JP2003058521A - Load management system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a load management system which easily sets a server requiring preferential processing requested from a preference client or maintenance. SOLUTION: A load management system 1-2 which is connected to a lot of clients 1-1 and a lot of servers 1-3 for processing requests of the clients and recognizes the load value of each server 1-3 and distributes requests from the clients 101 to respective servers 1-3 so as to uniform the load capabilities of respective servers 1-3 is provided with a load value management table 1-4 for managing the load value of each server 13, a setting means 1-5 which sets preference clients to which priority of processing should be given independently of the load values with respect to each server 1-4, and a reservation means 1-7 which falsely raises the load value of a set server, for which the requesting client 1-1 is set as a preference client, to suppress requests to the set server from other clients in the case of presence of the set server.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load management system which is connected between a plurality of servers and a plurality of clients and manages the load of each server in response to a request from the client to the server.

[0002]

2. Description of the Related Art FIG. 8 is a block diagram showing the configuration of a conventional distributed processing system. In the figure, 7-1 is a client computer, 7-2 is a server S2, which is provided with a server location management program, data of the server name executable for each service, the processing capacity of each server, and the load state of each server. A management database 7-3 that manages is stored.
7-4 is another server S1.

Next, the operation of the conventional distributed processing system configured as described above will be described. First, the client computer 7-1 sends a location inquiry request to the server 7-2 to search for a server that supports the service "B".
Call. Next, the data of each server in which the service "B" is entered as executable in the management database 7-3 of the location management program of the server 7-2 is acquired, and as a location inquiry response, the client computer 7-
Send to 1.

Next, the client computer 7-1 holds this data. Next, the client computer 7-1 searches for and selects a server with the lightest load (large remaining load capacity) from the server group providing the service "B". Here, the server 7-4 is selected as having a large remaining load capacity. Next, the service "B" is requested to the selected server 7-4. At this time, the value of the load state of the server 7-4 is also updated at the same time.

[0005]

In the conventional load management system configured as described above, the client computer acquires the load information of each server and determines the server requesting the service based on the load information. Therefore, unless a request is issued from the client computer, or load information of the server group is collected from the client,
Unless the client computer regularly collects the load information from the server group, the load information of each server held in the client computer is not revised.

Therefore, since the load value information is not appropriate at the present time, a large number of client computers may simultaneously judge that the load on the server is light based on this information, and the requests may momentarily concentrate on the server. Occurs. Then, the load on the server increases rapidly, and the request cannot be resolved within a desired processing time. Further, since the request is accepted even when the load on the server reaches the maximum value, an error does not occur and the processing efficiency of the client computer deteriorates (the processing time increases) until a timeout occurs at the protocol level.

Further, it is not possible to set the priority or access restriction by the client computer or processing.
Therefore, requests from client computers that preferentially process purchase contracts and payments on the system are
There has been a problem that a phenomenon occurs in which the load of a large number of client computers for the purpose of browsing is obstructed.

Further, when performing maintenance on a specific server, it is difficult to cut off the load on the server and remove it from the system.

The present invention has been made to solve the above problems, and provides a load management system capable of easily setting a server requiring priority processing of a request from a priority client or maintenance. The purpose is to do.

[0010]

[Means for Solving the Problems] Claim 1 according to the present invention
Load management system is connected to multiple clients and multiple servers that process requests from each client, grasps the load value of each server, and balances the remaining load capacity of each server from the clients. In a load management system that distributes requests to each server, a load value management table that manages the load value of each server, a setting unit that sets a priority client that gives priority to processing regardless of the load value for each server, and a request If there is a setting server in which the client that occurred is set as the priority client, it is equipped with a reservation unit that artificially increases the load value of the setting server and suppresses requests from other clients to the setting server. is there.

According to a second aspect of the load management system of the present invention, in the first aspect, when the load value of each server is regularly measured and updated, the reservation means pseudo-loads the load value. The ascended server does not periodically measure the load value when a request from the priority client is received, and when the request from the priority client ends, the load value is temporarily measured and the load value is updated. Shall be carried out on a regular basis.

According to a third aspect of the load management system of the present invention, the load management system is connected between a plurality of servers and a plurality of clients that make a request to each server, grasps the load value of each server, and In a load management system that distributes requests from clients to each server so that the remaining load capacity of the servers is evened out, the load value management table that manages the load values of each server and the load value in the load value management table have a predetermined value If the above is the case, the load value is artificially increased to a predetermined value with respect to the determination means that does not respond to the request from the client and the predetermined server that needs to perform maintenance, and each client transmits the predetermined server. And a maintenance means for performing a maintenance of a predetermined server.

According to a fourth aspect of the load management system of the present invention, in the third aspect, when the load value of each server is regularly measured and updated, the maintenance means artificially determines the load value. The server that has been raised does not perform periodic load value measurement when maintenance is started, and temporarily performs load value measurement when maintenance is completed to update the load value and periodically perform load value measurement. Is to be implemented.

