JP2001167074A - Load balancing processor - Google Patents

Abstract

(57) [Problem] To improve the throughput and processing capacity of the entire system by preventing uneven distribution or overload of a load generated between servers. A load distribution processing device in a client / server system in which a plurality of clients (1a, 1b) and a plurality of servers (2a, 2b) are connected via a network and processing requests from the plurality of clients are distributed among the plurality of servers. The server has priority server assignment information for determining a load assignment destination when the load of each server is substantially the same, and the client determines a connection destination server based on the priority server assignment information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load distribution processing layer value, and more particularly, to a load distribution processing apparatus which distributes processing requests from a plurality of clients equally among a plurality of servers.

[0002]

2. Description of the Related Art When processing requests from a large number of clients occur at the same time, there is a risk that a communication server responsible for this processing will be overloaded, and it is important to prevent this state beforehand. Japanese Unexamined Patent Publication No. Hei 9-106381 discloses that when the load on the server is different,
It shows that load distribution is performed between servers by allocating communication destinations from clients according to the load borne by the servers.

On the other hand, when the load imposed on the server is the same, the load is distributed to the server of the communication destination on the client side in a round robin manner or randomly, and no consideration is given to equally distributing the load among the servers. .

[0004]

SUMMARY OF THE INVENTION The above-mentioned Japanese Patent Application Laid-Open No. 9-106.
According to the technology disclosed in Japanese Patent Publication No. 381, when a processing request is issued from a plurality of clients to a server at the same time, if the load between the plurality of servers is the same, or if the client has just started a job, the client The server of the communication destination is selected in a round robin manner or randomly. For this reason, the load is unevenly distributed to a specific server, and the request to the server cannot be efficiently distributed. Further, if the load is unevenly distributed among the servers, communication processing and CPU processing for load distribution increase. Also, the load information between the servers needs to be updated periodically, but if the update period is too long, a state where appropriate load distribution is not performed continues. The present invention
An object of the present invention is to solve the above-described problems, thereby preventing uneven distribution of loads among servers, and improving the throughput and processing capacity of the entire system.

[0005]

The present invention employs the following means in order to solve the above-mentioned problems.

[0006] In a load distribution processing apparatus in a client / server system in which a plurality of clients and a plurality of servers are connected via a network and processing requests from the plurality of clients are distributed and processed by the plurality of servers,
The server has priority server allocation information for determining a load allocation destination when the load of each server is substantially the same, and the client determines a connection destination server based on the priority server allocation information. I do.

Further, a load distribution processing device in a client / server system having a load control unit for connecting a plurality of clients and a plurality of servers via a network and distributing processing requests from the plurality of clients to the plurality of servers. , The load control unit includes a load information storage unit, and the load information storage unit stores load level information of each server and priority server allocation state information substantially equally allocated to each client; Assigns a processing request to a server based on the load level information and the priority server allocation state information.

Further, a load distribution processing apparatus in a client / server system having a load control unit for connecting a plurality of clients and a plurality of servers via a network and distributing processing requests from the plurality of clients among the plurality of servers. , The load control unit of the client includes a load information storage unit, and the load information storage unit stores load level information of each server and priority server allocation state information that is almost equally allocated to each client. Acquired from a storage unit, the load control unit of the client allocates a processing request to a server based on the load level information and the priority server allocation state information.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. FIG. 1 is a diagram showing a client / server system according to an embodiment of the present invention.
In the figure, 1 is a client, 2 is a server, and 3 is a network connecting a plurality of clients 1 and a plurality of servers 2.

FIG. 2 is a diagram showing details of the client / server system according to the embodiment of the present invention. In the figure, 1a and 1b are clients, and 2a and 2b are servers. In the present embodiment, the client 1b and the server 2b have the same functions as the client 1a and the server 2a, respectively, and their functions are the same or symmetrical.

Reference numeral 11a denotes a communication control unit for communicating with the server 2a; 12a, an application program unit for performing user processing; 13a, a load information area for storing server load information; 14a, a server load information area 18a to be described later;
, A means for updating or referring to the load information of the server stored in the storage area of the load information area 13a, and a means for selecting a communication destination server by referring to the load information.

Reference numeral 15a denotes a communication control unit for communicating with the client 1a; 16a, an application program unit for performing server processing; 17a, a communication means for notifying another server of load information of the server stored in the load area at regular intervals. , And a load control unit including communication means for notifying the load information to another server to be load-balanced when the load of the server is changed, and a load information area 18a for storing the load information of the server.

FIG. 3 shows a load control unit 1 mounted on the server 2a.
It is a figure showing the operation flow of 7a. First, step 30
In step 31, when the online program on the server is started, in step 31, the start of the online program is notified to another server, and the operation information and load information of the other server are acquired. Step 32
, The load information stored in the load information area of the own server is updated based on the acquired operation information and load information of the other server. In step 33, it is determined whether a predetermined time has elapsed. If the predetermined time has elapsed, the process proceeds to step 34, where the load information is acquired from another server, and the load information stored in the load information area in the own server is updated. In step 35, the end of the online program on the server is determined. If the online program ends, the process proceeds to step 36; otherwise, the process proceeds to step 33.
Proceed to. In step 36, the other server is notified of the end of the program of its own server, the load information on the other server is updated, and the process ends.

Next, with reference to FIG. 4 and FIG. 5, processing of the server when a request message is generated from the client to the server will be described. FIG. 4 is a diagram showing a processing flow of the server when a request message is generated from the client to the server, and FIG. 5 is a diagram showing load information stored in a load information area of the server. In the figure, D61 indicates an application name, D62 indicates a distributed server name, D63 indicates a load level, and D64 indicates a priority server assignment state. Thus, the load information area 18a of the server 2a shown in FIG.
For each application program (AP1, AP2)
The server (server 1 and server 2) operating with the load distributed, the load level (F1, F2) of each server, and the current priority server allocation state of the application program are stored.

First, in step 40, the server 2a
Reads a message transmitted by the client 1a. In step 41, it is determined whether or not the client 1a has requested to acquire server priority server assignment state information indicating a communication destination server to be preferentially communicated when the load of each server is the same, If there is an acquisition request, proceed to step 42; otherwise, proceed to step 4
Go to 5. In step 42, the load control unit 17a reads, from the load information area 18a, server information indicating a communication destination server to be preferentially communicated when the load of each server is the same, that is, priority server assignment state information,
Add to the response message to the client. In step 43, the priority server information is updated and the load information area 18a is updated.
Write to. In step 45, the message is read, and it is determined whether there is a load information acquisition request for performing load distribution for each server. If there is a load information acquisition request, the process proceeds to step 46; Step 4
Proceed to 8. In step 46, the load control unit 17a reads the load information from the load information area 18a and adds it to the response message to the client. In step 48,
When the client 1a requests processing to the application program unit 16a of the server 2a, the client 1a passes the message from the client to the application program to perform the processing, and then, together with the priority server allocation state information and the load information, The response message is transmitted to the client 1a.

As described above, the load distribution control unit 17a can perform the load distribution processing based on the load level information for each application program. In addition, load distribution processing can be performed based on load level information for each server, not for each application program. In addition, load distribution processing can be performed based on both pieces of information.

FIG. 6 is a diagram showing details of the processing of step 43 (priority server information updating processing) in FIG. First, in step 81, when there is a processing request from a client, it is determined whether or not the request is the first request from the same client. Whether or not the clients are the same can be determined by an IP address or the like.
If the request is the first request from the same client, the process proceeds to step 82; otherwise, the process ends without updating the priority server assignment status information. In step 82, the own device determines whether or not it is the last server to be load-balanced. If it is determined that the server is the last server, the process proceeds to step 83; otherwise, the process proceeds to step 84. Whether or not the server is the last server is determined based on whether or not the priority server assignment state shown in FIG. 5 is the last server (server 2) in the case of the present embodiment in which the load is distributed among two servers. be able to.

In step 83, the first server is assigned as a priority server. In step 84, the next server is assigned as a priority server. In step 85, the priority server assignment information D64 is updated based on the assigned server information.

By performing such a processing flow,
Since the priority servers are distributed and assigned to each client, the load on the plurality of servers can be evenly distributed.

Next, with reference to FIGS. 7 and 8, the processing of the client when a request message is generated from the client to the server will be described. FIG. 7 is a diagram illustrating a processing flow of the client when a request message is generated from the client to the server, and FIG. 8 is a diagram illustrating the load information stored in the load information area 13a of the client. In the figure, D71 indicates an application name, D72 indicates a distributed server name, D73 indicates a load level, and D74 indicates a priority server assignment state. In the load information area 13a, for each application program (AP1, AP2), a server (server 1, server 2) operating with a distributed load, a load level (F1, F2) for each server, and an application program The current priority server assignment status information is stored.

In step 50, the client application program is started, and a processing request is sent from the application program unit to the server.
In step 51, the communication control unit 11 of the client
“a” makes a single priority server information acquisition request to an arbitrary or designated server, and acquires priority server information. At this time, the load information of the server may be obtained at the same time. In step 52, the acquired priority server assignment status information and load information are updated and stored in the load information area 13a. In step 53, when a second or subsequent processing request is issued to the server, in step 54, load information for each server is acquired based on the acquired information. In step 55, it is determined whether the load of each server is the same. If the load on each server is the same, the process proceeds to step 56; otherwise, the process proceeds to step 57. In step 56, a processing request is made to the priority server based on the priority server assignment information.
In step 57, a processing request is made to a low-load server with a low load.

In step 58, step 53 is executed.
Receives a response message to the processing request sent in step 2. If load information for each server is added to the received response message, the load information stored in the load information area 13a is updated based on the information. In step 59, it is determined whether there is a next request from the client to the server. If there is a processing request, the process proceeds to step 53; otherwise, the process proceeds to step 61 and ends.

As a method of determining the load level in the step of determining whether or not the load of each server is the same, the number of processings per unit time, the number of processing waits, the CPU utilization, and the like are used. it can. Further, the determination that the load of each server is the same may be made by strictly measuring the load level. Alternatively, the determination level may have a width and a width may be provided in a range where the determination is made that the load is the same.

As described above, the load control unit 14a can perform the load distribution processing based on the load level information for each application program. In addition, load distribution processing can be performed based on load level information for each server, not for each application program. In addition, load distribution processing can be performed based on both pieces of information. In updating the load information in step 58, the load level information is updated in D73. However, the priority server allocation status information D74 does not need to be updated unless the number of operating load distribution servers D72 is changed.

FIG. 9 shows a client /
FIG. 1 is a diagram illustrating an example in which a load distribution processing device of a server system is applied to an Internet banking system. Reference numeral 91 denotes a client, 92 denotes a circuit network such as the Internet or an intranet, 93 denotes a WWW server, 94 denotes an AP server that executes application processing such as a transfer business, and 95 denotes a DB server that stores customer information and the like. In this system, a plurality of clients, a WWW server, and an AP server are connected via the Internet.

In such a system, there is a high possibility that processing requests from a large number of clients 91 occur at the same time, so that efficient load distribution processing is required. In particular, a heavy load is expected on a WWW server that receives processing requests from a large number of clients and an AP server that can distribute processing from the WWW server. Therefore, by applying the load distribution processing device of the client / server system to such a system, uneven distribution or overload of a load generated between servers can be significantly prevented.

As described above, according to the present embodiment, when the load of each server is the same, each client selects a communication destination server based on the priority server information. As a result, it is possible to prevent load concentration on a specific server, which occurs when a communication destination server is arbitrarily selected. Further, it is possible to prevent the efficiency of the CPU and the network from being degraded due to the bias of the load, and to improve the throughput and the processing capacity of the entire system.

The above description has been made on the assumption that the client does not have the load information of each server in advance. On the contrary, each client may hold the priority server assignment information D74 in advance. . In this case, FIG.
In the processing flow on the client side shown in FIG.
The priority server information update processing shown in FIG. 2 is unnecessary.

[0029]

As described above, according to the present invention, when connection and processing requests are generated from a plurality of clients, uneven distribution or overload between servers is prevented beforehand, so that the load can be reduced. The distribution can be evenly performed, and the throughput and processing capacity of the entire system can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a client / server system according to an embodiment of the present invention.

FIG. 2 is a diagram showing details of a client / server system according to the embodiment;

FIG. 3 is a diagram illustrating an operation flow of a load control unit mounted on a server.

FIG. 4 is a diagram showing a processing flow of a server.

FIG. 5 is a diagram showing load information stored in a load information area of a server.

FIG. 6 is a diagram showing details of step 43 shown in FIG. 4;

FIG. 7 is a diagram showing a processing flow of a client.

FIG. 8 is a diagram showing load information stored in a load information area of a client.

FIG. 9 is a diagram showing an example applied to an Internet banking system.

[Explanation of symbols]

 1, 1a, 1b, 91 client 2, 2a, 2b server 11, 11a, 11b communication control unit 12, 12a, 12b application program unit 13, 13a, 13b load information area 14, 14a, 14b load control unit 15, 15a, 15b Communication control unit 16, 16a, 16b Application program unit 17, 17a, 17b Load control unit 18, 18a, 18b Load information area 1 92 Circuit network 93 WWW server 94 AP server 95 DB server

Claims (4)

[Claims] 1. A load distribution processing device in a client / server system in which a plurality of clients and a plurality of servers are connected via a network and processing requests from the plurality of clients are distributed and processed by the plurality of servers. Load distribution processing for determining priority server allocation information for determining a load allocation destination when each load is substantially the same, wherein the client determines a connection destination server based on the priority server allocation information. apparatus. 2. A load distribution processing apparatus in a client / server system having a load control unit that connects a plurality of clients and a plurality of servers via a network and distributes processing requests from the plurality of clients to the plurality of servers. In the above, the load control unit includes a load information storage unit, and the load information storage unit stores load level information of each server and priority server allocation state information substantially equally allocated to each client, Wherein a processing request to a server is allocated based on the load level information and the priority server allocation state information. 3. A load distribution processing device in a client / server system having a load control unit that connects a plurality of clients and a plurality of servers via a network and distributes processing requests from the plurality of clients to the plurality of servers. In the above, the load control unit of the client includes a load information storage unit, and the load information storage unit stores load level information of each server and priority server allocation state information that is almost equally allocated to each client. A load distribution processing device obtained from a storage unit, wherein the load control unit of the client allocates a processing request to a server based on the load level information and the priority server allocation state information. 4. The load distribution processing device according to claim 1, wherein the server system is an Internet banking system.