JP2002288148A - Distributed server system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a distributed server system capable of load distribution in response to the change of load state with the lapse of time. SOLUTION: The system monitors the load state of a server 101 which is executing processing of a client terminal 103, and to prevent the load from increasing with the lapse of time, performs transfer of incomplete processing between servers by an agent program automatically. A packet used in the first server is capsulated and communication with a server 102 taking over the processing is executed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a server-client type system for processing data, executing applications, and the like. In particular, in order to reduce the load on the server, a plurality of servers are installed and distributed. The present invention relates to an improvement in a distributed server system that performs a process so that a continuation of a process accompanying a server change can be taken over.

[0002]

2. Description of the Related Art A server / client type system includes:
When processing requests from a plurality of client terminals concentrate on one server, a load is generated on the server and the network,
There is a problem that the execution speed of the requested processing is significantly reduced.

For this reason, in order to distribute the generation of the load, a plurality of servers capable of performing the same processing are prepared, and when a plurality of processing requests are generated, the processing is distributed to the respective servers so that the concentration of the load is concentrated. The concept of avoiding distributed servers has been introduced.

[0004]

However, in the distributed processing performed in the conventional distributed server system, in a communication procedure performed at the time of the first processing request from a client, a server with a light load is detected and only the server is accessed. If the user decides to do so and wants to access another server, it is necessary to repeat the communication procedure. if,
When transferring access to another server in the middle of the process, perform the communication procedure to the other server again after completing the process, or discard the process in the middle of the process and perform the communication procedure to the other server again. Therefore, the discarded process is redone from the beginning.

[0005] In recent years, client terminals can be connected to a network regardless of indoors or outdoors and connection time by providing a LAN in a company or a school, or by wireless communication means such as a mobile phone and a wireless LAN. As a result, the load state associated with various processing performed by a single server also increases, such as that the server with a light load at a certain time increases in load with the increase in processing requests from client terminals within a few minutes. Large fluctuations occur every short time.

Conventionally, the number of client terminals requesting a server for processing has increased, and it has not been assumed that the load on the server changes in a short time. For this reason, in a conventional distributed server system, when a process is requested from a client terminal, a server with a light load only performs the process according to the load state of the server.

[0007] When the number of client terminals increases dramatically, requests from each client terminal become diversified, and the load value and processing time greatly differ depending on the type of processing performed by the server. For this reason, the load state of each server fluctuates greatly with the passage of time, and a significant difference occurs in the load between the servers.

In order to distribute the load among the servers in response to the load state which changes in a short time, the load on the server is constantly monitored, and when the load on the connected server increases, the other loads are light. You need to switch to the server.

However, the conventional distributed server system has not been designed to monitor the load that changes with time and to perform a procedure for appropriately taking over the connection of the server to another server according to the change in the load on the server. In addition, when the communication procedure is performed again and the server is moved, the processing is discarded halfway and the procedure is performed again on another server, which increases the load.

As a result, even if the load on the server changes in a short time and there is a server whose load is lighter than that of the connected server, the client will continue processing on the first server, and the load will be heavy. There is a problem in that work is continued on the server, and efficient load distribution cannot be performed.

[0011] Agent program technology is one that assists the processing request of the client terminal. The agent program technology automatically executes a process performed by a user of a client terminal on behalf of the user. At this time, since the processing performed by the user is programmed in advance, the processing is performed without the user being conscious of the processing.

However, in the conventional agent program technology, all processes required by a client terminal are determined at an initial stage, and all the processes are performed by an agent program. It transmits communication packets to terminals and does not have a function to distribute the load on the server.

As described above, in the current client-server system and agent program technology, continuation that does not depend on the processing state is applied to a load that greatly changes in a short time due to processing requests from various client terminals. It is hard to say that it was not designed to take the procedure into account and was not able to handle it.

[0014] Accordingly, an object of the present invention is to solve the above-mentioned problems of the prior art, taking into account the time change of the server load, and when the load on the server becomes heavier, the connection setting and the header information of the packet. The purpose of the present invention is to enable the effective use of a server to be easily performed by allowing another server to take over the processing in an intermediate state without performing a procedure such as a large-scale change.

[0015]

According to a first aspect of the present invention, there is provided a communication system comprising at least one client terminal capable of bidirectional communication and a plurality of servers connected via a network. The gist of the present invention is to provide a process continuation execution unit that allows another server to take over the above process from a server that is executing the process requested by the client terminal according to the load on the server.

