JPH07160647A - Distributed processing system - Google Patents

Abstract

PURPOSE:To easily change constitution, to perform service with an absolute minimum communication amount and to improve reliability as a system. CONSTITUTION:A client 10 is provided with a connection destination table 11 for indicating the server of a service request destination. When a service request is generated, the client 10 requests the service to the server 20 corresponding to the service based on the connection destination table 11. The server 20 is provided with a constitution information table 21 for indicating the connection destination for the service. When the service request is received, the server 20 obtains the connection destination based on the constitution information table 21. When the connection destination is the different server 20, the response of specifying connection to the different server 20 is performed. When the response is received, the client 10 requests the service of the specified different server 20.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed processing system for managing the configuration of a computer group requesting a service and a computer group providing a service in a system in which a plurality of computers are connected on a network.

[0002]

2. Description of the Related Art There is known a computer load balancing method for distributing various transactions to a plurality of computers to improve processing efficiency. Among the computers connected on the network, a computer that requests a service is called a client, and a computer that provides the service is called a server.

In such a client / server system, there is the following method for connecting (or finding) a server that provides a specific service from the client. (A) Give information to the client. The service and the address of the server providing the service are stored in advance on the client, and the client connects to the server when the service is required. (B) Name service (also referred to as directory service) This is described in, for example, Kiyoshi Kaji, "OSF DCE Technical Explanation" 1992.825 Soft Research Center Co., Ltd., p. As described in 125-127, the server stores in advance only the address of the server that provides the name service, and can provide the necessary service to the name server when the service becomes necessary. And ask for the address of the connection destination. Then, similar to the above (a), the target server is connected.

[0004]

However, the above method has the following problems. First,
In the method of (a), since the information of the connection destination server held by the client is fixed, for example, the service A is changed from the server X to the server Y as the number of clients and the number of services increase. We were unable to flexibly respond to configuration changes.

Next, in the case of the method (b), since only the name server needs to change the configuration, the above problem (a) does not occur, but it has another problem. That is, in the method (b), the target server cannot be known unless the name server is always inquired, and therefore the following problems occur.

Deterioration of performance Since the connection is always made to the name server, the connection time is doubled as compared with the case of directly connecting to the server. Increase in network traffic Each time a client requests a service, access to the name server occurs, which increases network traffic around the name server. Failure countermeasures When a failure occurs in the name server and the network around the name server, all clients cannot be used. Moreover, if the name server and the network are duplicated or a plurality of servers are distributedly arranged to deal with this problem, the system becomes expensive. Further, since the client and the name server cannot grasp the service status and the server status, it is impossible to dynamically arrange the clients and servers according to the situation.

The present invention has been made in order to solve the above-mentioned conventional problems, and can flexibly deal with a configuration change, reduce traffic, and dynamically change the client / server according to the situation. An object is to provide a distributed processing system that can be arranged.

[0008]

A distributed processing system according to a first aspect of the present invention is a distributed processing system in which a client, which is a computer that requests a service, and a plurality of servers, which are computers that provide a service, are connected on a network. When the client makes a service request, the client searches the connection destination table, makes a service request to the obtained connection destination server, and, as a result of the service request, the client. And a service requesting unit for making a service request to the connection-specified different server when the server sends a connection-specified response to the different server, wherein the server is a connection destination server for the requested service. The configuration information table shown below and the configuration information table when a service request is received. And a processing determination unit that responds to the request source client by responding to the request source client when the server that executes the service is a different server. Is.

A distributed processing system according to a second aspect of the present invention is the distributed processing system according to the first aspect of the present invention, wherein when the server designates another server for the requested service, the connection destination change to the other server is temporarily changed. A configuration information table having information indicating whether it is static or fixed, and when a service request is accepted, the configuration information table is searched, and if the server that executes the service is another server, the And a processing determination unit that transmits information indicating whether the change of connection destination is temporary or fixed to the request source client, and the client has a response of connection designation from the server to another server, Further, when this connection change is designated as fixed, a processing determination unit for updating the corresponding service request destination in the connection destination table to the server designated for the connection is provided. And it is characterized in and.

In the distributed processing system of the third invention, a client, which is a computer that requests a service, and a plurality of servers, which are computers that provide the service, are connected on a network, and one service is provided by each server. In a distributed processing system configured by combining services to be provided, the client searches a connection destination table indicating a server of a service request destination and the connection destination table when making a service request, and obtains the connection destination obtained. When a service request and service content information is sent to a server and a response to this server request to connect to another server is sent as a result of this service request, the service request is sent to the other server to which the connection is specified. And a service request unit that performs If, from the service content information,
Determine whether the service can be executed only on another server,
When it is determined that it can be executed only by another server, a processing determination unit that responds to the request source client to make a connection designation to the other server is provided.

A distributed processing system according to a fourth aspect of the present invention is a distributed processing system in which a client, which is a computer that makes a service request, and a plurality of servers, which are computers that provide a service, are connected on a network. When making a service request with the connection destination table indicating the destination server, the connection destination table is searched, and the service request and the service content information are sent to the connection destination server that was found, and as a result of this service request. When a response to the connection designation from the server to another server is made, a service request unit for making a service request to the connection-designated another server is provided.
When the server receives a service request, the server predicts the load level when the server executes the service from the service content information, and if the load level exceeds a preset upper limit, another It is characterized in that a processing determination unit for making a response to the server to specify a connection to the request source client is provided.

[0012]

In the distributed processing system according to the first aspect of the present invention, when a service request is generated by a client, first, the connection destination table is searched from the service contents to find the corresponding connection destination server. Next, a communication request is made to this server and a service request is made. As a result, the server searches the configuration information table and finds the connection destination corresponding to the service. When the obtained connection destination is another server, a message specifying connection to the other server is returned to the client.
If the obtained connection destination is the own server, a message that the own server is available is returned. When the client receives a telegram specifying the connection from the server to another server, the client makes a service request again to the specified another server.

In the distributed processing system of the second invention,
In the first aspect of the present invention, the server searches the configuration information table to obtain the connection destination, and when the connection destination is another server, obtains information as to whether the connection change is temporary or fixed. , Return this information to the client.
When the client receives a message for specifying a connection from a server to another server, the client determines from the information contained in the message whether the connection is temporary or fixed, and if it is fixed. Rewrites the corresponding connection destination server in the connection destination table.

In the distributed processing system of the third invention,
When a service request is generated by the client, first, the connection destination table is searched from the service contents to find the corresponding connection destination server. Next, after making a communication connection to this server, a telegram of the service content is transmitted. As a result, the server analyzes the service content information and determines whether the service can be executed only by another server or whether the services performed by a plurality of servers need to be combined. If the result of this determination is that it can be executed only by another server, a message specifying connection to that other server is returned to the client.

In the distributed processing system of the fourth invention,
When a service request is generated by the client, first, the connection destination table is searched from the service contents to find the corresponding connection destination server. Next, after making a communication connection to this server, a telegram of the service content is transmitted. As a result, the server analyzes the service content information, predicts the load level when the service is performed by the server itself, and when this exceeds the upper limit value, a predetermined connection specification to another server is specified. Return the message to the client.

[0016]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the distributed processing system of the present invention. Prior to this, the overall structure of the distributed processing system will be described.

FIG. 2 shows the overall structure. In the illustrated example, five computers 1 to 5 are connected to the network 6, respectively. These computers 1-5
Each of them has a function of a client or a server depending on the situation. For example, when a process occurs in the computer 1 and the process (called a service) is requested to the computer 2, the computer 1 is called a client and the computer 2 is called a server.

Further, the computer 2 (the computer 1
Computer 2 and computer 3) to call computer 3 (to process service requests from
Similarly, the relationship between the client and the server occurs. In this case, when viewed from the computer 1, both the computers 2 and 3 are servers. The present invention controls the configuration of a client and a server when a plurality of such computers are connected to a network.

In FIG. 1, the client 10 includes a connection destination table 11 and a service request unit 12.
The connection destination table 11 indicates the service request destination of the client 10 and is as follows in the first embodiment.

FIG. 3 shows the configuration of the connection destination table 11. As shown in the figure, the connection destination table 11 shows a combination of the service type and the connection destination address of the connection destination computer. For example, the service with the service number 2 makes a connection request to the computer 2. . The connection destination table 11 is specifically stored in a magnetic disk or memory (not shown) of the computers 1-5.

When requesting a service, the service request unit 12 in the client 10 searches the connection destination table 11 and sends a service request message to the requested server, and the server sends the service request message to another server. When a connection is designated, it has a function of making a service request again to the connection designated server.

The server 20 comprises a configuration information table 21 and a process determination section 22. The configuration information table 21 is
It shows a client 10 as a service request source and a server that performs the service based on the requested service.

FIG. 4 shows the configuration information table 21. That is, the configuration information table 21 is information indicating the service number indicating the service content, the connection destination (the number of the server providing the service), and the relationship between the server for the service is temporary or fixed. There is "temporary /
The item “permanent” is included. The configuration information table 21 is stored in a magnetic disk or memory (not shown) of the computers 1-5.

The processing determination unit 22 of the server 20 accepts the service request from the client 10 and searches the configuration information table 21 based on the content of the service and the requesting client 10 to obtain the requested server 20. Is a different server, it has a function of transmitting a message specifying connection to the different server to the requesting client 10.

Further, in FIG. 1, 30 is a network for connecting the client 10 and the server 20.
Although only one client 10 and one server 20 are shown in FIG. 1, there are actually a plurality of clients and a plurality of servers as shown in FIG.

Next, the operation of the distributed processing system of the first embodiment configured as described above will be described. FIG. 5 is an operation flowchart on the client 10 side. When processing occurs in the client 10, first, the service request unit 12
Accepts this (step S1), next, in order to determine where to request the processing, the connection destination table 11 is examined and the service request destination (connection destination) is determined (step S2). For example, when the service required by the client 10 is “2”, it is sufficient to connect to the computer 2. Then, the service requesting unit 12 of the client 10 connects to the server 20 (in this case, the computer 2) via the network 6 (step S3), and when connected, the connection result from the server 20 is converted into a message. It is returned (step S4). The operation of the server 20 will be described later.

FIG. 6 shows a format of a message transmitted and received between the client 10 and the server 20. In the figure, (a) shows a general form of a message, which includes a service type, a parameter 1, which is a message content, a parameter 2,
, And a footer indicating the end of the message. Further, (b) to (d) are electronic messages transmitted from the server 20 to the client 10, and (e) and (f) are electronic messages transmitted from the client 10 to the server 20.

Therefore, in step S3, (e)
A connection request message is sent to the server 20. Then, the message received in step S4 is (b) another server connection,
(C) There are two types of connection to the own server. Step S
When the message is received from the connection destination server 20 in step 4, the service request unit 12 next determines whether the message content is designated by the own server or another server, that is, whether the connection is OK (step S5). Here, if it is the own server connection,
Since the connection change from the client 10 to the server 20 is not necessary, the connection work ends here, and the client 10 subsequently sends (f) a service request message to the server 20.

On the other hand, when it is determined in step S5 that the received message type is another server connection, the service request unit 12 disconnects the connection to the server 20 (computer 2) that is currently connected (step S5). S6), message (b)
The connection to the other server (for example, the computer 4) indicated by the parameter 1 is performed (step S7). If the parameter 2 of the received message specifies permanent,
The connection destination table 11 is updated. That is, it is determined whether the parameter 2 is permanent or temporary (step S8). If it is permanent, for example, the connection destination of the service number 1 is set to the computer 2
To computer 4 (step S9).

As described above, since the contents of the connection destination table 11 can be changed by the electronic message from the server 20, even if the server 20 providing the service is changed, it is necessary to purposely prepare a new connection destination table 11. In addition, it is possible to deal flexibly with changes in the system status.

Next, the operation on the server 20 side will be described. FIG. 7 shows a server 2 corresponding to the operation of the client 10.
It is an operation flowchart of 0. The server 20 first waits for a connection from the client 10 (step S
1). When there is a connection request from the client 10 and this connection is established, the processing determination unit 22 obtains the connection destination corresponding to the requested service from the configuration information table 21 (step S2). That is, the information regarding whether the connection destination is the own server or another server and the information regarding whether the connection destination is temporary or fixed when the other server is extracted.

It is determined from the retrieved information whether the connection destination is the own server or another server (step S3), and if the connection destination is the own server, the message (c) is returned to the client 10 (step S4). ). On the other hand, if it is another server, the electronic message (b) is returned to the client 10 (step S5).

As described above, the configuration information table 2
In addition to the connection destination extracted from 1, the item of the maximum number of connections of the client 10 in the server 20 is further provided, and a counter showing the current number of connections is provided. If this count value exceeds the maximum number of connections, the service Connection to another server may be designated regardless of the number.
In this case, the other server that specifies the connection is determined in advance. In this way, by considering the maximum number of connections, the processing can be distributed among the plurality of servers 20, and the efficiency of the system can be improved.

Furthermore, considering the load status of the own server 20,
The connection to the other server 20 may be designated according to this load condition. That is, when a connection request from the client 10 is received, the current load level of the server computer is obtained from, for example, the operating rate of the CPU, and if this load level exceeds a predetermined threshold value, the configuration information table 21 is used. Even in the case of the connection to the own server, the connection to the other server which is determined in advance is designated.

Then, the connection request source (client) may be added to the item of the configuration information table 21, and the connection destination may be determined by the combination of the service number and the connection request source. FIG. 8 shows the configuration information table 21. That is, when a plurality of clients 10 can request the same service, the computer serving as the server 20 is changed by the clients 10. For example, when the client computer 1 makes a service request for service number 3,
The connection destination is the computer 4, but when the client computer 4 makes the connection, the server itself (computer 2)
Is the connection destination, and the connection destination is determined according to the connection request source.

Next, a second embodiment will be described. In the first embodiment, the connection request from the client 10 {telegram (e)}
While the server 20 returns the connection configuration information (telegram) to the server, in the second embodiment, the server 2 responds to the result of the service request {telegram (f)} from the client 10.
The difference is that 0 returns connection configuration information. That is, in the second embodiment, the connection destination server is determined based on the service content.

For example, a book search service will be taken up and described as an example. It is assumed that the server for providing the book search service is the computer 2. However, it is assumed that the computer 2 has only the search information of books due to the resource, and the computers 3 and 4 are in charge of the search for other newspapers and magazines. Therefore, in such a case, the client 1
When the computer 2 receives a book search service request from 0, the computer 2 returns a telegram of the server connection itself to the client 10. In reality, however, the computer 2 inquires of the computers 3 and 4 and sends the result to the requesting client. Performs operations such as sending. For example, the processing is performed as follows.

The service request A: "output a list of computer-related materials" is received by the computer 2, the book portion is processed by the computer, the newspaper portion is processed by the computer 3, and the magazine portion is processed by the computer. Each processing is performed in 4, and the results are once collected in the computer 2, edited, and returned to the client.

In the service request B: "output a list of computer-related magazines", the computer 2 simply sends this request to the computer 4, and the result is sent to the computer 4.
From the computer 2, the computer 2 receives it and transfers it to the service requesting client as it is. Thus, in the service request B, passing through the computer 2 obviously causes waste of communication. Therefore, in the second embodiment, the service is divided into connection destinations for each component, and the connection destination server is determined by a service request.

The specific operation of the second embodiment will be described below. FIG. 9 shows an operation flowchart of the client.
First, in step S, the service request is accepted, the connection destination is obtained from the service content, and the connection destination is connected.
The operations of 1 to S3 are the same as in the first embodiment. That is,
The client 10 sends a message (e) to the connection destination server 20. Next, unlike the first embodiment,
Send the message (f) without waiting for the result from the server 20,
A service request is made (step S4). As a result, the electronic messages (b) to (d) are sent from the server 20 (step S5).

When the message (d) is sent in the determination in step S6, it means that the service is completed in the connected server, and the service is finished. When the message (b) is sent, the connection to the server 20 is disconnected (step S7) and the new connection destination indicated in the message is connected, as in steps S6 and S7 of FIG. Is performed (step S8). Here, the “temporary / permanent” item of the electronic message (b) is temporary, as described in the operation of the server. After that, a service request is made to the new connection destination (step S9), and the result is received from the server 20 (step S19). Although it is assumed here that the message (d) is returned from the new connection destination, the message (b) is sent in a more general way, as shown by the broken line A in the figure. It is also possible to execute from step S6.

Next, the operation of the server in the second embodiment will be described. FIG. 10 is a flowchart of the operation. First, the server 20 waits for a connection from the client 10 (step S1), and when the connection is made from the client 10, accepts the service request content from the client 10 (step S2). That is, the message (e) is received in step S1, and the message (f) is received in step S2.

Then, the analysis is performed based on the "service request content" of the electronic message (f) (step S3). That is,
As in the case of the book search service described above, when a certain service (this is referred to as service X) is configured to include a plurality of services (each of which is referred to as X1, X2, ..., Xn), each of them is Each server has information as to which server provides the services X1, X2, ..., Xn of. Then, in step S4, the connection destination server is determined from the result of the analysis of the service content. That is,
If the requested service includes services provided by multiple servers, the server becomes the connection destination, and if the requested service can be provided by another single server, the server becomes the connection destination. .

Next, in the connection destination determination in step S5, if the server is the own server, the requested service is implemented and the result is returned to the client (step S).
6). On the other hand, if it is another server, the electronic message (b) is transmitted to the client (step S7). Note that this message (b) may have different connection destinations with the same service number, so as described above, "temporary / permanent"
Items are temporary.

In the second embodiment, the server analyzes the constituent elements of the requested service and determines the connection destination. However, the processing amount of the requested service and the load status of the server itself. It is also possible to determine the connection destination based on, which will be described below as a third embodiment.

FIG. 11 is an operation flowchart of the server in the third embodiment. First, waiting for a connection from a client (step S1) and then accepting the service request content (step S2) is the same as in the second embodiment. Then, the received service request content is analyzed (step S3), and based on the analysis result of step S3, in order to perform the service, the amount of resources used such as the disk and the memory capacity and the processing time of the server are calculated. Calculates how much is required, and predicts the load degree of the own server when this service is performed (step S4).

For example, it is assumed that the service request content from the client is “obtain the average age of men”. Here, the data is for 100,000 people (50% for men and 50% for women), and the processing time is 100 ms per record.
In such a case, the time to execute this service is 50
It can be estimated as 00 seconds.

Next, it is judged whether or not the load level of the self server thus obtained exceeds a predetermined upper limit value (step S5). It carries out and returns the result to a client (step S6). On the other hand, when the load exceeds the upper limit, the connection destination is set as another predetermined server, and the electronic message (b) having the server as the connection destination is returned (step S7). Also in this case, the "temporary / permanent" item is "temporary".

The operation on the client side in the third embodiment is substantially the same as that in the second embodiment shown in FIG. 9, but the operation after step S10 is different. That is, as indicated by a broken line A in FIG. 9, when a message from the server is received in step S10, the process returns to step S6, and it is determined whether the service result at the connection destination is OK. this is,
This is because the new connection destination server also performs the operation shown in FIG. 11, and the result may occur even in the case of another server. In such a case, the above operation is repeated until a server capable of executing the service is found.

[0050]

As described above, according to the distributed processing system of the first aspect of the invention, the client has the connection destination table indicating the server of the service request destination, and when the service request is made, the client uses this connection destination table. The server makes a service request based on the configuration information table indicating the connection destination server for the requested service. When the service request is received, the configuration information table is searched, and the connection destination is another server. Makes a response that specifies the connection to another server, so even if the destination server changes, it is only necessary to send this information in the response from the server to the client. It is possible to flexibly respond to a configuration change. Also, unlike the conventional method, a name server is not required, so the connection time between the client and server is shortened, and
It is possible to obtain a highly reliable client / server system without requiring the minimum amount of network traffic and without stopping the entire system due to a server down.

Further, according to the distributed processing system of the second invention, in the first invention, information indicating whether the connection change is temporary or fixed is added to the connection designation information given from the server to the client to another server, If the connection change to this different server is fixed, the client updates the connection destination table, so even if the connection change from the client to the server occurs, that information is reflected on the client side. Therefore, in addition to the effect of the first aspect of the invention, it is possible to flexibly and easily deal with the configuration change.

According to the distributed processing system of the third invention, when one service is composed of a combination of services provided by the respective servers, the server determines whether the services are different from each other based on the service contents from the client. It is judged whether it can be executed by only one server, and if it is judged that it can be executed only by another server, the connection to that other server is specified to the client, so network traffic is reduced. Moreover, since unnecessary communication is not performed, it is possible to provide the service promptly.

According to the distributed processing system of the fourth aspect of the invention, when the server receives a service request from a client, the load level of its own server when the service is executed is predicted from the service content, and this load level is estimated. When the value exceeds the preset upper limit, the connection to another server is specified to the client, so the load is not biased to a specific server among multiple servers, and the efficiency is high. A distributed processing system can be provided.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a distributed processing system of the present invention.

FIG. 2 is an overall configuration diagram of a distributed processing system of the present invention.

FIG. 3 is an explanatory diagram of a client connection destination table in the distributed processing system of the present invention.

FIG. 4 is an explanatory diagram of a server configuration information table in the distributed processing system of the present invention.

FIG. 5 is an operation flowchart of the client of the first embodiment in the distributed processing system of the present invention.

FIG. 6 is an explanatory diagram of a format of each message in the distributed processing system of the present invention.

FIG. 7 is an operation flowchart of the server of the first embodiment in the distributed processing system of the present invention.

FIG. 8 is an explanatory diagram of a configuration information table including client items of the first embodiment in the distributed processing system of the present invention.

FIG. 9 is an operation flowchart of the clients of the second and third embodiments in the distributed processing system of the present invention.

FIG. 10 is an operation flowchart of the server of the second embodiment in the distributed processing system of the present invention.

FIG. 11 is an operation flowchart of the server of the third embodiment in the distributed processing system of the present invention.

[Explanation of symbols]

 10 client 11 connection destination table 12 service request unit 20 server 21 configuration information table 22 processing determination unit 30 network

Claims (4)

[Claims] 1. A distributed processing system in which a client, which is a computer that requests a service, and a plurality of servers, which are computers that provide a service, are connected on a network, wherein the client is a connection destination indicating a server to which the service is requested. When making a table and a service request, the connection destination table is searched, and a service request is made to the connection destination server that was obtained, and as a result of this service request, a response to the connection designation from that server to another server is made. In the case where the connection request is made, a service request unit for making a service request to the other server specified by the connection is provided, and the server, when receiving the service request, and a configuration information table showing a connection destination server for the requested service, A server that searches the configuration information table and executes the service Is a separate server, the distributed processing system is provided with a processing determination unit that makes a response to the requesting client to specify a connection to the different server. 2. The server, when a different server is designated for the requested service, has configuration information having information indicating whether the change of connection destination to the different server is temporary or fixed. Table and information that, when a service request is received, the configuration information table is searched, and if the server that executes the service is another server, whether the connection destination change is temporary or fixed Is sent to the request source client, the client receives a response from the server to specify a connection to another server, and if the connection change is specified to be fixed, ,
The distributed processing system according to claim 1, further comprising: a processing determination unit that updates the corresponding service request destination in the connection destination table to the server that specifies the connection. 3. A client, which is a computer that makes a service request, and a plurality of servers, which are computers that provide services, are connected on a network, and one service combines the services provided by each server. In the distributed processing system configured, the client searches a connection destination table indicating a server of a service request destination and the connection destination table when making a service request, and makes a service request to the obtained connection destination server. In addition to transmitting the service content information, if a response to this server request for connection designation from another server is made as a result of this service request, a service request unit for making a service request to the other server designated for connection When the server receives the service request, the server provides the service content information. Et al., If the service is determined it can run only on a separate server, is determined can be executed only in a different server,
A distributed processing system, comprising: a processing determination unit that responds to the request source client with a connection designation response to the another server. 4. A distributed processing system in which a client that is a computer that makes a service request and a plurality of servers that are computers that provide a service are connected on a network, wherein the client is a connection destination indicating a server to which a service is requested. When making a table and a service request, the connection destination table is searched, and the service request and service content information are sent to the connection destination server that was found, and as a result of this service request, the server sends the service request to another server. And a service request unit for making a service request to another server for which connection is specified, when the server accepts the service request, the server determines from the service content information Predict the load level when the service is executed, and set this load level in advance. Distributed processing system if it exceeds the upper limit, which is characterized by comprising a determination unit that performs connection specification of the response to another server to the requesting client that.