According to a fifth aspect of the load management system of the present invention, the load management system is connected between a plurality of servers and a plurality of clients that make a request to each server, grasps the load value of each server, and In a load management system that distributes requests from clients to each server so that the remaining load capacity of the servers is evened out, the load value measurement program is executed periodically regardless of whether or not there are requests to each server. And load value updating means for updating the load value of each server.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiment 1. 1 is a block diagram showing a configuration of a load management system according to a first embodiment of the present invention. FIG. 2 is a diagram showing data registered in a load value management table of the load management system shown in FIG. Three
FIG. 2 is a diagram showing a flowchart for explaining the operation of the load management system shown in FIG. FIG. 5 is a view showing a flow chart for explaining a load value measuring method by the load value updating means.

In the figure, 1-1 is a plurality of client computers (hereinafter referred to as clients), 1-3 is a server computer (hereinafter referred to as a server) that provides a service (request) requested by these clients 1-1. ). Reference numeral 1-2 is a load management system connected between the client 1-1 and the server 1-3, which grasps the load value of each server 1-3 and equalizes the remaining load capacity of each server 1-3. The request from the client 1-1 is distributed to each server 1-3 as described above. The load management system 1-2 includes the following.

Reference numeral 1-4 is a load value management table for managing the load value of each server 1-1, and 1-5 is a setting for setting a priority client that gives priority to processing regardless of the load value for each server 1-3. If there is a setting server in which the requesting client 1-1 is set as a priority client, the means 1-6 artificially increases the load value of the setting server and requests from other clients to the setting server. Reservation means 1-7 that suppresses the load value of each server is periodically measured by, for example, a load value measurement program (benchmark) as shown in FIG.
-4 is a load value updating means for updating the load value of each server 1-1.

The operation of the load management system of the first embodiment configured as described above will be described. First, the data registered in the load value management table 1-4 will be described with reference to FIG. For each server, the load value updating means 1-7 measures the processing capacity (the value of the load that can be processed) in the unloaded state and registers it as the server maximum performance value.
Next, the load value updating means 1-7 periodically measures the current load value of each server and registers it as the server load current value.

Next, the setting means 1-5 sets a priority client which gives priority to processing regardless of the load value for each server. As the priority client, as shown in FIG. 2, for example, the address of each client is “172.2.
6.1.1.1 ”, respectively.
Next, whether or not the load value is to be measured regularly for each server is set to On or Off in the load measurement stop flag. Normally, the flag is displayed as Off. Also, O
The case where the n flag is displayed will be described later.

Next, a method of periodically measuring the load value by the load value updating means 1-7 will be described with reference to FIG. First, a load value measurement instruction is given periodically (step S-11 in FIG. 5). Next, it is determined whether or not the load measurement stop flag of the server is On (step S-1 in FIG. 5).
2). When the stop flag is On (No), that is, when the load measurement stop flag in FIG. 2 is determined to be Off, the benchmark is executed for that server to calculate the load value (FIG. 5).
Step S-13). Next, the calculated load value is set as the load current value of the server, and the server load current value shown in FIG. 2 is updated. In this way, the load value of the server is regularly updated. In addition, the stop flag is Off (Y
es), that is, when the load measurement stop flag in FIG. 2 is determined to be On, the measurement of the load value for the server is postponed.

Next, the operation of requesting the client of the load management system 1-2 will be described. First, the client 1-1 outputs a request to the load management system 1-2 (step S-1 in FIG. 3). Next, it is determined whether or not the requesting client is registered in the priority client by checking the address of the priority client registered in the load value management table 1-4 (FIG. 3).
Step S-2).

Next, when the server is not registered as a priority client (NO), a server having a large remaining load capacity is selected by judging from the server load current value and the server maximum performance value in the load value management table 1-4. (Step S- in FIG. 3)
3). In the case of the load value as shown in FIG. 2, the server B is selected as having the largest remaining load capacity. Then, the client request is transmitted to the selected server (step S-4 in FIG. 3).

If the client is registered as a priority client in step S-2 (YES), the current load value of the specific server registered in the priority client is artificially increased to the maximum performance value. At the same time, the current load value update stop flag is set to On to cancel the periodic load value update (step S-5 in FIG. 3). Next, the request from the client is transmitted to the specific server (step S-6 in FIG. 3). Next, when the request from the priority client is completed, the load value of the specific server is measured at that time. Along with this, the update stop flag is turned off.
Then, the load value is regularly measured (step S-7 in FIG. 3).

Specifically, the client is "172.2.
In the case of 6.1.11 ″, the server A that has registered this client address as the priority client is selected.
Then, the current load value 10 of the server A is artificially increased to the server maximum performance value 100. Then, the load measurement stop flag is set to On. By doing this, even if a request is made from another client next time, the remaining load capacity becomes 0, and the remaining load capacity becomes the smallest. Therefore, the requests from other clients are not distributed, and the requests of only the priority client can be dealt with.

Further, when the request from the priority client is being dealt with, it is not necessary to update the server load current value (if updated, it makes no sense to artificially increase the load value).
Therefore, it is necessary to stop measuring the load value. However, when the request from the priority client ends, the load value is measured at that time and the server load current value is updated. This is because the server load current value is updated regularly, so it may take some time before the next load value is updated.

The load management system of the first embodiment configured as described above can preferentially process the request of the priority client set by the server. Further, during the processing, since the requests from other clients are not processed, the processing can be speeded up. Further, since the load value is updated regularly, it is easy to understand the load state of the server, and the request processing of the client can be distributed to an appropriate server.

In the first embodiment, one client is set as the priority client. However, the present invention is not limited to this, and it goes without saying that a plurality of clients can be set. Further, it goes without saying that the priority client can be changed appropriately according to the intention of the server.

Embodiment 2. 6 is a block diagram showing a configuration of a load management system according to a second embodiment of the present invention,
FIG. 7 is a view showing a flowchart for explaining the operation of the load management system shown in FIG. In the figure,
Reference numeral 2-1 is a plurality of client computers (hereinafter, referred to as clients), and 2-3 is a server computer (hereinafter, referred to as a server) that provides a service (request) requested by these clients 2-1.

Reference numeral 2-2 is a load management system connected between the client 2-1 and the server 2-3, which grasps the load value of each server 2-3 and determines the load capacity remaining of each server 2-3. The requests from the client 2-1 are distributed to the respective servers 2-3 so as to be equalized. The load management system 2-2 includes the following.

Reference numeral 2-4 is a load value management table for managing the load value of each server 2-1. Reference numeral 2-5 is not compatible with the request from the client when the load value in the load value management table becomes a predetermined value or more. The judging means 2-6 artificially raises the load value to a predetermined server for which maintenance is required, cuts off the request from each client to the predetermined server, It is a maintenance means for performing maintenance. What is maintenance?
There may be a case where the server is adjusted practically, or a case where the server is temporarily separated from the client to see the situation.

The operation of the load management system of the second embodiment configured as described above will be described. First, as the data registered in the load value management table 2-4, the data shown in FIG. 2 is registered as in the first embodiment. The load value updating method is the same as that in the first embodiment.

Next, a maintenance operation for the server of the load management system 2-2 will be described. First, a maintenance request is output from the predetermined server 2-3 to the load management system 2-2 (step S-2 in FIG. 7).
1). Next, the maintenance means 2-6 artificially raises the current load value of the server that has issued the maintenance request to the maximum performance value. At the same time, the current load value update stop flag is set to On to cancel the periodic load value update (step S-22 in FIG. 7).

Next, the predetermined server is maintained (step S-23 in FIG. 7). Next, when the maintenance is completed, the load value of the predetermined server is measured at that time. At the same time, the update stop flag is turned off to periodically measure the load value (step S- in FIG. 7).
24).

Specifically, when the server D requests maintenance, the load present value 0 of the server D is artificially increased to the server maximum performance value 100. Then, the load measurement stop flag is set to On. By doing so, even if a request is issued from the client, the determining means 2-5 determines that the load value in the load value management table 2-4 is equal to or greater than a predetermined value (maximum server performance value) and the request from the client Not compatible with. Therefore, the request from the client is not distributed, and the server can be substantially separated from the network.

Further, when the server is performing maintenance, it is not necessary to update the server load current value (if it is updated, the load value is artificially increased and it becomes meaningless to disconnect from the network), so the load value is measured. Need to be stopped. However, when maintenance is completed,
At that time, the load value is measured and the server load current value is updated. This is because the server load current value is updated regularly, so it may take some time before the next load value is updated.

The load management system according to the second embodiment configured as described above can process a request for maintenance of the server substantially separated from the network.

Although the case of setting the priority client of the first embodiment and the case of performing the maintenance of the second embodiment have been described separately, the load management system has the functions of both the embodiments. It goes without saying that you can do so. Further, it goes without saying that a load management system that regularly updates the load value of the server may be used for the conventional relationship between the client and the server.

[0038]

As described above, according to claim 1 of the present invention, a plurality of clients and a plurality of servers that process requests from each client are connected, and the load value of each server is grasped, In a load management system that distributes requests from clients to each server so that the remaining load capacity of the servers is equalized, a load value management table that manages the load values of each server and processing is performed regardless of the load value for each server. If there is a setting means that sets each priority client and a setting server in which the requesting client is set as the priority client, the load value of the setting server is artificially increased and the setting server from other clients is set. Since it has a reservation means that suppresses the request to the client, it processes the request of the priority client set by the server with priority. Possible to provide a load management system that can become.

According to claim 2 of the present invention, in claim 1, when the load value of each server is regularly measured and updated, the load value is artificially increased by the reservation means. When the request from the priority client is received, the server does not measure the load value periodically, and when the request from the priority client is completed, the load value is temporarily measured to update the load value and the load value is periodically updated. Since the measurement is performed, it is possible to provide a load management system capable of surely prioritizing the processing of the priority client set by the server.

Further, according to claim 3 of the present invention, the load value of each server is grasped by being connected between the plurality of servers and the plurality of clients making requests to each server. In a load management system that distributes requests from clients to each server so that the remaining capacity is equalized, the load value management table that manages the load values of each server and the load value in the load value management table are equal to or greater than a predetermined value. And a determination means that does not respond to the request from the client,
Maintenance means for artificially increasing a load value to a predetermined value for a predetermined server that needs to be maintained, blocking requests from each client to the predetermined server, and performing maintenance on the predetermined server. Since it is equipped with, it becomes possible to provide a load management system that can easily perform server maintenance.

According to claim 4 of the present invention, in claim 3, when the load value of each server is regularly measured and updated, the maintenance means artificially increases the load value. When maintenance starts, the server does not perform periodic load value measurement, and when maintenance ends, it temporarily measures the load value to update the load value and also performs the periodic load value measurement. Because
It is possible to provide a load management system that enables easy and reliable server maintenance.

Further, according to claim 5 of the present invention, the load value of each server is grasped by being connected between the plurality of servers and the plurality of clients making a request to each server and grasping the load value of each server. In a load management system that distributes requests from clients to each server so that the remaining capacity is equalized, the load value measurement program is executed regularly to calculate the load value regardless of whether there is a request to each server. Since the load value updating means for updating the load value of each server is provided, it is possible to provide the load management system in which the load status of the server can be easily grasped.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a load management system according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a load value management table according to the first embodiment of the present invention.

FIG. 3 is a diagram showing a flowchart for explaining the operation of the load management system according to the first embodiment of the present invention.

FIG. 4 is a diagram showing an example of a benchmark.

FIG. 5 is a diagram showing a flowchart for explaining a load value updating operation according to the first embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of a load management system according to a second embodiment of the present invention.

FIG. 7 is a diagram showing a flowchart for explaining the operation of the load management system according to the second embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a conventional load balancing system.

[Explanation of symbols]

1-1, 2-1 Client computer, 1-2, 2-2
Load management system, 1-3, 2-3 server computer,
1-4, 2-4 load value management table, 1-5 setting means, 1-6 reservation means, 1-7 load value updating means, 2-
5 Judgment means, 2-6 Maintenance means.

Claims (5)

[Claims] 1. Connected to a plurality of clients and a plurality of servers that process requests from each of the clients, grasp the load value of each of the servers, and equalize the remaining load capacity of each of the servers. In a load management system that distributes requests from clients to each of the above servers, a load value management table that manages the above load values of each of the above servers, and a priority client that prioritizes processing regardless of the load values are set for each of the above servers. And a setting server in which the client making the request is set as the priority client, the load value of the setting server is artificially increased and a request from another client to the setting server is made. A load management system comprising: a reservation unit that suppresses the load. 2. When the load value of each server is regularly measured and updated, the server whose load value is artificially increased by the reservation means periodically loads the load value when a request from the priority client is received. The measurement is not performed, and when the request of the priority client is completed, the load value is temporarily measured to update the load value and the load value is periodically measured. The load management system described in. 3. Connected between a plurality of servers and a plurality of clients that make requests to each of the servers, grasp the load value of each of the servers, and balance the remaining load capacity of each of the servers. In a load management system that distributes requests from the clients to the servers, a load value management table that manages the load values of the servers,
Determination means for not responding to the request from the client when the load value in the load value management table is equal to or greater than a predetermined value,
For a given server that requires maintenance, increase the load value to the given value in a pseudo manner,
A load management system comprising: a maintenance unit that cuts off requests from the clients to the predetermined server and performs maintenance on the predetermined server. 4. When the load value of each server is regularly measured and updated, the server whose load value is artificially increased by the maintenance means periodically measures the load value when maintenance is started. 4. The load management according to claim 3, wherein the load value is temporarily measured when the maintenance is completed, the load value is updated, and the load value is periodically measured when the maintenance is completed. system. 5. A server is connected between a plurality of servers and a plurality of clients that make a request to each server so that the load value of each server can be grasped and the remaining load capacity of each server can be equalized. In a load management system that distributes requests from the clients to the servers, the load value measurement program is periodically executed to calculate the load values regardless of whether or not there is a request to the servers. A load management system comprising a load value updating means for updating the load value of the load management system.