According to a second aspect of the present invention, in the first aspect of the present invention, a threshold value of an executable process load is set in the server, and the processing continuation executing means sets the load of the process being executed to the threshold value. The point is that when it exceeds the limit, it will start taking over the process to another server.

According to a third aspect of the present invention, in the second aspect of the invention, the processing continuation executing means is configured to cause the other server to take over the processing within a range not exceeding the threshold value. And

According to a fourth aspect of the present invention, in any one of the first to third aspects of the present invention, the processing continuation executing means detects a processing load of the server, and executes processing in an intermediate state in load distribution between servers. The point is to have an agent program that executes takeover between servers.

According to a fifth aspect of the present invention, in any one of the first to fourth aspects of the present invention, the client terminal communicates a packet encapsulating a communication packet to the server to the other server. It is assumed that it has become.

According to a sixth aspect of the present invention, in the fifth aspect of the present invention, the encapsulated packet is obtained by encapsulating, as transmission data, a communication packet including a header which is an address of the server and processing data. The gist of the invention is to add a header which is the address of the other server.

With the configuration of each of the above-mentioned inventions, when the load on the connected server increases, the procedure for setting up the connection and changing the header information of the packet on a large scale is not performed.
Intermediate processing can be taken over by another server, the server connected to the client terminal can be switched to another server with a lighter load, the load can be efficiently distributed among the servers, and server resources can be saved. Enable effective use.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention is applied to, for example, a server-client type system comprising a plurality of clients and a plurality of servers which can be communicated bidirectionally via a network. This system includes a client terminal and two or more servers (server 1 and server 2) that can perform processing in response to a request from the client terminal. While the server 1 is executing a process based on a request from the client terminal, the process takeover execution means:
When the load accompanying the request is taken over from the server 1 to the server 2 for the purpose of distributing the load between the server 1 and the server 2 when the load on the server 1 increases, the server 1 Irrespective of the processing state of the server 2, the server 2 continuously takes over the processing.

In this case, a threshold value of a load that can withstand all servers may be set, and when the current load value exceeds the threshold value, the processing may be started. Also,
In all the servers, the threshold value is compared with the current load value of the server 2, and when the current load value does not reach the threshold value, the server 2 takes over the processing within a range not exceeding the threshold value. May be accepted. Further, the means may include an agent program which detects a load of the server on behalf of a user using the client terminal and automatically takes over the processing in the middle state in the load distribution between the servers between the servers. You may have it.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the distributed server system of the present invention will be described below with reference to the drawings. FIG. 1 shows how an agent moves between a client and a server in the distributed server system of the present invention. Thick wavy line (10
5, 110) indicates that communication is in progress, and solid lines (106 to 109) indicate movement of the agent program. Server A (10
1) and the client C (103) are connected to the network 10
4 (105), the server A (10)
It is assumed that the load of 1) exceeds a threshold. Client C (1
03), the server A (101) continues the processing that was being performed with the server A (101), and the client A (103) notifies the client C (103) that the processing is to be taken over between the servers in order to distribute the load. ), An agent for performing a continuation procedure of processing to another server is transmitted (106).
The client C (103) that has received the agent sets in advance whether or not to take over, and when taking over, notifies the agent of the location information of the client, and hands over the agent to the server A (101) (107). ). Server A that received the agent
(101), when a request for takeover is received from the client C (103), the server A (101) fetches processing variables representing the processing being performed at that time and the state of the processing up to now, and searches the network 104. If the agent finds a server (here, the server B (102)) whose load status is more than the threshold value than the server A (101), the agent sends the processing variable and the position information of the client to the server B (102). Notification is made (108). The agent takes in the location information (address) of the server B (102) and returns to the client C (103) (10
9). The client C (103) acquires the position information of the server B (102) from the agent, and
In 2), the process is continuously executed (110).

FIG. 2 shows the flow of the agent program. In the figure, when the load of the server A exceeds the allowable load threshold (S201), the agent program notifies the client C that the processing is to be taken over (S202). The client C determines in advance whether to take over the processing (S203),
If the handover is not performed, the server A is notified of the fact (S204). When taking over, the agent program automatically acquires the address of the client C and returns it to the server A (S205). The agent program acquires a processing variable indicating the processing state at that time from the server A (S206), and searches for a server on the network (S207). Here, it is determined whether or not there is a server with a sufficient load up to the threshold (S208). If there is a server B with a sufficient load up to the threshold, the server B is notified of the address and the processing variable of the client C. (S209), the address of the server B is obtained (S21)
0). Then returns to Client C, Client C
By acquiring the address of the server B, the server B can continue the processing (S211). When it is determined whether there is a server having a load with a margin up to the threshold (S208), if there is no server with a load with a margin up to the threshold, the agent program returns to the server A and returns the processing variable (S212). . Thereafter, the agent program is transmitted to the client, but at this time returns without having the address information, which means that the server A continues the processing (S21).
3).

As described above, the content of the agent program used in the present invention is such that when the load on the server exceeds a set threshold value, the server notifies the client of the processing takeover for load distribution, Instead, it takes in the client address and the processing variables generated by that server, searches the network for a server with a load that can afford the threshold, and if present, notifies the new server of the client address and processing variables. If not, the method further includes a step of notifying the original server of the client address and the processing variable.

Each of the above processing variables is a numerical data of an application name, data being processed, and a working state being processed, which are information required for the server to be able to take over the processing. is there.
This processing variable is common to the servers A and B.

FIG. 3 shows an example of a coating for encapsulating a communication packet after taking over to the server B. When the client sends out the processing data 34 after being handed over to the server B, the client 36 encapsulates the packet 36 to which the header information 31 that has been sent so far is added as transmission data 35, and A new packet 33 to which the information 32 is added is generated and transmitted to the server B. The server B that has received the transmission data 35 from the received packet 33. From the header information 31 of the transmission data 35, the processing data 34
Continuation of the processing based on the As a result, the client can communicate with the server B using the header information 31 of the packet communicated with the server A without performing a procedure such as address conversion.

As described above, if a packet for communication with the server B is obtained by encapsulating the packet, it is not necessary to directly modify the original packet for communication. As a result, even if the server is changed in the middle of the processing, the packet information is not changed at all, so that the processing between the servers is taken over according to the load state, and the packet being transmitted to the server A is Even if the data is divided and transmitted to the server B, the continuation of the transmission can be determined based on the header information of the original communication packet.

After the encapsulated packet is transmitted to the server B by the outer packet information 32, the server B decapsulates and reads the internal packet to perform processing. Since the server B can know the internal packet information, the server B can determine which client has taken over the processing in the server A.

[0031]

As described above, according to the distributed server system of the present invention, procedures such as connection setting and large-scale change of packet header information are performed in accordance with the load situation that changes greatly in a short time. Without interruption, it is possible to take over the processing in the middle state to another server, switch the connected server to a server with a lighter load, perform efficient load distribution between servers, and Enable effective use.

[Brief description of the drawings]

FIG. 1 is a diagram showing a flow of processing continuation in one embodiment of the present invention.

FIG. 2 is a diagram showing a flowchart of an agent program in the embodiment.

FIG. 3 is a schematic diagram showing encapsulation of a processing packet after movement in the embodiment.

[Explanation of symbols]

 31 header information on server A and client C 32 header information on server B and client C 33 transmission packet to server B 34 processing data 35 transmission data 36 transmission packet to server A 101 server A 102 server B 103 client terminal C104 network

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Katsuki Inamura 6-27-8 Jingumae, Shibuya-ku, Tokyo F-term in Kyocera D.I.

Claims (6)

[Claims] 1. A server, comprising at least one client terminal capable of bidirectional communication and a plurality of servers connected via a network, wherein a server executing a process requested by the client terminal loads the server. A processing continuation executing means for causing another server to take over the above processing in response to the request. 2. A threshold value of an executable process load is set in the server, and the process continuation executing means takes over the process to another server when the load of the process being executed exceeds the threshold value. 2. The distributed server system according to claim 1, wherein the distributed server system is started. 3. The distributed server system according to claim 2, wherein said processing continuation executing means causes the other server to take over the processing within a range not exceeding the threshold value. 4. The processing continuation execution means includes an agent program for detecting a processing load of the server and executing a transfer of a processing in an intermediate state between servers in load distribution between the servers. The distributed server system according to claim 1. 5. The server according to claim 1, wherein the client terminal communicates a packet encapsulating a communication packet to the server to the other server. A distributed server system as described. 6. The encapsulated packet is obtained by encapsulating, as transmission data, a communication packet including a header, which is an address of the server, and processing data, and adding a header, which is an address of the other server, to transmission data. 6. The distributed server system according to claim 5, comprising: