchart showing a distributed file reference / update algorithm of the distributed file management unit of the present invention.

FIG. 30 is a block diagram showing a conventional distributed file management device.

[Explanation of symbols]

101 network 102-104, 802-804, 1102-110
4, 2002-2004, 2602-2604, 270
2-2704, 2710 Computer site 105-107, 805-807, 1105-110
7, 2005-2007, 2605-2607, 270
5 to 2707 Server computers 108, 109, 110 Client computer groups 108-1 to 108-n Client computers 111, 116, 121 Partial file management units 112, 117, 122, 1112, 1117, 112
2, 2012, 2017, 2022, 2612, 261
7, 2622, 2712 Distributed file management unit 113, 118, 123, 2713 Network interface 114, 119, 124, 814, 819, 824, 2
714 Status management unit 115, 120, 125 Storage device 126a to 126f, 202-1 to 202-n Partial file 131 to 133 Internal network 401 Load information table 402 Storage device load information table 403 Network load information table 501 Distributed file management table 601 , 1301, 1401-1403, 1601-1
603, 1701-1703, 2201-2203, 2
301 to 2303 Partial file management tables 811, 812, 813 External status management unit 901 External load information table 1131, 1132, 1133 Distributed file transfer unit 1201 Access information table 1901 Connection information table 2031, 2032, 2033 Distributed file copy unit 2631, 2632 , 2633 Partial file size determination unit 2711 Management server computer

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Ryuji Yasukochi 1006 Kadoma, Kadoma, Osaka Prefecture Inside Matsushita Electric Industrial Co., Ltd. (72) Noriko Tanaka 1006 Odaka, Kadoma, Kadoma, Osaka Pref. Matsushita Electric Industrial Co., Ltd. Terms (reference) 5B045 BB49 DD16 GG02 GG09 JJ08 5B082 HA01 HA08

Claims (31)

[Claims] 1. A distributed file management device connected to a network in which a plurality of server computers having storage means for storing data and one or more client computers are connected, wherein a load on the plurality of server computers is provided. State management means for holding and managing information; and in response to a distributed file processing request from the client computer, a partial file of the distributed file is specified, and the load information managed by the state management means is specified. Distributed file management means for determining a server computer that processes the partial file based on the distributed file management device. 2. The external device according to claim 1, wherein said status management means notifies said distributed information to another distributed file management device, and stores load information of said server computer notified from said another distributed file management device as external load information. Comprising state management means,
2. The distributed file management device according to claim 1, wherein: 3. The distributed file management apparatus according to claim 2, wherein said external state management means notifies said another distributed file management apparatus of said load information by multicast. 4. The apparatus according to claim 2, wherein the external status management means notifies the load information to another distributed file management apparatus adjacent to the other distributed file management apparatuses. Distributed file management device. 5. A distributed file management system comprising: a plurality of server computers; one or more client computers; and a network connecting the plurality of server computers and the one or more client computers. Each of the computers includes storage means for storing partial files constituting part or all of the distributed file, state management means for holding and managing the load information, and responding to a distributed file processing request from the client computer. Distributed file management means for specifying a partial file of the distributed file and determining a server computer that processes the partial file based on the load information managed by the state management means. Characterized distributed file management system. 6. An external state for notifying the load information to another server computer and holding the load information of the other server computer notified from the other server computer as external load information. 6. The distributed file management system according to claim 5, further comprising management means. 7. The distributed file management system according to claim 6, wherein said external status management means notifies said another server computer of said load information by multicast. 8. The plurality of server computers are grouped into one or a plurality of server computer groups, and the external state management means is provided in a predetermined server computer group among the one or a plurality of server computer groups. The distributed file management system according to claim 6, wherein the load information is notified to another server computer to which the server belongs. 9. The plurality of server computers are grouped into one or a plurality of server computer groups, and the external state management means is provided for the one or a plurality of server computer groups in an adjacent server computer group. The distributed file management system according to claim 6 or 7, wherein the load information is notified to another server computer to which the server belongs. 10. The distributed file management means determines a partial file to be moved and another server computer to be moved based on the access information, the load information, and the external load information for each of the partial files. The distributed file management apparatus according to any one of claims 2 to 4, further comprising a distributed file moving unit configured to move a file to the server computer. 11. The distributed file moving unit detects that the load of the storage unit included in the load information is larger than a predetermined value, and moves based on the external load information and the access information. 11. The distributed file management apparatus or distributed file management system according to claim 10, wherein a partial file and another destination server computer are determined, and the partial file is moved to the other server computer. 12. The distributed file moving means detects that the remaining capacity of the storage means included in the load information is smaller than a predetermined value, and moves based on the external load information and the access information. 11. The distributed file management apparatus or distributed file management system according to claim 10, wherein a partial file and another destination server computer are determined, and the partial file is moved to the other server computer. 13. The distributed file moving means detects that the load of the network included in the load information is larger than a predetermined value, and moves based on the external load information and the access information. 11. The distributed file management apparatus or distributed file management system according to claim 10, wherein a file and another destination server computer are determined, and the partial file is moved to the other server computer. 14. The distributed file moving means holds the partial file based on the load information, the external load information, the access information, and connection information between the client computer and the plurality of server computers. The communication cost between the server computer having the storage means and the client computer which has issued the processing request is determined, another server computer having a communication cost smaller than the communication cost is determined, and the other server computer is determined. 11. The distributed file management device or distributed file management system according to claim 10, wherein the partial file is moved to a storage device. 15. The method according to claim 15, wherein the distributed file moving unit confirms in advance whether the partial file can be moved with respect to the other server computer to which the partial file is to be moved. Item 15. The distributed file management device or the distributed file management system according to any one of Items 10 to 14. 16. The distributed file moving means moves another partial file from the other server computer to the server computer when the partial file is moved to the another server computer. Claims 10 to 15
The distributed file management device or the distributed file management system according to the above. 17. The distributed file moving means makes a list of other server computers that can move the partial file, and determines another server computer to move the partial file based on the list. 17. The distributed file management device or distributed file management system according to claim 10, wherein: 18. The server according to claim 10, wherein said distributed file transfer means sends information on the server computer that created said partial file to said another server computer together with said partial file transfer. Distributed file management device or distributed file management system. 19. The distributed file management means determines a partial file to be copied and another server computer of a copy destination based on the access information for each of the partial files, the load information, and the external load information, 19. The distributed file management apparatus or distributed file management system according to claim 2, further comprising a distributed file copy unit that copies a partial file to said another server computer. 20. The distributed file copy unit detects that the load of the storage unit included in the load information is larger than a predetermined value, and copies based on the external load information and the access information. 20. The server according to claim 19, wherein the partial file and another copy destination server computer are determined, and the partial file is copied to the other server computer.
The distributed file management device or the distributed file management system according to the above. 21. A distributed file copying means for detecting that the load of the network included in the load information is larger than a predetermined value, and for copying based on the external load information and the access information. 2. A file and a destination other server computer are determined, and the partial file is copied to the other server computer.
10. The distributed file management device or distributed file management system according to item 9. 22. The distributed file copy unit holds the partial file based on the load information, the external load information, the access information, and connection information between the client computer and the plurality of server computers. The communication cost between the server computer having the storage means and the client computer which has issued the processing request is determined, another server computer having a communication cost smaller than the communication cost is determined, and the other server computer is determined. 20. The distributed file management device or the distributed file management system according to claim 19, wherein the partial file is copied to a storage device. 23. The distributed file copying means according to claim 15, wherein said another server computer, to which said partial file is copied, checks in advance whether said partial file can be copied. Item 23. The distributed file management device or the distributed file management system according to any one of Items 19 to 22. 24. The distributed file copying means, when copying the partial file to the other server computer,
24. The distributed file management apparatus or distributed file management system according to claim 19, wherein another partial file is copied from said another server computer to said server computer. 25. The distributed file copying means makes a list of other server computers that can copy the partial file, and determines another server computer to copy the partial file based on the list. 25. The distributed file management device or the distributed file management system according to claim 19, wherein: 26. The distributed file copying means selects a plurality of copy destination other server computers to copy the partial file, and simultaneously copies the partial file to the selected plurality of other server computers by multicast. 26. The distributed file management apparatus and distributed file management system according to claim 19, wherein: 27. The load information managed by the state management means includes a capacity and a load of the storage means, and a communication load between the network and the plurality of server computers. The distributed file management device or the distributed file management system according to claim 1. 28. The apparatus according to claim 1, wherein said server computer further comprises a partial file management means for writing said partial file into said storage means and reading said partial file from said storage means. The distributed file management device or the distributed file management system according to the above. 29. The distributed file management means, when the processing request from the client computer is a request for creating a distributed file, divides the distributed file into a plurality of partial files, and holds the divided partial files. A server computer is determined based on the load information managed by the state management unit. If the processing request from the client computer is a request to update or request a distributed file, the request to update or the request to update is issued. The existence of a partial file to be processed is determined, and a server computer that processes the processing request is determined based on the load information managed by the state management unit. The distributed file management device or the distributed file management system according to the above. 30. The distributed file management means includes a partial file size determination means for determining the size of the partial file constituting a part or all of the distributed file based on information from the client computer. 30. The distributed file management device or the distributed file management system according to claim 1, wherein: 31. A partial file size determining means for determining the size of the partial file constituting a part or all of the distributed file based on the type of data recorded in the distributed file. 30. The distributed file management apparatus or distributed file management system according to claim 1, further comprising: