JP2008210110A - Business system and its execution request management method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a business system and an execution request management method, for reducing a time when response is performed to an execution request from a client, and not causing a high load state of a server by a second time execution request from the client. <P>SOLUTION: When the execution request is outputted from a queue, the server calculates and holds a latest queue residence time when the execution request is stored in the queue. The server calculates and holds a latest value of an execution time of each operation executed in the past. When receiving the execution request of the operation from the client, addition of the latest queue residence time and the execution time of the operation executed in the past corresponding to the execution request is performed to estimate the execution time of the requested operation. When an estimation value exceeds a preset upper limit value, a response including an error message of execution impossibility is returned to the client of an execution request source of the operation. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a business system including a server and a client and an execution request management method thereof.

  Generally, a computer system suitable for business in various industrial fields is called a business system. A business system is composed of a computer (server) that centrally manages hardware resources such as printers and modems and information resources such as application software and databases, and a plurality of clients that are computers that use resources managed by the server. Is done.

  In such a business system, processing (operation) using various hardware resources and information resources is requested from each client to the server via the network.

  When the server receives an operation execution request from the client, the client until an application that does not execute the operation (hereinafter referred to as “free process”) among the plurality of server applications occurs. Temporarily store requests in the queue. Then, the requested operation is executed when a free process occurs, and the result (response) is returned to the client. At this time, if the server is in a heavy processing load state (hereinafter referred to as a high load state), a response to the execution request is hardly returned to the client. In addition, when the server is stalled, no response is returned to the client.

  In order to cope with such a high load state, in the conventional business system, an upper limit value of the response time to the client is set in the server in advance by an operator's operation, and the execution time of the operation exceeds the upper limit value Sends a response containing a non-executable error message to the client so that some response is always sent back to the client.

  However, in such a conventional business system, there is a problem that the client user must wait for a response from the server until the upper limit of the response time preset in the server is exceeded. In addition, when a response including an error message is received by a client user operation, the same operation execution request is sent again to the server, which increases the processing load on the server and leads to a high load state. is there. In particular, if a response is not returned from the server for a long time, the client user may send an execution request to the server many times without waiting for a response, which will spur the high load state of the server. . Furthermore, since the server tries to execute an operation whose execution is not expected to be completed in response to a request from the client, there is a problem that the server resources are wasted.

In order to deal with such a problem, for example, in Patent Document 1, a request from a client is held in a request queue, the request residence time, and the request transfer frequency (the number of requests from a client that can be processed per unit time). Are measured, and based on these measurement results and a preset transfer frequency upper limit value, the request is forwarded to the server, or the client is instructed to retransmit, thereby suppressing generation of useless retransmission requests. In addition, it is described that unnecessary generation of an unpleasant waiting time of the user is suppressed.
JP 2006-201970 A

  As described above, the conventional business system has a problem that the client user has to wait for a response from the server until the upper limit of the response time set in the server is exceeded. In particular, when the server is in a high load state, the client user may not only wait for a response from the server, but may not execute the requested operation. Therefore, a means for returning some response to the client at an earlier stage has been desired.

  The present invention has been made in order to solve the above-described problems of the prior art, and shortens the time required to respond to an execution request from a client, and also executes an execution request from a client again. An object of the present invention is to provide a business system and an execution request management method that do not cause a high load state of the server.

To achieve the above object, the business system of the present invention includes a client that transmits an operation execution request,
A queue for temporarily storing the execution request, and when the execution request is output from the queue, a latest queue residence time that is a time during which the execution request has been stored in the queue is calculated. The latest value of the execution time of the operation executed in the past is calculated and held for each operation, and when an operation execution request is received from the client, the operation executed in the past corresponding to the execution request is executed. Calculate the predicted value of the execution time of the operation according to the execution request by adding the time and the latest queue residence time, and when the predicted value exceeds a preset upper limit value of the response time for the client, The client sends a response including an unexecutable error message to the client that sent the operation execution request. And a server connected through the ant and network,
Have

On the other hand, the execution request management method of the present invention is an execution request management method for managing an execution request of an operation transmitted from a client,
When the execution request is output from the queue for temporarily storing the execution request, the latest queue residence time that is the time that the execution request has been stored in the queue is calculated and held,
Calculate and hold the latest value of the execution time of operations executed in the past for each operation,
When an operation execution request is received from the client, an estimated value of the operation execution time according to the execution request is calculated by adding the execution time of the operation executed in the past corresponding to the execution request and the latest queue residence time. And
When the predicted value exceeds a preset upper limit value of the response time for the client, a response including an unexecutable error message is returned to the client that has transmitted the operation execution request.

  In the configuration and method as described above, the execution time from when the operation execution request is received by the server to when the response is returned to the client is predicted, and the predicted value of the execution time is set to a preset upper limit of the response time. Otherwise, the server returns a response containing an unexecutable error message to the client without performing the requested operation. Therefore, the time until the server responds to the execution request from the client is shortened. Therefore, the client user does not need to wait for a response for a long time after transmitting the operation execution request to the server, and is freed from unpleasant response waiting time stress. In addition, even if the client sends an operation execution request to the server again after receiving a response including an error message, the server executes the requested operation if the predicted execution time of the execution request exceeds the upper limit. Send back the response containing the error message to the client again. Therefore, the processing load on the server does not increase due to a re-execution request from the client, and resources are not wasted.

  According to the present invention, the time until the server responds to the execution request from the client is shortened. In addition, a high execution state of the server is not caused by a re-execution request from the client.

  Next, the present invention will be described with reference to the drawings.

  In the business system of the present invention, the server predicts the time from when the operation execution request is received until the response is returned to the client, and the predicted value of the execution time exceeds the preset upper limit of the response time. In this case, a response including an unexecutable error message is returned to the requesting client without executing the requested operation. In this way, a mechanism is provided in which the client user does not wait for a long time response even when the server is in a high load state.

  FIG. 1 is a block diagram showing a configuration example of a business system of the present invention.

  As shown in FIG. 1, the business system of the present invention includes a server 80 that manages the business system, a plurality of clients 10 that are connected to the server 80 via a network, and that send execution requests for various operations to the server 80. It is the structure which has.

  The server 80 includes a queue 30, a server application 40, a queue residence time monitoring unit 50, an operation execution time monitoring unit 60, and an execution time prediction unit 70.

  The queue 30 sequentially stores execution requests (requests) transmitted from the client 10, and outputs the requests to the server application 40 in the order in which the requests are received each time a free process occurs in the server application 40.

  The server application 40 executes the operation according to the request received from the queue 30 and returns a response including the processing result to the requesting client 10.

  The queue residence time monitoring unit 50 monitors the queue residence times of requests transmitted from the plurality of clients 10. The queue residence time monitoring unit 50 includes a queue-in time for each operation that has received a request, and a queue residence time that is the time that the request output from the queue 30 was last stored in the queue 30, that is, the latest queue residence time. Are managed and maintained in tabular form. A management example of the queue-in time is shown in Table 1, and a management example of the latest queue residence time is shown in Table 2.

  The queue residence time monitoring unit 50 notifies the execution time prediction unit 70 of the latest queue residence time managed using Table 2. The queue retention time monitoring unit 50 monitors the number of queued requests (“4” in the example shown in Table 1), and when the number of queued requests remains “0”, the latest information shown in Table 2 is displayed. The queue retention time is reset to “0”.

  The operation execution time monitoring unit 60 manages and holds the execution start time, execution end time, and execution time of operations executed in the past in a table format for each operation. The execution start time, execution end time, and execution time of each operation are updated to the latest values every time the same operation is executed. The operation execution time monitoring unit 60 calculates an execution time for each operation from the execution start time and the execution end time, and notifies the execution time prediction unit 70 of the calculation result. Table 3 shows an example of managing the execution time for each operation.

  The execution time prediction unit 70 obtains the latest queue residence time from the queue residence time monitoring unit 50 and obtains the latest value of the operation execution time corresponding to the currently received request from the operation execution time monitoring unit 60. . Then, these values are added, and an execution prediction time that is a prediction value of the execution time of the currently received request is calculated. In addition, the execution time prediction unit 70 compares the execution prediction time of the currently received request with the preset upper limit value of the response time, and the execution prediction time of the currently received request exceeds the upper limit value. In this case, a response including an unexecutable error message is returned to the client 10 that sent the request.

  The server 80 is realized by, for example, a computer, and as illustrated in FIG. 2, a processing device 1 that executes predetermined processing according to a program, and an input device 2 that inputs commands and information to the processing device 1. And an output device 3 for monitoring the processing result of the processing device 1.

  The processing device 1 includes a CPU 11, a main storage device 12 that temporarily stores information necessary for the processing of the CPU 11 and also used as the queue 30, and a program for causing the CPU 11 to execute processing as a server 80 described later. A recorded recording medium 13, a data storage device 14 for storing information such as the queue-in time, the latest queue residence time, and the execution time of each operation shown in Tables 1 to 3, the main storage device 12, The memory control interface unit 15 that controls data transfer with the recording medium 13 and the data storage device 14, the I / O interface unit 16 that is an interface device with the input device 2 and the output device 3, and the client 10 transmit and receive information. And a communication control device 17 that is an interface with the network for this purpose. The CPU 11, the memory control interface unit 15, the I / O interface unit 16, and the communication control device 17 are connected via a bus 18.

  The processing apparatus 1 includes a queue residence time monitoring unit 50, an operation execution time monitoring unit 60, and the like according to a program recorded on the recording medium 13. The functions of the execution time prediction unit 70 and the server application 40 are realized. The recording medium 13 may be a magnetic disk, a semiconductor memory, an optical disk, or other recording medium.

  The server 80 includes the queue 30, the server application 40, the queue residence time monitoring unit 50, the operation execution time monitoring unit 60, and the execution time prediction unit 70 shown in FIG. 1, and can manage the business system. For example, it need not be limited to computers. For example, in addition to the communication control device 17 that is an interface with a network, the server 80 includes a semiconductor integrated circuit device such as an LSI (Large Scale Integration) or a DSP (Digital Signal Processor) configured by a logic circuit, a memory, and the like. The queue residence time monitoring unit 50, the operation execution time monitoring unit 60. The structure which implement | achieves the function of the execution time estimation part 70 and the server application 40 may be sufficient.

  The client 10 can also be realized by the same configuration as the server 80 shown in FIG. 2 except that the program stored in the recording medium 13 and the data stored in the data storage device 14 are different.

  Next, the operation of the business system of the present invention will be described with reference to the drawings.

  FIG. 3 is a flowchart showing a processing procedure of the execution time prediction unit shown in FIG. 4 is a flowchart showing a processing procedure of the queue residence time monitoring unit shown in FIG. 1, and FIG. 5 is a flowchart showing a processing procedure of the operation execution time monitoring unit shown in FIG.

  In the server 80, it is assumed that the upper limit value of the response time for the request from the client 10 is stored in a predetermined storage device (for example, the data storage device 14 shown in FIG. 2) in advance by an operator's operation.

  As shown in FIG. 3, the server 80 first reads the upper limit value of the response time from the predetermined storage device by the execution time prediction unit 70 (step A1).

  When the request transmitted from the client 10 is received (step A2), the execution time prediction unit 70 acquires the latest queue residence time from the queue residence time monitoring unit 50 (step A3).

  As shown in FIG. 4, when the queue residence time monitoring unit 50 receives a request from the client 10 and stores the request in the queue 30 (step B1), the queue storage time (queue-in time) is set to the above-described time. Record in Table 1 (step B2). When a request is output from the queue 30 (step B3), the queue residence time is calculated from the output time (queue-out time) of the request and the queue-in time of the request, and the latest stored in Table 2 above. The queue residence time is updated (step B4).

  Next, the queue residence time monitoring unit 50 deletes the information of the queued request from the table 1 (step B5), and whether or not the queue residence number of the request managed in the table 1 is “0”. Is determined (step B6). If the queue retention number is not “0”, the process returns to step B1 to wait for the next request to be queued. If the queue retention number is “0”, the latest queue retention time stored in Table 2 is reset to “0” (step B7), and the process returns to step B1 to return to the next request. Wait for the queue. Note that the queue residence time is not limited to the method according to the procedure shown in FIG. 4, but can be calculated from the product of the average execution time and the number of queue residences by calculating the average execution time of all operations.

  As shown in FIG. 3, when the execution time prediction unit 70 acquires the latest queue residence time from the queue residence time monitoring unit 50 in step A3, the execution time prediction unit 70 obtains the latest execution time of the operation from the operation execution time monitoring unit 60. (Step A4).

  As shown in FIG. 5, when the operation is started (step C1), the operation execution time monitoring unit 60 updates the execution start time of the corresponding operation stored in Table 3 (step C2).

  Next, the operation execution time monitoring unit 60 determines whether or not the operation has ended normally (step C3). If the operation has ended normally, the execution end time of the corresponding operation stored in Table 3 above is determined. Is updated (step C4). Further, the execution time of the corresponding operation is calculated from the execution start time and execution end time of the updated operation, and the execution time of the corresponding operation stored in Table 3 is updated (step C5). When the update of the execution time is completed, the process returns to step C1 and waits for the start of operation execution. When the operation is not normally completed in the process of Step C3, the operation execution time monitoring unit 60 clears the operation execution start time of the corresponding operation stored in Table 3 (Step C6), and performs the process of Step C1. Return and wait for the operation to start.

  As shown in FIG. 3, when the execution time prediction unit 70 acquires the latest execution time of the corresponding operation from the operation execution time monitoring unit 60 in step A4, the execution time prediction unit 70 acquires the queue residence time and the operation acquired from the queue residence time monitoring unit 50. The operation execution time acquired from the execution time monitoring unit 60 is added to calculate the operation execution prediction time for the currently received request (step A5).

  Next, the execution time prediction unit 70 compares the calculated execution prediction time with the upper limit value of the response time read in step A1 (step A6). If the execution prediction time does not exceed the upper limit value, the execution time prediction unit 70 The request is stored in the queue 30. The operation according to the request is executed after the elapse of a required queue residence time (step A7). If the predicted execution time exceeds the upper limit value, a response including an unexecutable error message is returned to the client 10 that sent the request (step A8).

  The server 80 may return the queue residence time calculated by the queue residence time monitoring unit 50 or the execution time for each operation calculated by the operation execution time monitoring unit 60 in response to a request from the client 10. In this case, since the user of the client 10 can check the processing load of the server 80 before sending the operation execution request to the server 80, the time required for the operation that can be executed quickly, the end of execution of the desired operation, etc. I can know.

  According to the business system of the present invention, the execution time from when the operation execution request is received by the server 80 until the response is returned to the client 10 is predicted, and the predicted value of the execution time is set to a preset response time. When the upper limit value is exceeded, the server 80 returns a response including an unexecutable error message to the client 10 without executing the requested operation. Therefore, the time until the server 80 responds to the execution request from the client 10 is shortened. Therefore, the user of the client 10 does not have to wait for a response for a long time after transmitting an operation execution request to the server 80, and is freed from stress of unpleasant response waiting time. Further, even if the client 10 transmits an operation execution request again to the server 80 after receiving a response including an error message, if the predicted value of the execution time of the execution request exceeds the upper limit value, the server 80 requests the requested operation. The response including the error message is sent back to the client 10 again without executing. Therefore, the processing load on the server 80 does not increase due to a re-execution request from the client 10, and resources are not wasted.

It is a block diagram which shows the example of 1 structure of the business system of this invention. It is a block diagram which shows the implementation example of the server shown in FIG. It is a flowchart which shows the process sequence of the execution time estimation part shown in FIG. It is a flowchart which shows the process sequence of the queue residence time monitoring part shown in FIG. It is a flowchart which shows the process sequence of the operation execution time monitoring part shown in FIG.

Explanation of symbols

1 Processing Device 2 Input Device 3 Output Device 11 CPU
12 main storage device 13 recording medium 14 data storage device 15 memory control interface unit 16 I / O interface unit 17 communication control unit 18 bus 10 client 30 queue 40 server application 50 queue residence time monitoring unit 60 operation execution time monitoring unit 70 execution time Prediction unit 80 server

Claims (4)

A client sending an operation execution request;
A queue for temporarily storing the execution request, and when the execution request is output from the queue, a latest queue residence time that is a time during which the execution request has been stored in the queue is calculated. The latest value of the execution time of the operation executed in the past is calculated and held for each operation, and when an operation execution request is received from the client, the operation executed in the past corresponding to the execution request is executed. Calculate the predicted value of the execution time of the operation according to the execution request by adding the time and the latest queue residence time, and when the predicted value exceeds a preset upper limit value of the response time for the client, The client sends a response including an unexecutable error message to the client that sent the operation execution request. And a server connected through the ant and network,
Having a business system. A server that receives an operation execution request sent from a client;
A queue for temporarily storing the execution request;
When the execution request is output from the queue, the latest queue residence time, which is the time that the execution request has been stored in the queue, is calculated and held, and the latest value of the execution time of operations executed in the past Is calculated and held for each operation, and when an operation execution request is received from the client, the execution time of the operation executed in the past corresponding to the execution request and the latest queue residence time are added to perform the execution. A predicted value of the execution time of the operation according to the request is calculated, and when the predicted value exceeds a preset upper limit value of the response time for the client, an error message indicating that the operation is not executed is sent to the client that has transmitted the operation execution request. A processing device for returning a response including:
Server with. An execution request management method for managing an execution request of an operation transmitted from a client,
When the execution request is output from the queue for temporarily storing the execution request, the latest queue residence time that is the time that the execution request has been stored in the queue is calculated and held,
Calculate and hold the latest value of the execution time of operations executed in the past for each operation,
When an operation execution request is received from the client, an estimated value of the operation execution time according to the execution request is calculated by adding the execution time of the operation executed in the past corresponding to the execution request and the latest queue residence time. And
An execution request management method for returning a response including an unexecutable error message to a client that has transmitted the operation execution request when the predicted value exceeds a preset upper limit value of the response time for the client. A program for causing a computer to receive an operation execution request transmitted from a client,
When the execution request is output from the queue for temporarily storing the execution request, the latest queue residence time that is the time that the execution request has been stored in the queue is calculated and held,
Calculate and hold the latest value of the execution time of operations executed in the past for each operation,
When an operation execution request is received from the client, an estimated value of the operation execution time according to the execution request is calculated by adding the execution time of the operation executed in the past corresponding to the execution request and the latest queue residence time. And
When the predicted value exceeds a preset upper limit value of the response time for the client, the computer is caused to execute a process of returning a response including an unexecutable error message to the client that has transmitted the operation execution request. program.

Images