JP2002373102A - File access control program, file access controller and file access control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase scalability and node side processing efficiency without operating any complicate setting with respect to various access patterns in the case of sharing a file among a plurality of clients. SOLUTION: This file access controller is provided with a plurality of nodes 6001 -600m whose files are respectively accessed according to access requests from clients 1001 -100n for outputting at least load distribution information for each access and a load distributing part 400 for transferring the access requests to the nodes fulfilling a prescribed condition among those nodes 6001 -600m based on the load distribution information each time the access requests are issued.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a file access control program, a file access control device, and a file access control method used for access control when sharing a file among a plurality of clients. The present invention relates to a file access control program, a file access control device, and a file access control method capable of improving scalability and efficiency on a node side without making complicated settings for various access patterns when files are shared by a user. is there.

In recent years, networking of computers has been rapidly progressing, and accordingly, a demand for a file server for collectively managing files shared between computers has been increasing. This is because a file system can be constructed at low cost. That is, in the file system, the use of the file server enables the file to be shared among a plurality of computers, so that the same file need not be copied and owned by a plurality of computers.

[0003]

2. Description of the Related Art A conventional file system comprises a plurality of clients and a file server connected to the clients via a network and for sharing files among the plurality of clients. Here, as the conventional file system, the following items (1) to (1) are used.
(3).

(1) A file system in which a plurality of file servers exist independently from a client. (2) A file system in which a plurality of file servers have the same directory structure as viewed from a client. (3) A file system having a load distribution device that performs load distribution for file access based on the IP address and port number of the client.

In a conventional file system, access requests to nodes to which a file is accessed are distributed in order to avoid a system down due to a concentration of accesses. In the file systems of (1) and (2), the system administrator individually sets which nodes and data areas are used in a distributed manner for each client. Further, such a file system requires a network design corresponding to the above settings.

In the file system of the item (3),
The load distribution can be automatically performed by the load distribution device. In this file system, since the information of the file is not recognized by the load distribution device, the exclusive access of the file access is executed on the node side.

[0007]

As described above, in the file systems of the items (1) and (2), the system administrator can individually determine which nodes and data areas are to be used in a distributed manner. Stated that the setting was made for each client. However, such a file system requires advanced skills and complicated settings as the number of clients increases. Therefore, there is a problem that it is difficult to exhibit high scalability with the skill of a general system administrator.

In the file system of the item (3),
Since the load balancer does not recognize the file information,
Since exclusive processing of file access must be executed on the node side, there has been a problem that processing efficiency regarding access on the node side is reduced.

[0009] The present invention has been made in view of the above,
Without complicated settings for various access patterns when sharing files between multiple clients,
An object of the present invention is to provide a file access control program, a file access control device, and a file access control method capable of improving scalability and processing efficiency on a node side.

[0010]

In order to achieve the above-mentioned object, the present invention provides a computer which accesses a file in response to an access request from a client,
A plurality of nodes that output at least load balancing information for each access, and a load balancing that transfers the access request to a node satisfying a predetermined condition among the plurality of nodes based on the load balancing information each time the access request is issued. means,
This is a file access control program for functioning as a program.

Further, the present invention provides a plurality of nodes each of which accesses a file in response to an access request from a client and outputs at least load distribution information for each access; Load distribution means for transferring the access request to a node satisfying a predetermined condition based on the plurality of nodes.

Further, the present invention provides a load balancing information output step in which a file is accessed in response to an access request from a client and at least load balancing information is output from a plurality of nodes for each access, and the access request is provided. A load balancing step of transferring the access request to a node satisfying a predetermined condition among the plurality of nodes based on the load balancing information.

According to this invention, the access request is transferred to a node satisfying a predetermined condition among a plurality of nodes based on the load distribution information output from the node each time an access request is received from the client. As a result, scalability and processing efficiency at the node side can be improved without complicated settings for various access patterns when files are shared at multiple nodes between multiple clients. .

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of a file access control program, a file access control device and a file access control method according to the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment according to the present invention. The file system shown in this figure includes clients 100 _{1 to} 100 _n , an external network 200, and a file server 300. The clients 100 _{1 to} 100 _n are n computer devices, and can access the file server 300 via the external network 200.

These clients 100 _{1 to} 100 _n
Share a file in the file server 300 and execute file reading from the file server 300, writing of a file to the file server 300, file search, and the like. FIG. 2 shows the client 100 ₁
From to 100 _n via the external network 200, the file sharing request information 7 that is sent to the file server 300
It is a figure showing an example of 00.

The file sharing request information 700 is information for requesting the file server 300 to share a file, and includes file handle information, file name information, and offset information. The file handle information is information including a byte string for uniquely specifying a file. The file name information is information including a character string relating to the name of the file. The offset information is information including a numerical value indicating an access range.

Returning to FIG. 1, the file server 300 is a server that performs a file sharing response (file transfer, etc.) in response to a file sharing request (read, write, file search, etc.) from the clients 100 _{1 to} 100 _n. is there.
The files existing in the file server 300 are shared by the clients 100 _{1 to} 100 _n as described above.

Further, the file server 300 is used for load balancing.
Unit 400, internal network 500 and node 600
₀ ~ 600_m It is composed of Load distribution unit 400
Are the external network 200 and the internal network 500
Between the client 100₁ -10
0_n Access (load) from the node 600 to be described later.
₀ ~ 600_m To any of

In the load distribution unit 400, the external network interface 401 is connected to the external network 200
Takes a communication interface between The control unit 402
Performs various controls related to load distribution. Details of the operation of the control unit 402 will be described later. In the storage unit 403,
The various score tables shown in FIG. 3 and the various statistical tables shown in FIGS. 4 and 5 are stored.

FIGS. 3 (a), 3 (b) and 3 (c)
8 shows the cache-oriented point table 800, the load factor-oriented point table 810, and the intermediate point table 8.
Reference numeral 20 denotes a table for determining a load distribution method based on a point system. Here, there are three types of load distribution methods in one embodiment: a cache-oriented type, a load factor-oriented type, and an intermediate type.

The cache emphasis type is a load distribution method that emphasizes that data (file) corresponding to a file sharing request exists in a cache memory described later. The cache-oriented score table 800 shown in FIG.
Is a table showing the correspondence between the "condition" of the scoring system and the score when the "condition" is "match" or "mismatch", and is used in a cache-oriented load distribution method.

The condition is that the node where the file exists is an access control node, that the data exists in the cache memory, that the load factor of the node is less than 80%, and that the load factor of the node is 100 %. The access management node is a node in which the correspondence between the client and the node is defined by the system administrator.

The load factor emphasis type score table 810 shown in FIG. 3B is a table showing the correspondence between the "condition" of the score system and the score when "match" or "mismatch" with "condition". And is used in a load distribution method of a load factor emphasis type. "Conditions" in the load factor emphasis type score table 810
Is the same as the “condition” in the cache-oriented score table 800. However, the load factor emphasis type score table 810
In some cases, as compared with the cache-oriented score table 800, there is a portion where the score differs when the "condition" is "match" or "mismatch".

The intermediate type score table 8 shown in FIG.
Reference numeral 20 denotes a table indicating the correspondence between the "condition" of the score system and the score when the "condition" is "matched" or "mismatched", and is positioned intermediately between the cache-oriented type and the load factor-oriented type. Is used in the load distribution method.

"Conditions" of the intermediate type score table 820
Are the same as the “conditions” in the cache-oriented point table 800 and the load factor-oriented point table 810. However, in the intermediate type score table 820, there is a portion where the score when the “condition” is “match” or “mismatch” is different from the cache-oriented score table 800 and the load factor-oriented score table 810.

The hash value table 9 shown in FIG.
Reference numeral 00 denotes a table used for determining the condition of being the access management node described above, and is a table indicating a correspondence between a node number corresponding to the access management node and a hash value. Here, the node numbers 0 to _m correspond to the nodes 600 ₁ to 600 _m (see FIG. 1). The hash value is the file sharing request information 700 shown in FIG.
This is a value obtained by applying the file handle information and the file name information to a hash function that is a one-way function, and is a value that uniquely identifies a node number corresponding to an access management node.

The cache statistics table 910 shown in FIG. 4B is a (cache) node representing a sharable file AZ existing in the file server 300 and a node having a cache memory in which the file is stored. It is a table showing the correspondence with a number.

In the example shown in FIG. 3, file A is stored in cache memory 608 of node 600 ₀ (see FIG. 1) corresponding to node number 0. The file B is stored in the cache memory 608 of the node 600 ₀ (see FIG. 1) corresponding to the node number 0 and the cache memory (not shown) of the node 600 ₂ corresponding to the node number 2
Is stored in

The load ratio statistics table 9 shown in FIG.
20, a node number 0~m corresponding to node 600 ₀ to 600 _m, a table establishing the relationship between the load factor of the node 600 ₀ to 600 _m. In the example shown in the figure, the load factor of the node 600 ₀ corresponding to the node number 0, 30
%.

The file access count statistics table 930 shown in FIG. 5A includes the sharable files AZ existing in the file server 300 and the client 100 _1.
It is a table showing the correspondence with the number of accesses from 100 _n . In the example shown in the figure, the access number of the file A is 1520.

The access type statistics table shown in FIG.
Bull 940 is the client 100 ₁ ~ 100_n From
Type of access to the file server 300 (read, write
And file and file searches) and the number of accesses
It is a table showing a person in charge. In the example shown in FIG.
The number of accesses for 1024 is 1024.

The request number statistical table 9 shown in FIG.
Reference numeral 50 denotes node numbers ₀ to _m corresponding to the nodes 600 ₀ to 600 _m (see FIG. 1), and clients 100 ₁ to 100 _m .
11 is a table showing a correspondence relationship from 0 _n to the number of requests for a file sharing request (file access request) to the node. In the example shown in the figure, the number of requests for the node 600 ₀ corresponding to the node number 0 is 1.

The node access count shown in FIG.
The total table 960 includes the node 600 ₀ ~ 600_m(Figure 1
Node number 0-m corresponding to client 1)
00 ₁ ~ 100_n Vs. the number of accesses to the node from
It is a table showing a response. In the example shown in FIG.
Node 600 corresponding to node number 0₀ Access
The number is 102. Here, the various statistical tables mentioned above
Every time the file is accessed,
(See FIG. 1).

Returning to FIG. 1, the internal network interface 404 takes a communication interface with the internal network 500. The nodes 600 ₀ to 600 _m correspond to the node numbers ₀ to _m , and the clients 100 ₁ to 100 ₁
00 _{n is} the access destination. These nodes 600 _0-
The 600 _m has a disk for storing files and a cache memory for temporarily storing files and speeding up access, and can be regarded as a virtual single file storage unit as a whole.

[0036] In the node 600 _0, the network processing unit 601 takes a communication interface with the internal network 500. When there is a file sharing request, the data integration unit 602 integrates various information and data shown in FIG. 6A as a file sharing response message,
This is transmitted to the load distribution unit 400 as response information 1000.

In the response information 1000 shown in FIG. 6A, the load distribution information 1001 includes a cache statistics table 910 (see FIG. 4B) and a load ratio statistics table 92.
0 (see FIG. 4C), file access number statistics table 930 (see FIG. 5A), access type statistics table 940 (see FIG. 5B), request number statistics table 9
50 (see FIG. 5 (c)) and load distribution information (statistical information) for updating the node access count statistics table 960 (see FIG. 5 (d)).

The header information 1002 is information relating to the header (destination, data type, etc.) of the data 1004. The file ID information 1003 is information for identifying a file to be accessed. Data 1004 is data of a file to be accessed.

Returning to FIG. 1, the file sharing processing unit 603
Executes a file sharing process according to a file sharing protocol. Load monitoring unit 604 monitors the load factor of the node 600 _0. The monitoring result of the load monitoring unit 604 is a load ratio statistical table 920 (see FIG. 4C).
Is reflected in

The file processing unit 605 executes processing related to reading, writing, searching, and the like of a file. In the file processing unit 605, the access management unit 606
Manage file access. The disk 607 stores files.

The cache memory 608 has a disk 6
Among the files stored in 07, the file which has been read is stored. Here, the cache memory 608
The reason that the file is stored in the cache memory 608 is that the access speed is faster than when the file is read from the disk 607 as compared with the case where the file is read from the disk 607.

The cache control unit 609 determines whether or not a file exists in the cache memory 608 (cache hit determination) and controls read / write of a file to / from the cache memory 608. File input / output unit 6
10 is a disk 607 or a cache memory 608
Input and output files related to The node 600
₁ to 600 _m are also the same configuration as node 600 ₀ as described above.

Next, the operation of the embodiment will be described with reference to the flowcharts shown in FIGS. FIG. 7 is a flowchart illustrating a load distribution method selection process according to one embodiment. This load distribution method selection processing is executed at predetermined time intervals by the control unit 402 (see FIG. 1), and the cache-oriented type (see FIG. 3A) and the load factor-oriented type (see FIG.
This is a process for selecting any one of (b) and the intermediate type (see FIG. 3 (c)).

In other words, the load distribution method selection processing is as follows.
One of the cache-oriented point table 800 (see FIG. 3A), the load factor-oriented point table 810 (see FIG. 3B), and the intermediate point table 820 (see FIG. 3C). This is a process for selecting one.

FIG. 8 is a flowchart for explaining the operation of the load distribution unit 400 shown in FIG. FIG. 9 is a flowchart illustrating the file sharing request process shown in FIG. FIG. 10 is a flowchart illustrating the file sharing response process shown in FIG.

At step SA1 shown in FIG. 7, the control unit 402 of the load distribution unit 400 (see FIG. 1)
Is referred to the file access number statistics table 930 shown in FIG. In step SA2, the control unit 402 determines whether access is concentrated on a small number (for example, three) of the files based on the file access number statistical table 930.

If the decision result in the step SA2 is "Yes", in a step SA3, the control section 402 sets the load distribution method to the load factor emphasizing type, and the load factor emphasizing type point table 810 shown in FIG. Select Step S
In A12, the control unit 402 waits for a predetermined time and then repeats Step SA1 and subsequent steps. Thereafter, the load distribution unit 400
Here, the load distribution processing is executed in accordance with the load factor emphasis type score table 810 until another load distribution method is selected.

On the other hand, if the result of the determination in step SA2 is "N
o ”, the control unit 402 proceeds to step SA4.
Is a file access count statistics table 930 (FIG. 5).
(See (a)), it is determined whether or not access is distributed to a large number (for example, 10) of files.

If the decision result in the step SA4 is "Yes", in a step SA5, the control unit 402 sets the load distribution method to the cache-oriented type and selects the cache-oriented type score table 800 shown in FIG. I do. In step SA12, after waiting for a predetermined time, the control unit 402 repeats step SA1 and subsequent steps. Thereafter, the load distribution unit 400 executes the load distribution processing according to the cache-oriented type score table 800 until another load distribution method is selected.

On the other hand, if the result of the determination at step SA4 is "N
o ”, the control unit 402 proceeds to step SA6.
Is the access type statistics table 94 shown in FIG.
0. In step SA7, the control unit 402
Based on the access type statistics table 940, it is determined whether or not the number of accesses related to writing is greater than the number of accesses for other read and file searches.

If the decision result in the step SA7 is "Yes", in a step SA8, the control unit 402 selects the intermediate type score table 820 shown in FIG. In step SA12,
After waiting for a predetermined time, the control unit 402 proceeds to step SA1.
Repeat thereafter. Thereafter, in the load distribution unit 400, until another load distribution method is selected, the intermediate type point table 82
0, the load distribution processing is executed.

On the other hand, if the result of determination at step SA7 is "N
o ”, the control unit 402 proceeds to step SA9.
Determines, based on the access type statistics table 940, whether or not the number of accesses related to read is greater than the number of accesses for other write and file searches.

If the result of the determination in step SA9 is "Yes", in step SA10, the control unit 402 selects the cache-oriented type score table 800 shown in FIG. I do.
In step SA12, after waiting for a predetermined time, the control unit 402 repeats step SA1 and subsequent steps. Thereafter, in the load distribution unit 400, until another load distribution method is selected,
The load distribution processing is executed according to the cache-oriented point table 800.

On the other hand, if the result of determination at step SA9 is "N
o ”, that is, when the number of accesses related to the file search in the access type statistics table 940 is larger than the number of accesses of other reads and writes, in step SA11, the control unit 402 sets the load distribution method to the load factor emphasis type. As a result, the load factor emphasis type score table 810 shown in FIG. 3B is selected.

At step SA12, the control section 402
After waiting for a predetermined time, step SA1 and subsequent steps are repeated.
Thereafter, the load distribution unit 400 executes the load distribution processing in accordance with the load factor emphasis type score table 810 until another load distribution method is selected.

As described above, in one embodiment, the load distribution method selection processing is executed every predetermined time, so that the load distribution method can be appropriately changed according to the number of accesses to the file and the number of accesses for each file type. Selected.

The control unit 402 executes processing based on the flowchart shown in FIG. 8 in parallel with the load distribution method selection processing described above. That is, in step SB1 shown in FIG.
_It is determined whether there is a file sharing request (file access request) from any one of the clients _{1 to} 100 _n . If the determination result of step SB1 is "Yes", in step SB2, the control unit 402 executes a file sharing request process described later.

In this case, the control unit 402 determines in step SB
The determination result of No. 1 is “No”. In step SB3,
Control unit 402, from one of the nodes of the nodes 600 ₀ to 600 _m, it is determined whether there is a file sharing response corresponding to the file sharing request described above, in this case,
The determination result is “Yes”. In step SB4, the control unit 402 executes a file sharing response process described later.

In this case, the control unit 402 determines in step SB
No. 3 is determined as “No”, and one of Step SB1 and Step SB3 is determined as “Yes”.
The determinations in step SB1 and step SB3 are repeated until.

Then, the client 100 shown in FIG.
_{When a} file sharing request is issued from ₁ to the file server 300, the control unit 402 sets the determination result of step SB1 to "Yes". In this case, the client 10
0 ₁ From as information of a file sharing request, the file sharing request information 700 is an external network 2 shown in FIG. 2
00 to the load distribution unit 400.

The file sharing request information 700 includes header information (not shown). Here, at this time, the load balancing method selection processing described above
It is assumed that the cache-oriented type score table 800 shown in FIG.

At step SB2, the control section 402 executes a file sharing request process. Specifically, in step SC1 shown in FIG. 9, control unit 402 executes a header process for confirming the header information of file sharing request information 700. In step SC2, the control section 402 extracts the header information.

In step SC3, the control unit 402 initializes (= 0) the total score corresponding to each of the nodes 600 ₀ to 600 _m . The total score is used when determining the transfer destination node of the file sharing request, that is, the load distribution destination node. In step SC4, the control unit 402 performs a hash calculation for obtaining a hash value from the file handle information and the file name information of the file sharing request information 700 shown in FIG. In this case, the hash value is 5.

In step SC5, the control unit 402 refers to the hash value table 900 shown in FIG.
An inquiry is made as to whether there is a hash value identical to the hash value (5 in this case) obtained in step SC4. In this case, the hash value of the node number 0 in the hash value table 900 (=
5) applies. That is, the node 600 ₀ corresponding to the node number 0 is the access control node.

In step SC6, the control unit 402 sets the cache-oriented type score table 800 shown in FIG.
And the total score of each node is updated based on the reference result of step SC5. In this case, with respect to node 600 ₀ corresponding to the node number 0, to match the condition "is an access management node", the total number +10
Points. On the other hand, each of the other nodes 600 ₂ to 600 _m has a score of −10 because it does not match the above condition.

In step SC7, the control unit 402 refers to the load ratio statistical table 920 shown in FIG. 4C, and sets the load ratio of each node to the cache-oriented type score table 800 (see FIG. 3A). (“The load factor is less than 80%”, “the load factor is 100”) are determined.

[0067] In this case, with respect to the node 600 ₀ corresponding to the node number 0 shown in FIG. 4 (c), the load factor 30
% And matches the condition that the load factor is less than 80%, so that the score is +5 points. Further, since the condition does not match the condition that the load factor is 100%, the score is set to 0. Hereinafter, similarly, node numbers 1 to m
Are similarly specified for each of the nodes 600 ₁ to 600 _m corresponding to.

At step SC8, the control unit 402 updates the total score of each of the nodes 600 ₀ to 600 _m corresponding to the node numbers ₀ to _m . That is, the control unit 40
In step 2, the score specified in step SC7 is added to the total score. Respect nodes 600 ₀ corresponding to the node number 0, a total score is +15 points (+ 10 + 5 + 0).

In step SC9, the control unit 402 refers to the cache statistics table 910 shown in FIG.
It is determined whether or not the condition (refer to FIG. 3A) (“data exists in the cache”) is met.

[0070] In this case, and FIG. 4 (b) to show the files A, B and Z (in this case, the file A is assumed to be the access request) node 600 ₀ of the cache memory 608 corresponding to the node number 0 , And matches the condition that “the data exists in the cache”, the score is set to +10 points. Hereinafter, similarly, the scores are similarly specified for each of the nodes 600 ₁ to 600 _m corresponding to the node numbers ₁ to _m .

At step SC10, the control section 402
Nodes 600 ₀ to 600 _m corresponding to node numbers ₀ to _m
Update the total score of each. That is, the control unit 4
02 adds the score specified in step SC9 to the total score. Respect nodes 600 ₀ corresponding to the node number 0, a total score is +25 points (+ 15 + 10).

At step SC11, the control section 402
Nodes 600 ₀ to 600 _m corresponding to node numbers ₀ to _m
Of the total scores for each of the nodes, the node with the highest total score is selected. In this case, it is assumed that the node 600 ₀ is selected. In step SC12, the control unit 402 sets the node 600 selected in step SC11.
₀ , the file sharing request information 700 (see FIG. 2) is transferred. That is, in this case, is the load distribution processing to a node 600 ₀ Load balancing method of the cache-oriented is executed.

At step SC13, the control section 402
The request number statistics table 950 shown in FIG. 5C is updated. In this case, the control unit 402, for transferring the file sharing request information 700 to the node 600 _0, after incrementing the number of request corresponding to the node number 0 number of requests statistics table 950, and returns to the main routine shown in FIG. 8 .

Then, the file sharing request information 700 is
It is received by the node 600 ₀ shown in FIG. Accordingly, the file sharing processing unit 603 converts the access type information (for example, read) from the header information (not shown) of the file sharing request information 700 into the file sharing request information 700.
, The information of the access target file (in this case, file A) is obtained.

Next, the cache control unit 609 reads the file from the cache memory 608 based on the above information. If the file does not exist in the cache memory 608, the file is read from the disk 607. Next, the file input / output unit 6
10 outputs the read file to the data integration unit 602.

As a result, the data integration unit 602
The response information 1000 shown in FIG. In the response information 1000, the load distribution information 1001 includes load ratio information from the load monitoring unit 604, cache node number, access type (read) information, node number information, and the like.

The header information 1002 includes a destination (in this case,
Information of the client 100 ₁ ). File ID
Information 1003 is information for identifying the read file. Data 1004 is the file. Next, the data integration unit 602 outputs the response information 1000
Is transmitted to the load distribution unit 400.

Then, the response information 1000 is stored in the load distribution unit 4
When received at 00, the control unit 402 sets the determination result of step SB3 shown in FIG. 8 to “Yes”. In step SB4, the control unit 402 executes a file sharing response process.

More specifically, step SD shown in FIG.
In FIG. 1, the control unit 402 sets the response information 1000 (FIG.
(See (a)) header processing for confirming the load balancing information 1001 and the header information 1002. Step S
In D2, the control unit 402 extracts the load balancing information 1001, the header information 1002, and the file ID information 1003.

At step SD 3, the control unit 402 checks the load distribution information 1001 (cache node number, load factor information, access type (read) information, node number information, etc.), header information 1002 and file ID information 1
Based on 003, a cache statistics table 910 (see FIG. 4B) and a load ratio statistics table 920 (FIG. 4B).
(See (c)), file access number statistics table 93
0 (see FIG. 5A), access type statistics table 94
0 (see FIG. 5B), the request number statistics table 950 (see FIG. 5C) and the node access number statistics table 960 (see FIG. 5D), respectively.

In the request number statistical table 950, the request number corresponding to the node number 0 is decremented by one. In tables other than the request number statistics table 950, the corresponding numerical value is incremented by one.

At step SD4, the control unit 402
From the response information 1000 shown in FIG.
001 is removed, and the response information 1100 shown in FIG.
Generate In step SD5, the control unit 402
Based on the destination information obtained from the
Client 100 via external network 200 ₁ 
To the response information 1100.

Also, in the file sharing request for the file search, the client sends the file ID information 1300 shown in FIG. 6C to the file server 300. The file ID information 1300 includes parent directory / file ID information 1200 and file name information 1201.

Thereafter, each time there is a file sharing request (access request) from the client, the above-mentioned various statistical tables (such as the cache statistical table 910) are updated, and the load distribution method (cache-oriented type score) is updated according to the degree of update. Table 800, load factor emphasis type score table 810,
The intermediate type score table 820) changes dynamically.

As described above, according to one embodiment,
For example, client 100₁ ~ 100 _n Any one of
Each time there is an access request from one client,
The load balancing information 1001 (FIG. 6
(See (a))), a plurality of nodes 600₀ ~ 60
0_m Out of the nodes that meet the predetermined conditions
Requests to achieve load balancing.
Client 100₁~ 100_n Between multiple nodes
Access patterns when sharing files
Scalability without complicated settings.
And the processing efficiency on the node side can be improved.

According to one embodiment, the node 60
Each time the load balancing information 1001 (see FIG. 6A) is input from any one of the nodes from ₀ to 600 _m ,
Since various statistics tables (such as the cache statistics table 910: see FIG. 4B) for storing the load balancing information are updated, according to the status of the node in real time,
Load balancing can be performed.

Further, according to one embodiment, as described with reference to FIG. 7, at predetermined time intervals, a plurality of load distribution methods are selected according to the load distribution information stored in the various statistical tables. Since a predetermined load distribution method is selected and the access request is transferred to the node based on the selected load distribution method, it is possible to flexibly cope with various access patterns. .

According to one embodiment, FIG.
6B, the load balancing information 1001 is deleted from the response information 1000, and the response information 1100 shown in FIG. 6B is transferred to the client.
00 without requiring any special settings for ₁ to 100 _n, it is possible to enhance the processing efficiency of the scalability and node side.

Further, according to the embodiment, since the load distribution information 1001 includes at least the information on the load ratio and the number of file accesses in the node, the load distribution information
Load distribution in consideration of the number of accesses can be performed efficiently.

Further, according to the embodiment, since the load distribution information 1001 includes at least the information of the file stored in the cache memory in each node, the load distribution taking the access speed into consideration is effective. Can be done

According to one embodiment, FIG.
As described with reference to the above, the points are totaled for each node depending on whether or not a predetermined condition corresponding to the selected load distribution method is met, and based on the total points, the total points among a plurality of nodes are calculated. Since the access request is transferred to the highest node, the scalability can be further improved by setting the predetermined condition in detail.

Although the embodiment of the present invention has been described in detail with reference to the drawings, a specific configuration example is not limited to this embodiment and a range that does not depart from the gist of the present invention. Even if there is a change in the design, the present invention is included in the present invention.

For example, in the above-described embodiment, the file access control program for realizing the function of the file server 300 shown in FIG. 1 is recorded on the computer-readable recording medium 1500 shown in FIG. The file access control program recorded on the recording medium 1500 is stored in the computer 14 shown in FIG.
A series of processes related to file access control may be executed by causing the file to be read and executed.

The computer 1400 includes a CPU 1410 for executing the file access control program, an input device 1420 such as a keyboard and a mouse, a ROM (Read Only Memory) 1430 for storing various data, and a RAM (RAM) for storing operation parameters and the like. Random Access Memo
ry) 1440, a reading device 1450 that reads a file access control program from the recording medium 1500, an output device 1460 such as a display or a printer, and a bus 1470 connecting each unit.

The CPU 1410 reads the file access control program recorded on the recording medium 1500 via the reading device 1450, and executes the file access control program to execute a series of processes related to the file access control described above. Execute Note that the recording medium 1500 includes not only a portable recording medium such as an optical disk, a floppy (registered trademark) disk, and a hard disk, but also a transmission medium such as a network for temporarily recording and holding data. It is.

(Supplementary Note 1) A plurality of nodes each of which accesses a file in response to an access request from a client and outputs at least load distribution information for each access, A file access control program for functioning as a load distribution unit for transferring the access request to a node satisfying a predetermined condition among the plurality of nodes based on the plurality of nodes. (Supplementary note 2) The file access control program according to Supplementary note 1, wherein the computer is caused to function as updating means for updating a table storing the load balancing information every time the load balancing information is input. (Supplementary Note 3) The computer causes the computer to function as a selection unit that selects a predetermined load distribution method from a plurality of load distribution methods at predetermined time intervals according to the load distribution information stored in the table. The file access control program according to claim 2, wherein the means transfers the access request to the node based on the load distribution method selected by the selection means. (Supplementary Note 4) Each of the plurality of nodes outputs response information obtained by adding the load distribution information to file information regarding the file to the load distribution unit, and the load distribution unit outputs the load distribution information from the response information. 3. The file access control program according to any one of supplementary notes 1 to 3, wherein the file information is transferred to the client. (Supplementary note 5) The file access control program according to any one of Supplementary notes 1 to 4, wherein the load balancing information includes at least information of a load factor and a number of file accesses in the node. (Supplementary note 6) The file access according to any one of Supplementary notes 1 to 5, wherein the load balancing information includes at least information of a file stored in a cache memory in each node. Control program. (Supplementary Note 7) The load distribution unit sums up the points for each node based on whether a predetermined condition corresponding to the load distribution method selected by the selection unit is met, and based on the total points, a plurality of nodes are provided. Wherein the access request is transferred to a node having the highest total score among the nodes.
5. The file access control program according to any one of 5. (Supplementary Note 8) A plurality of nodes each of which accesses a file in response to an access request from a client and outputs at least load balancing information for each access, and the plurality of nodes each time the access request is made based on the load balancing information. A load distribution means for transferring the access request to a node satisfying a predetermined condition among the nodes. (Supplementary Note 9) The apparatus further includes a selection unit that selects a predetermined load distribution method from a plurality of load distribution methods according to the load distribution information stored in the table storing the load distribution information at predetermined time intervals. 9. The file access control device according to claim 8, wherein the distribution unit transfers the access request to the node based on the load distribution method selected by the selection unit. (Supplementary Note 10) A load balancing information output step in which a file is accessed in response to an access request from a client, and at least load balancing information is output from each of a plurality of nodes for each access, A load distribution step of transferring the access request to a node satisfying a predetermined condition among the plurality of nodes based on information. (Supplementary Note 11) The method further includes a selecting step of selecting a predetermined load distribution method from a plurality of load distribution methods according to load distribution information stored in a table storing the load distribution information at predetermined time intervals. The file access control method according to claim 10, wherein in the distribution step, the access request is transferred to the node based on the load distribution method selected in the selection step.

[0097]

As described above, according to the present invention,
Since the access request is transferred to a node satisfying a predetermined condition among a plurality of nodes based on the load distribution information output from the node each time an access request is issued from the client, the load distribution is realized. The scalability and the processing efficiency on the node side can be improved without making complicated settings for various access patterns when files are shared by a plurality of nodes between clients.

Further, according to the present invention, the table for storing the load distribution information is updated every time the load distribution information is input from the node, so that the load distribution can be performed in accordance with the status of the node in real time. This has the effect that it can be performed.

Further, according to the present invention, a predetermined load distribution method is selected from a plurality of load distribution methods in accordance with the load distribution information stored in the table at predetermined time intervals, and the selected load distribution method is selected. , The access request is transferred to the relevant node, so that it is possible to flexibly cope with various access patterns and load distribution.

Further, according to the present invention, the load distribution information is deleted from the response information obtained by adding the load distribution information to the file information, and the file information is transferred to the client. There is an effect that scalability and processing efficiency on the node side can be improved without the need for setting.

Further, according to the present invention, the load balancing information includes
Since at least information on the load ratio and the number of file accesses in the node is included, it is possible to achieve an effect that the load can be efficiently distributed in consideration of the load ratio and the number of accesses.

According to the present invention, the load balancing information includes:
Since at least the information of the file stored in the cache memory in each node is included, it is possible to effectively perform the load distribution in consideration of the access speed.

Further, according to the present invention, the points are totaled for each node depending on whether a predetermined condition corresponding to the load distribution method selected by the selecting means is met, and a plurality of nodes are calculated based on the total points. Since the access request is transferred to the node having the highest total score, the scalability can be further improved by setting the predetermined conditions finely.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment according to the present invention.

FIG. 2 is a diagram showing an example of file sharing request information 700 in the same embodiment.

FIG. 3 is a diagram showing various score tables in the same embodiment.

FIG. 4 is a diagram showing various statistical tables in the same embodiment.

FIG. 5 is a diagram showing various statistical tables in the same embodiment.

FIG. 6 is a diagram showing various information in the same embodiment.

FIG. 7 is a flowchart illustrating a load distribution method selection process according to the same embodiment.

FIG. 8 is a flowchart illustrating an operation of the load distribution unit 400 illustrated in FIG.

FIG. 9 is a flowchart illustrating a file sharing request process shown in FIG. 8;

FIG. 10 is a flowchart illustrating a file sharing response process illustrated in FIG. 8;

FIG. 11 is a block diagram showing a configuration of a modified example of the same embodiment.

[Explanation of symbols]

100 _{1 to} 100 _n client 200 external network 300 file server 400 load distribution unit 401 external network interface 402 control unit 403 storage unit 600 ₀ to 600 _m node 602 data integration unit 603 file sharing processing unit 604 load monitoring unit 605 file processing unit 607 Disk 608 Cache memory 609 Cache control unit 610 File input / output unit

Claims (10)

[Claims] 1. A computer comprising: a plurality of nodes each of which accesses a file in response to an access request from a client and outputs at least load balancing information for each access; And a load distribution means for transferring the access request to a node satisfying a predetermined condition among the plurality of nodes. 2. The file access control according to claim 1, wherein the computer functions as updating means for updating a table storing the load balancing information every time the load balancing information is input. program. 3. The computer according to claim 1, wherein said computer functions as a selecting means for selecting a predetermined load distribution method from a plurality of load distribution methods at predetermined time intervals according to the load distribution information stored in said table. 3. The file access control program according to claim 2, wherein the distribution unit transfers the access request to the node based on the load distribution method selected by the selection unit. 4. Each of the plurality of nodes outputs response information obtained by adding the load distribution information to file information on the file to the load distribution unit, and the load distribution unit outputs the load distribution information from the response information. The file access control program according to any one of claims 1 to 3, wherein the information is deleted and the file information is transferred to the client. 5. The file access control according to claim 1, wherein the load distribution information includes at least information of a load factor and a number of file accesses in the node. program. 6. The load distribution information according to claim 1, wherein at least information of a file stored in a cache memory of each node is included.
5. The file access control program according to any one of 5. 7. The load distribution means sums up points for each node depending on whether a predetermined condition corresponding to the load distribution method selected by the selection means is met, and a plurality of points are calculated based on the total points. The file access control program according to any one of claims 3 to 5, wherein the access request is transferred to a node having the highest total score among the nodes. 8. A plurality of nodes each of which accesses a file in response to an access request from a client and outputs at least load balancing information for each access; and And a load distribution means for transferring the access request to a node satisfying a predetermined condition among the plurality of nodes. 9. A method for selecting a predetermined load distribution method from a plurality of load distribution methods in accordance with load distribution information stored in a table storing the load distribution information at predetermined time intervals, comprising: 9. The file access control device according to claim 8, wherein the load distribution unit transfers the access request to the node based on the load distribution method selected by the selection unit. 10. A load distribution information output step in which a file is accessed in response to an access request from a client, and at least load distribution information is output from each of a plurality of nodes for each access; A load distribution step of transferring the access request to a node satisfying a predetermined condition among the plurality of nodes based on the distribution information.