JP2008077266A - Service control unit, distributed service control system, service control method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a service control unit which achieves processing which satisfies a service quality condition as a whole system in a system which makes a plurality of servers cooperate. <P>SOLUTION: A message reception controller 110 receives a process request message and stores it in a message queue 130. A message weight calculation section 121 calculates a process volume to a processing request message in the server. A thread 142 takes out a process request message from a message queue 130 and makes it processed by the server. A control rule determination section 150 decides the size of the message queue 130, and the number of parallel processing of the thread 142 so as to satisfy the service quality condition based on a retaining period of a message queue of the process request message, a processing volume in the server calculated in the message weight calculation section 121 and the service quality requirement stored in a service quality performance storage section 160. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a service control apparatus, a distributed service control system, a service control method, and a program. More specifically, the present invention relates to a service control apparatus and method for controlling a processing request message for a server that performs processing by transmitting and receiving processing request messages. Further, the present invention relates to a distributed service control system in which a program and such a service control apparatus are linked.

  With the spread of structured documents that can be easily processed by computers such as XML, the exchange of structured document messages has led to the spread of business systems based on server collaboration along a series of flows such as order receiving servers, inventory management servers, and delivery servers. I am doing. Conventionally, a method of performing batch processing in which each server receives a plurality of requests received from other servers in a night time zone with sufficient server resources has been widely used. As the immediacy of shortening the overall processing time is required, each server is replacing the batch processing with the sequential processing of individual requests.

  In business systems, especially in the case of collaboration over the Internet, various communications such as e-mail are used at the same time, so the congestion status of the communication network varies widely, and transmission of processing request messages from the partner server The pattern may not be uniform, and a large amount of processing request messages may arrive at the same time. Therefore, in order to maintain the service quality even in such a situation, we assumed a fluctuation in the number and amount of processing requests at the system design stage and introduced a server with sufficient performance in advance to ensure the service quality. . In operation, service quality is ensured by dynamically allocating computer resources, priority processing of specific processing requests, and the like.

Based on the above problems, an example of a technique for optimizing the execution of a conventional distributed business process is described in Patent Document 1. In Patent Document 1, in business process processing, process definitions and data are carried between servers, and the introduction of a computer node that controls the entire business process independent of the server is unnecessary. An example of a technique for optimizing the execution of a workflow executed on a plurality of computer nodes is described in Patent Document 2. In Patent Document 2, the workflow server centrally manages the configuration and performance of the computer and the overall workflow definition in advance, and the execution distribution is determined while monitoring the execution state by the workflow server.
JP 2004-258823 A Japanese Patent Laid-Open No. 2002-259639

  The first problem of the prior art is that, in a system in which a plurality of servers operate in cooperation by exchanging messages, the server downstream of the cooperation flow receives the number of processing request messages exceeding the server performance and performance degradation occurs. In this case, the service quality related to the processing time in the entire system linked with the server cannot be maintained. This is because the processing request message cannot be accepted by this server due to the performance degradation that occurred in the downstream server, and as a result, the processing request process for the downstream server stays in the upstream server, and the performance degradation of the downstream server This is because it is difficult to estimate the processing performance of a plurality of servers as a whole.

  The second problem is that the system requires extra computer resources assuming a sufficiently large number of processing request messages compared to the number of processing request messages arriving per unit time. The reason is that processing request messages through a network are likely to have a difference in communication quality due to coexistence with various communications such as e-mails, so that processing requests do not always arrive uniformly. In particular, since the arrival of request processing messages may be intermittent and have burstiness that concentrates at a time, a system that sequentially processes the processing request messages as in the above-described system requires sufficient computer resources.

  The present invention solves the above-described problems of the prior art, and in a system in which a plurality of servers operate in cooperation by exchanging messages, a service control apparatus, method, program, and program capable of executing processing satisfying service quality as a whole system, and An object is to provide a distributed service control system.

  Another object of the present invention is to provide a service control device, a method, a program, and a service control device that prevent the server from being disabled with a minimum amount of computer resources even when a large amount of processing request messages are temporarily transmitted to the server. It is to provide a distributed service control system.

  Another object of the present invention is to provide a service control device, a method, a program, and a distributed service control system that can quickly respond to fluctuations in processing request messages.

  In order to achieve the above object, the service control device of the present invention uses a message reception control unit that performs reception control of a processing request message and stores the received processing request message in a message queue, and the structure of the processing request message. A message weight storage unit that stores a message weight rule that defines a rule for calculating the processing time at the server, and based on the processing request message and the message weight rule, for the processing request message A message weight calculation unit for calculating a processing time at the server, and a thread group control unit for controlling a thread group having a plurality of threads for transmitting the processing request message to the server and executing the processing. And assigning the thread to the parallel processing of the thread Referring to a thread group control unit that controls the upper limit value and a quality requirement information storage unit that stores quality requirement information related to a quality requirement to be satisfied by the server, a residence time of the processing request message in the message queue, A control rule determining unit that controls the size of the message queue and the upper limit value of the parallel processing number of threads in the thread group control unit based on the calculated processing time in the server and the quality requirement information; It is characterized by providing.

  The service control method of the present invention is a service method for controlling a processing request message for a server using a service control device, wherein the service control device receives the processing request message and sends the received processing request message to the message. A message receiving step for storing in a queue; and a message weight storage unit for storing a message weight rule in which the service control device defines a rule for calculating a processing time in the server using the structure of the processing request message. A message weight calculating step of calculating a processing time at the server for the processing request message based on the processing request message and the message weight rule; and the service control device sends the processing request message to the server. Multiple threads that send and execute processing A thread assignment step for controlling a thread group and assigning the thread to the processing request message; and a quality requirement information storage unit in which the service control apparatus stores quality requirement information related to a quality requirement to be satisfied by the server. The size of the message queue based on the retention time in the message queue of the processing request message, the calculated processing time in the server, and the quality requirement information, and the thread group control unit And a control rule determining step for controlling an upper limit value of the number of parallel processes of the thread.

  A program according to the present invention is a program for causing a computer to execute processing for controlling a processing request message for a server, wherein the computer receives a processing request message and stores the received processing request message in a message queue. A message weight storage unit for storing a message weight rule for defining a rule for calculating a processing time in a server using a structure of the reception step and the processing request message, the processing request message, and the message Based on the weight rule, a message weight calculation step for calculating a processing time at the server for the processing request message, and a thread group having a plurality of threads for transmitting the processing request message to the server and executing the processing, In response to the processing request message A thread allocation step for allocating the thread, a quality requirement information storage unit that stores quality requirement information relating to a quality requirement to be satisfied by the server, a residence time of the processing request message in the message queue, and the calculation A control rule determining step for controlling the size of the message queue and the upper limit value of the number of parallel processes of threads in the thread group control unit based on the processing time in the server and the quality requirement information It is characterized by making it.

  In the service control device, method, and program of the present invention, whether or not a predetermined service quality requirement is satisfied based on the processing amount at the server for the received processing request message and the message queue residence time of the processing request message. To determine the size of the message queue and the upper limit value of the number of parallel processes in the server. By controlling the size of the message queue and the upper limit value of the number of parallel processes so that the service quality requirement is satisfied at the server, for example, a large number of processing request messages are transmitted, and the processing for the processing request message is processed by the service quality requirement If it is predicted that the message will not be satisfied, the server load increases by reducing the message queue size and delaying reception to reduce the number of input processing request messages and decreasing the upper limit of the number of parallel processing To achieve processing that satisfies service quality requirements. Also, when there is enough processing capacity, increase the number of processing request messages by increasing the size of the message queue and increase the upper limit of the number of parallel processings to increase the number of processing request messages processed by the server . By performing such control, it is possible to realize processing that maximizes the processing capability of the server while satisfying service quality.

  In the service control apparatus of the present invention, the control rule determination unit performs processing on the processing request message based on the retention time of the message, the calculated processing time at the server, and the quality requirement. If it is determined whether or not the above condition is satisfied, and if it is not satisfied, it is possible to adopt a configuration in which at least one of a decrease in the size of the message queue and a decrease in the upper limit value of the number of parallel processes of the thread are employed. If the size of the message queue is reduced, the number of input processing request messages can be suppressed by delaying processing request messages newly received by the message reception control unit. Further, when the upper limit value of the number of parallel processes of threads is reduced, the number of parallel processes in the server can be reduced, the server load can be reduced, and the processing time for each processing request message can be shortened. As a result, it is possible to reduce the time from the reception of the processing request message to the completion of the processing at the server, and it is possible to realize processing that satisfies the service quality requirements.

  In the service control device of the present invention, when the control rule determination unit determines that the processing for the processing request message does not satisfy the quality requirement, the upper limit value of the parallel processing number of the thread is reduced, and the parallel processing number of the thread is decreased. The processing request message in a state where the upper limit value of the number of parallel processes of the thread is reduced based on the processing time at the server after the decrease of the upper limit value of the message, the residence time of the message, and the quality requirement If it is determined whether or not the processing satisfies the quality requirement and it is determined that the processing does not satisfy the requirement, a configuration for reducing the size of the message queue can be adopted.

  In the service control apparatus of the present invention, the control rule determination unit performs processing on the processing request message based on the retention time of the message, the calculated processing time at the server, and the quality requirement. It is possible to adopt a configuration in which it is determined whether or not the condition is satisfied, and if it is determined that the condition is satisfied, at least one of an increase in the size of the message queue and an increase in the upper limit value of the parallel processing number of the thread can be employed. When processing for processing request messages satisfies the quality requirements, the message queue size is increased to increase the number of processing request messages input, the number of parallel processing of threads is increased, and the number of processing request messages processed simultaneously by the server is increased. increase. By increasing the number of processing request messages input and the number of simultaneous executions of processing in the server within the range that satisfies the quality requirements, the server performance can be utilized to the maximum.

  In the service control device of the present invention, when the control rule determination unit determines that the processing for the processing request message satisfies the quality requirement, the control rule determination unit checks whether the size of the message queue is smaller than a predetermined initial value, When the size is smaller than the initial value, the size of the message queue is increased. When the size of the message queue is not smaller than the initial value, the upper limit value of the number of parallel processes of the thread can be increased. When the quality requirement is satisfied and the current state is a state in which the message queue size is reduced and the number of processing request messages input is suppressed, the reception delay is increased by increasing the message queue size. Is released and the suppression of the input processing request message is released. When the input processing request message is not suppressed, the upper limit value of the parallel processing number of threads is increased to increase the parallel processing number in the server. By controlling in this way, the server performance can be utilized to the maximum.

  In the service control apparatus of the present invention, the thread measures an actual measurement value of a processing time from the start of processing to the end of processing for the processing request message in the server, and the control rule determination unit includes the calculation The message weight rule is set so that a difference between the calculated processing time at the server and the measured value of the processing time is reduced based on the processing time at the calculated server and the measured value of the processing time. A configuration for correcting can be adopted. In this case, the accuracy of the processing time at the server calculated based on the received processing request message can be improved.

  In the service control apparatus of the present invention, when the message reception control unit receives the processing request message, the message reception control unit determines whether or not there is a vacancy in the message queue. It is possible to adopt a configuration that delays reception processing of a processing request message. In this case, when the message reception control unit stores the processing request message in the message queue, the message reception control unit can transmit a response indicating the completion of reception to the transmission source of the processing request message. In this case, the transmission source server cannot start processing for the next processing request message because the message transmission is not completed until the transmitted processing request message is stored in the message queue. Thereby, reception delay according to the size of the message queue can be performed, and the number of processing request messages transmitted from the transmission source server can be suppressed according to the size of the message queue.

  In the service control method and the program according to the present invention, in the control rule determination step, processing for the processing request message is performed based on the retention time of the message, the calculated processing time at the server, and the quality requirement. If it is determined whether or not the quality requirement is satisfied, and it is determined that the quality requirement is not satisfied, a configuration in which at least one of the size reduction of the message queue and the upper limit value of the parallel processing number of the thread can be adopted.

  In the service control method and the program of the present invention, in the control rule determination step, when it is determined that the processing for the processing request message does not satisfy the quality requirement, the upper limit value of the parallel processing number of the thread is reduced, and the parallel processing of the thread is performed. The processing request message in a state where the upper limit value of the parallel processing number of the thread is reduced based on the processing time at the server after the upper limit value of the number is decreased, the residence time of the message, and the quality requirement If it is determined whether or not the processing for the above satisfies the quality requirement and it is determined that the processing does not satisfy the quality requirement, a configuration for reducing the size of the message queue can be adopted.

  In the service control method and the program according to the present invention, in the control rule determination step, processing for the processing request message is performed based on the retention time of the message, the calculated processing time at the server, and the quality requirement. If it is determined whether or not the quality requirement is satisfied and it is determined that the quality requirement is satisfied, a configuration in which at least one of an increase in the size of the message queue and an increase in the upper limit value of the parallel processing number of the thread can be employed.

  In the service control method and program of the present invention, in the control rule determination step, when it is determined that the processing for the processing request message satisfies the quality requirement, it is determined whether or not the size of the message queue is smaller than a predetermined initial value. As a result, when the size is smaller than the initial value, the size of the message queue is increased, and when the size of the message queue is not smaller than the initial value, the upper limit value of the parallel processing number of the thread can be increased.

  In the service control method of the present invention, the service control device measures an actual measurement value of a processing time from the start of processing to the end of processing for the processing request message in the server in the thread; The service control device reduces a difference between the calculated processing time at the server and the measured value of the processing time based on the calculated processing time at the server and the measured value of the processing time. And a step of correcting the message weight rule. The program of the present invention includes a step of measuring an actual measurement value of a processing time from the start of processing to the end of processing in the thread for the processing request message in the server, and the calculated server Correcting the message weight rule so that a difference between the calculated processing time at the server and the measured value of the processing time is reduced based on the processing time at the time and the measured value of the processing time. It is possible to adopt a configuration that further executes

  In the service control method and program of the present invention, in the message reception step, when the processing request message is received, it is determined whether or not there is an empty space in the message queue. A configuration for delaying the reception processing of the processing request message can be employed.

  The service control method and program of the present invention can employ a configuration in which, in the message reception step, when a processing request message is stored in the message queue, a response indicating reception completion is transmitted to the transmission source of the processing request message.

  The distributed service control system of the present invention is a distributed service control system that delivers a processing request message between a plurality of servers and operates in cooperation with each other, and each of the servers at the input stage of the processing request message It has a control device.

  In the distributed service control system of the present invention, the service control device controls the upper limit of the reception delay and the parallel processing number of threads by changing the size of the message queue so that the service quality requirement is satisfied at each server. For this reason, the process which satisfy | filled service quality requirements is realizable as the whole system. At this time, in the cooperation flow between the servers, the reception delay and the number of parallel processes influence each other between the distributed service control devices of adjacent servers, and the appropriate execution distribution is automatically reached. Thereby, as a whole system, processing that utilizes the capabilities of each server can be realized while satisfying the service quality requirements.

  In the service control device, method, and program of the present invention, whether or not a predetermined service quality requirement is satisfied based on the processing amount at the server for the received processing request message and the message queue residence time of the processing request message. To determine the size of the message queue and the upper limit value of the number of parallel processes in the server. When it is expected that the service quality requirement will not be satisfied, the number of processing request messages is suppressed, and when it is expected to be satisfied, the number of processing request messages is increased to satisfy the service quality while processing the server. Processing that maximizes its capabilities can be realized.

  In the distributed service control system according to the present invention, the service control apparatus according to the present invention realizes processing satisfying the service quality in each server, thereby realizing processing satisfying the service quality as a whole system. Also, in the cooperation flow between servers, the reception delay and the number of parallel processes affect each other between the distributed service control devices of adjacent servers, and automatically reach an appropriate execution distribution, so that the entire system, Processing that utilizes the capabilities of each server can be realized while satisfying service quality requirements.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows the configuration of a distributed service control apparatus according to an embodiment of the present invention. The distributed service control apparatus 100 includes a message reception control unit 110, a message state management unit 120, a message queue 130, a processing request management unit 140, a control rule determination unit 150, and a service quality performance storage unit 160. The distributed service control device 100 is arranged as a relay device in the preceding stage of a server (message processing unit) that executes a service in a system in which a plurality of servers operate in cooperation by exchanging messages, and transmits and receives processing request messages. Relay. The distributed service control apparatus 100 automatically controls message communication and server processing so that the server maintains quality related to service performance. The distributed service control apparatus 100 may be constructed in the same computer system as the server, or may be constructed using a computer different from the server.

  The message reception control unit 110 controls reception of processing request messages from other servers. The message state management unit 120 includes a message weight calculation unit 121 and a message weight storage unit 122. The message state management unit 120 calculates a server processing amount for each processing request message. The message weight storage unit 122 stores a message weight rule for calculating the processing amount at the server based on the structure of the processing request message. The message weight calculation unit 121 calculates a processing amount for the processing request message, that is, a predicted value of processing time at the server, based on the processing request message and the message weight rule. The message weight calculation unit 121 stores the calculated processing amount at the server in the message weight storage unit 122 as message weight information.

  The message queue 130 temporarily stores processing request messages to the server. The size of the message queue 130 is set to be variable. The processing request management unit 140 transmits a processing request message to the server, and causes the server to execute processing on the processing request message. The processing request management unit 140 includes a plurality of executable threads 142 and a thread group control unit 141. The thread group control unit 141 manages the thread 142 and assigns processing to the thread 142. The thread 142 transmits a processing request message to the server to execute processing. The number of threads 142 is the number of parallel processes on the server.

  The service quality performance storage unit 160 stores information regarding service quality requirements to be satisfied as server performance requirements. The control rule determination unit 150 controls the size of the message queue 130 and the number of parallel processing of the processing request message in the processing request management unit 140 based on the processing time and the service quality requirement for the processing request message. Further, the control rule determination unit 150 uses the message weight information (server processing amount) calculated using the message weight rule and the measurement result of the processing time in the actual server to correct the message weight rule ( Learning).

  The message reception control unit 110 receives a processing request message. The received processing request message is stored in the message queue 130 after the message weight calculation unit 121 calculates the amount of processing for the processing request message. At that time, the message reception control unit 110 checks whether or not the message queue 130 has a vacancy, and if there is no vacancy, the message reception control unit 110 delays reception of the processing request message until the vacancy occurs. When the message reception control unit 110 stores the processing request message in the message queue 130, the message reception control unit 110 transmits a response indicating completion of message reception to the message transmission source.

  The thread group control unit 141 takes out one processing request message from the message queue 130 and assigns a process for transferring the processing request message to the server and monitoring the thread 142. The message reception control unit 110 measures the time (queue retention time) from when the processing request message is stored in the message queue 130 until it is taken out after reception of the processing request message is completed. The thread 142 transmits a processing request message to the server to perform processing. At that time, the thread 142 measures the processing time in the server.

  The control rule determination unit 150 periodically performs the processing amount for the processing request message calculated by the message weight calculation unit 121, the queue residence time, and the service quality requirement information stored in the service quality performance storage unit 160. Based on the above, the size of the message queue 130 and the upper limit value of the parallel processing number of the threads 142 in the processing request management unit 140 are controlled so that the processing time at the server satisfies the service quality requirement. For example, when the server load is high and the processing time does not satisfy the service quality performance requirement, the upper limit value of the number of parallel processes of the thread 142 is decreased to reduce the number of parallel processes in the server, thereby reducing the server load. Further, the size of the message queue 130 is reduced, the reception of the processing request message is delayed, and the reception amount of the processing request message is suppressed.

  The control rule determination unit 150 compares the predicted value of the processing time at the server calculated by the message weight calculation unit 121 with the measured value of the processing time at the server measured by the thread 142, and calculates the message weight. The message weight rule used for the correction is corrected. In this correction, the message weight rule is corrected so that the difference between the predicted value of the server processing time and the actual measurement value is reduced. By this correction, an error between the predicted value of the server processing time calculated by the message weight calculation unit 121 and the actual measurement value can be reduced.

  FIG. 2 shows an operation procedure of the reception process. The message reception control unit 110 starts receiving the processing request message (step S111), and determines whether or not the message queue 130 is empty (step S112). In step S112, for example, when the size of the message queue 130 is set to “20”, it is determined whether or not 20 entries of the message queue 130 are already used. If there is no space, the reception delay is performed until the space is uttered (step S121).

  If there is an empty space in the message queue 130, that is, if the processing request message can be stored in the message queue 130, the message reception control unit 110 transmits a reception completion message to the transmission source of the processing request message ( Step S113). In step S113, for example, when a processing request message is transmitted by an HTTP protocol POST request, a value indicating reception completion is returned. Thereafter, the message reception control unit 110 sends the received request processing message to the message weight calculation unit 121. The message weight calculation unit 121 analyzes the message structure of the received request processing message, and processes the processing request message on the server based on the analyzed message structure and the message weight rule stored in the message weight storage unit 122. The amount is calculated (step S114).

  FIG. 3 shows a specific example of a general processing request message envelope. The processing request message 180 has the destination of the processing request message in the header of the envelope. It also has a structured document in the envelope body and optionally includes binary data such as images in the attached data. If the structured document is data corresponding to, for example, a device purchase order, tags are repeated by the number of parts. The amount of processing (processing time) at the server depends on the number of tags, presence / absence of attached data, size, and the like. The part related to the server processing amount is stored as a message weight rule and used for prediction of the server processing time.

  FIG. 4 shows the structure of the message weight rule. The message weight rule 190 manages the rules (191 to 19n) for each message type. In each rule, a rule for calculating the amount of processing corresponding to the message type is described. For example, for a certain message type, when the number of tags is 0 to 5, the number of tags × time A is used as the server processing amount, and when the number of tags exceeds 5, the number of tags × time B is calculated as the server processing amount. A rule to do so is described. In addition, when “attached file exists”, a rule that increases the processing amount by the processing amount A is described.

  Returning to FIG. 2, in step S <b> 114, first, the message type of the processing request message is determined, and which of the message weight rules 190 is to be used is determined. Thereafter, the server processing amount is calculated based on the message structure such as the number of tags included in the processing request message, the presence / absence of an attached file, and the rules corresponding to the message type of the processing request message. The message weight rule can be set from the outside when the processing content of the service is known in advance. When the processing contents of the service are not known, the processing at the server may be actually performed and derived based on the actually measured value of the processing time.

  When calculating the server throughput, the message weight calculation unit 121 stores the calculation result in the message weight storage unit 122 as message weight information (step S115). The structure of message weight information is shown in FIG. The message weight information 200 includes a message ID of the processing request message and message weights 201 to 20n indicating the processing amount at the server. After storing the message weight information, the message weight calculation unit 121 transfers the processing request message to the message queue 130 and passes the processing request message to the message queue 130 (step S116). Note that either the storage of the processing request message in the message queue or the calculation of the message weight may be performed first.

  FIG. 6 shows a processing procedure for thread assignment. The thread assignment is executed independently of the reception process shown in FIG. The thread group control unit 141 has a plurality of executable threads 142 set in advance before the start of system operation. After starting the operation of the system, the thread group control unit 141 determines whether there is an empty thread in the thread group managed by the thread group control unit 141 (step S211). If there is no free thread, step S211 is repeatedly executed to continue monitoring the free thread.

  If the thread group control unit 141 determines that there is an empty thread in step S211, the thread group control unit 141 checks whether a message is retained in the message queue 130 (step S212). If the message remains, the activation notification is notified to the thread 142 (step S213). If the message is not retained, the process returns to step S211 to continue monitoring the free thread.

  FIG. 7 shows an operation procedure of the thread 142. When the thread 142 receives the activation notification from the thread group control unit 141 (step S311), the thread 142 acquires one processing request message from the message queue 130 (step S312). Thereafter, the service start time is measured (step S313), and the processing request message is transferred to the server that executes the service (step S314). After transferring the processing request message, the thread 142 determines whether or not the processing at the server is finished, and waits until the processing (service) at the server is finished (step S315).

  If the thread 142 determines that the process is completed at the server, the thread 142 measures the end time of the service (step S316). The thread 142 calculates the service execution time from the difference between the service start time measured in step S313 and the service end time measured in step S316. When the thread 142 calculates the service execution time, the thread 142 notifies the control rule determination unit 150 of the service execution time (step S317). Thereafter, the completion of the service is notified to the thread group control unit 141 (step S318). Receiving this notification, the thread group control unit 141 manages the thread 142 that has been processed as an empty thread.

  FIG. 8 shows an operation procedure of message weight rule correction in the control rule determination unit 150. The control rule determination unit 150 acquires the service execution time, that is, the processing time at the actual server for the processing request message, from the thread 142 (step S351). Further, the control rule determination unit 150 acquires the message weight information 200 (FIG. 5) and the message weight rule 190 (FIG. 4) from the message weight storage unit 122 (step S352).

  The control rule determination unit 150 corrects the message weight rule based on the difference between the actual service execution time for the processing request message and the estimated processing time at the server calculated using the message weight rule (step S353). In this correction, when the message weight is calculated using the message structure of the processing request message, the message weight rule is corrected so that the message weight approaches the actual service execution time. By repeatedly correcting the message weight rule, the accuracy of the predicted value of the processing time at the server is improved.

  FIG. 9 shows an operation procedure for deriving a control rule in the control rule determination unit 150. This control rule derivation operation is periodically repeated, and is performed independently of the message weight rule correction operation. The control rule determination unit 150 acquires the message weight information 200 (FIG. 5) from the message weight storage unit 122 (step S411), and acquires the average message residence time of the message queue 130 from the message reception control unit 110 (step S412). ).

  Subsequently, the control rule determination unit 150 acquires service quality requirement information set in advance as a service level requirement from the service quality performance storage unit 160 (step S413). FIG. 10 shows the structure of service quality requirement information. The service quality requirement information 170 has requirements 171 to 17n as service quality requirements to be satisfied for each service. The control rule determination unit 150 determines whether or not the service quality requirement can be satisfied for the acquired message weight information and message retention time (step S414).

  FIG. 11 shows the processing execution time for the processing request message. The processing execution time can be roughly divided into four times: time required for reception processing, time required for processing waiting (queue retention time), time required for service execution, and time required for transmission processing to the next service (next server). There are elements. The message weight acquired in step S411 corresponds to a predicted value of the service execution time from the start of service execution to the completion of message transmission to the next server. The service execution time (predicted value) is obtained by adding the queue residence time (processing waiting time) to the service execution time (predicted value). In step S414, it is determined whether or not the predicted value of the apparent service execution time satisfies the service quality requirement.

  If the control rule determination unit 150 determines that the service quality requirement cannot be satisfied in step S414, the control rule determination unit 150 notifies the thread group control unit 141 of a decrease in the number of threads that can be executed in parallel (step S421). Upon receiving this notification, the thread group control unit 141 decreases the upper limit value of the executable thread 142 in accordance with the notification content. By reducing the upper limit value of the thread 142, the number of parallel processes in the server is reduced, and the server load is reduced, so that the service execution time can be expected to be shortened. In step S421, for example, based on the difference between the message weight and the service quality requirement, the decrease in the number of threads is increased as the difference increases.

  Subsequently, the control rule determination unit 150 determines whether or not the service quality requirement can be satisfied after the decrease in the number of threads (step S422). In step S422, for example, it is estimated in advance how much the service execution time is shortened by reducing the number of threads, and a value obtained by subtracting the shortened service execution time from the message weight acquired in step S411 and the message queue residence time. It is determined whether or not the sum can satisfy the service quality requirement.

  If the control rule determination unit 150 determines in step S422 that the service quality requirement cannot be satisfied even after the number of threads is decreased, the control rule determination unit 150 determines to perform reception delay and decreases the size of the message queue 130 (step S423). In step S423, for example, the size of the message queue 130 is set to be smaller as the difference between the predicted service execution time and the service quality requirement when it is determined in step S422 that the service quality requirement can be satisfied. When it is determined in step S422 that the service quality requirement can be satisfied, no reception delay is performed.

  In step S423, for example, the size of the message queue 130 is set to a size smaller than a predetermined initial value. As a result, after the size reduction, the message reception control unit 110 delays reception by an amount corresponding to the size reduction (step S121 in FIG. 2), and accepts the processing request message per unit time. The number decreases. In this way, by suppressing the number of received processing request messages, the time for which the processing request message stays in the message queue 130 is reduced, and the apparent service time (FIG. 11) can be shortened.

  If it is determined in step S414 that the service quality requirement can be satisfied, the control rule determination unit 150 checks whether or not the reception delay is performed in the message reception control unit 110 (step S415). In step S415, for example, if the size of the message queue 130 is smaller than a predetermined initial value, it is determined that reception delay is performed. If reception delay is performed, the size of the message queue 130 is increased by a predetermined number, and the reception delay is canceled (step S431). Thereafter, the message reception control unit 110 receives the number of processing request messages corresponding to the increased size of the message queue 130.

  When the control rule determination unit 150 determines in step S415 that reception delay is not performed, the control rule determination unit 150 notifies the thread group control unit 141 of an increase in the upper limit value of the number of threads that can be executed in parallel (step S416). . In accordance with the notification, the thread group control unit 141 increases the upper limit value of the number of threads, and increases the maximum value of the parallel processing number in the server. At this time, the size of the message queue 130 may be increased from the initial value in accordance with the increase in the number of parallel processes.

  If the size of the message queue 130 is reduced in step S423 and the reception of the processing request message is performed by the message reception control unit 110, the server that has transmitted the processing request message does not complete the message transmission processing immediately after the message transmission. It appears that the processing request message has not reached the destination server. Therefore, in the server that is the message transmission source, the time required for message transmission is extended, and the service execution time (FIG. 11) is increased. When the service execution time becomes long and the service quality requirement cannot be satisfied in the message transmission source server, the number of threads or the size of the message queue is reduced. At this time, in the message transmission destination server, the apparent service execution time after receiving the message is not affected by the reception delay of the processing request message, so that the service processing performance is not deteriorated by the message reception delay.

  Hereinafter, a specific example will be described. FIG. 12 shows the configuration of a distributed service control system using a plurality of servers and a distributed service control device. Each server 10 exchanges processing request messages via the network 20 and performs a task (service) assigned to each server 10. Each server 10 has a distributed service control device 11 at the message input stage of the service unit 12 that executes the service. The configuration of the distributed service control apparatus 11 is the same as the configuration of the distributed service control apparatus 100 shown in FIG. The service unit 12 of each server 10 includes a communication unit 21 and a service execution unit 22. Processing for the processing request message is executed by the service execution unit 22.

  FIG. 13 shows the exchange of processing request messages between servers. Servers 51, 52, and 53 in FIG. 13 correspond to the servers 10 in FIG. Also, the distributed service control devices 101, 102, 103 of each server correspond to the distributed service control device 11 of each server in FIG. Here, it is assumed that an order receiving operation is considered as an operation, and a series of operations such as inventory confirmation, delivery, invoicing is executed across a plurality of servers in a distributed service control system. The processing request message includes information related to part ordering. Each server extracts and stores part data from the processing request message, and then transfers the processing request message as it is to the next server. In each server, it is assumed that a plurality of processing request messages are generated by dividing a processing request message, or batch processing is not performed by combining a plurality of processing request messages into one.

  Regarding the server performance, the performance of the servers 51, 52, and 53 is “high”, “medium”, and “low”, respectively. In the distributed service control apparatuses 101, 102, and 103 of each server, “3 seconds” is set as the quality requirement for the processing time for the processing request message. The initial size of the message queue 130 is “20”. The upper limit value of the number of threads in the processing request management unit 140 is “8”. In the message weight rule stored in the message weight storage unit 122 (FIG. 1) of the message state management unit 120, the processing time for the number of tags indicating the part data included in the XML document of the body of the processing request message is set. Yes.

  A processing request message is transmitted from the server 50 to the server 51 at a rate of 5 messages / second, and 5 processing request messages are transmitted from the server 51 to the server 52, from the server 52 to the server 53, and from the server 53 to the server 54, respectively. Suppose that it is transmitted at / second. Each processing request message includes 3 to 5 partial data (tags). In this state, in the distributed service control apparatus 103 of the server 53, the average processing waiting time of the processing request message staying in the message queue 130 is 1 second, and the service time in the service execution unit 22 is 1.5 seconds. To do.

  It is assumed that the number of processing request messages transmitted from the server 50 is increased and 10 processing request messages are transmitted from the server 50 to the server 51 per second. As the number of processing request messages increases on the upstream side, the number of processing request messages to the downstream servers 52, 53, and 54 per unit time increases. As a result, in the server 53 having the lowest server performance among the servers 51, 52, and 53, the processing for the processing request message cannot catch up, and the number of processing request messages staying in the message queue 130 of the distributed service control apparatus 103 increases. Thus, the average processing waiting time increases.

  In the distributed service control apparatus 103, the apparent service execution time (FIG. 11) increases as the average processing wait time increases. If the control rule determination unit 150 determines in step S414 in FIG. 9 that the apparent service execution time does not satisfy the service quality requirement “3 seconds”, the control rule determination unit 150 proceeds to step S421 to reduce the number of threads. In S422, it is determined that it is necessary to suppress the reception delay, that is, the number of input processing request messages, in order to make the apparent service time within 3 seconds, and the size of the message queue 130 is changed from “20”. The message is reduced to “10” and the message reception control unit 110 performs reception delay.

  In order to maintain the message queue size “10” after the change, the message reception control unit 110 delays the reception processing of the processing request message received thereafter, and suppresses the number of processing request messages received per unit time. . As a result, in the service execution unit 22 of the server 52 in the preceding stage of the server 53, the message transmission time (FIG. 11) becomes longer due to the completion of the process request message transmission process, and the service execution time is extended. As a result, the time during which the processing request message stays in the message queue 130 of the distributed service control apparatus 102 of the server 52 also becomes longer.

  When the control rule determination unit 150 determines that the service execution time (apparent service execution time) of the server 52 does not satisfy the quality requirement “3 seconds”, the control rule determination unit 150 performs the same as the operation of the server 53 in the distributed service control device 103. The number of threads is reduced and the size of the message queue 130 is changed. As a result, the server 51 on the upstream side of the server 52 also extends the service execution time due to the reception processing not being completed as in the case of the server 52, and the queue retention time of the processing request message becomes longer. As a result, the server 51 reduces the number of threads and changes the size of the message queue 130 so as to satisfy the service quality requirement.

  As described above, when the service quality requirement cannot be satisfied by the server 53, the reception time of the processing request message is delayed, so that the service processing time is extended by the upstream server 52 and the service quality requirement can be satisfied by the server 52. The server 52 sets the processing request message delay. By performing the reception delay in the server 52, when the service execution time of the server 51 is extended and the service quality requirement is not satisfied, the reception delay is performed to suppress the number of input processing request messages. As described above, when the load is concentrated on a specific server, the distributed service control system (FIGS. 12 and 13) sequentially delays the reception of processing request messages up to the upstream server, and is configured with a plurality of servers. Operates to maintain overall system performance. As a result, each server can perform processing that satisfies the service quality requirements.

  Next, let us consider a case where the service quality requirement is satisfied in the server 51 when the processing request message is transmitted from the server 50 to the server 51 at 5 messages / second. If the control rule determination unit 150 of the distributed service control device of the server 51 determines that the service quality performance requirement is satisfied in step S414 of the control rule determination process shown in FIG. The number of processes is gradually increased, and the number of parallel processes of the service execution unit 22 is increased within a range where the service time of the server 51 does not deteriorate. By doing in this way, the performance of the server 51 can be utilized to the maximum extent.

  If the number of parallel processes increases too much at the server 51 and a large number of processing request messages are sent to the downstream server 52 and the server 52 cannot meet the service quality requirements, the distributed service control device 102 of the server 52 will delay the reception. Do. In this case, the server 51 cannot satisfy the service quality requirement, and the control rule determination unit 150 decreases the number of threads. As described above, the server 51 increases the parallel processing number of the service execution unit 22 and controls the number of processing request messages to be processed within a range in which the service time of the server 51 and the service time of the downstream server do not deteriorate. In this way, it is possible to realize processing that satisfies the service quality requirements at each server while realizing processing according to server performance at each server.

  In summary, in the present embodiment, when the distributed service control apparatus 100 (FIG. 1) fails to satisfy the service quality requirement for the processing request message, the server (service execution unit 22 in FIGS. 12 and 13) The processing load on the server is reduced by reducing the number of parallel processes or reducing the size of the message queue 130 and performing a reception delay in the message reception control unit 110. By performing such control at the input stage of the processing request message to the server, it is possible to avoid the situation where the service execution time (apparent service execution time in FIG. 11) becomes long and the service quality requirement is not satisfied. , Processing that satisfies service quality requirements can be realized.

  In this embodiment, a simple structural analysis is performed on the received processing request message, the service execution time in the server is predicted according to the content, and the service quality requirement is satisfied using the predicted service execution time. The upper limit value of the parallel processing number of the thread 142 is controlled. By doing in this way, the processing request message processed simultaneously by the server can be suppressed within a certain range, and the server performance can be utilized to the maximum extent. Also, even if the number of processing request messages input per unit time temporarily increases, processing request messages that have arrived in large quantities are not processed at a time exceeding the upper limit of the parallel processing count of the thread 142. Processing that does not occur in the process and that satisfies the service quality requirements can be realized.

  In this embodiment, the server processing time calculated using the message weight rule is used when determining whether or not the processing time for the processing request message satisfies the service quality requirement. By performing control based on the load predicted based on the message structure before the processing request message is actually processed by the server, it becomes easy to quickly follow the fluctuation of the processing request message. In addition, the control rule determination unit 150 corrects the message weight rule based on the difference between the processing time at the actual server and the predicted value of the processing time, thereby improving the accuracy of the processing time prediction. Can be raised.

  In a system in which a plurality of servers operate in cooperation, when the distributed service control device 100 is arranged corresponding to each server, the distributed service control device 100 determines the reception delay and the parallel processing number of threads by changing the size of the message queue. Changes are made, and the reception delay and the number of parallel processes affect each other between the distributed service control devices of adjacent servers in the cooperation flow between the servers. As a result of the individual servers operating to meet the quality of service requirements, the optimal performance distribution is automatically reached and the service time of each server is guaranteed to meet the quality of service requirements. Time can be guaranteed. In addition, serialization by storing processing request messages prevents the server from being disabled due to a large number of processing requests, and at the same time, prevents the amount of stored messages transferred to the server from becoming a system bottleneck. The number of parallel processes can be changed.

  Further, in the present embodiment, a distributed service control device is arranged at the message input stage of each server to realize processing that satisfies service quality requirements. Therefore, unlike the technique described in Patent Document 2, a plurality of independent services are provided. It is not necessary to introduce a server that centrally manages monitoring and control of a plurality of services in order to guarantee a service level agreement (SLA) related to performance for a business composed of services. In the present embodiment, the distributed service control apparatus 100 controls each server so that the service time satisfies the service quality requirement, so that the processing time of the entire business can be kept within a certain range. Unlike the described technology, there is no need to carry relationship information between services corresponding to information managed by the central management server between services.

  In this embodiment, the distributed service control apparatus 100 increases the number of threads as long as the server performance has sufficient capacity. Therefore, the server processing performance is not suppressed by the message queue 130 and the initial number of threads. You can make the most of performance. In a system using a plurality of distributed service control devices, if the performance of the server downstream of the linkage flow is lower than the performance of the upstream server, the upstream You can take full advantage of server performance. As described above, the distributed service control device automatically and slowly adjusts the service execution, so that the server performance of each server can be utilized to the maximum while realizing the processing satisfying the service quality requirement in each server.

  Although the present invention has been described based on the preferred embodiments, the service control apparatus, method, program, and distributed service control system of the present invention are not limited to the above-described embodiments. Those in which various modifications and changes are made from the configuration of the embodiments are also included in the scope of the present invention.

  The present invention can be applied to an application such as a control program for maintaining a service quality by cooperating a plurality of services based on the processing performance of each other. Further, the present invention can also be applied to a service control device that operates on a computer independent of a server that executes a service.

The block diagram which shows the structure of the distributed service control apparatus of one Embodiment of this invention. The flowchart which shows the operation | movement procedure of a reception process. The figure which shows the specific example of the envelope of a general process request message. The figure which shows the structure of a message weight rule. The figure which shows the structure of message weight information. The flowchart which shows the process sequence of thread allocation. The flowchart which shows the operation | movement procedure of a thread | sled. The flowchart which shows the operation | movement procedure of the message weight rule correction | amendment in a control rule determination part. The flowchart which shows the operation | movement procedure of control rule derivation | leading-out in a control rule determination part. The figure which shows the structure of service quality requirement information. The figure which shows the execution time of the process with respect to a process request message. The block diagram which shows the structure of the distributed service control system using a some server and the distributed service control apparatus. The block diagram which shows the exchange of the process request message between the servers in a distributed service control system.

Explanation of symbols

10, 51-54: Server 12: Service unit 21: Communication unit 22: Service execution unit 11, 100-103: Distributed service control device 110: Message reception control unit 120: Message state management unit 121: Message weight calculation unit 122: Message weight storage unit 130: Message queue 140: Processing request management unit 150: Control rule determination unit 160: Service quality performance storage unit 170: Service quality requirement information 180: Processing request message 190: Message weight rule 200: Message weight information

Claims (25)

A message reception control unit that performs reception control of a processing request message and stores the received processing request message in a message queue;
Refer to a message weight storage unit that stores a message weight rule that defines a rule for calculating the processing time at the server using the structure of the processing request message, and to the processing request message and the message weight rule A message weight calculator for calculating a processing time at the server for the processing request message,
A thread group control unit that controls a thread group having a plurality of threads that execute a process by transmitting the process request message to a server, assigning the thread to the process request message, and the number of parallel processes of the thread A thread group control unit for controlling
With reference to a quality requirement information storage unit that stores quality requirement information related to quality requirements to be satisfied by the server, the residence time in the message queue of the processing request message, the calculated processing time in the server, A service control apparatus comprising: a control rule determining unit that controls a size of the message queue and an upper limit value of the number of parallel processes of threads in the thread group control unit based on quality requirement information.   The control rule determination unit determines whether processing for the processing request message satisfies a quality requirement based on the message residence time, the calculated processing time at the server, and the quality requirement. 2. The service control apparatus according to claim 1, wherein if it is determined that the value is not satisfied, at least one of a reduction in the size of the message queue and a reduction in an upper limit value of the parallel processing number of the thread is performed.   When the control rule determination unit determines that the processing for the processing request message does not satisfy the quality requirement, the control rule determination unit decreases the upper limit value of the parallel processing number of the thread and reduces the upper limit value of the parallel processing number of the thread. Whether processing for the processing request message in a state where the upper limit value of the number of parallel processes of the thread is reduced satisfies the quality requirement based on the processing time in the server, the retention time of the message, and the quality requirement The service control apparatus according to claim 2, wherein the size of the message queue is reduced when it is determined that the message queue is not satisfied.   The control rule determination unit determines whether processing for the processing request message satisfies a quality requirement based on the message residence time, the calculated processing time at the server, and the quality requirement. The service control apparatus according to claim 1, wherein if it is determined that the condition is satisfied, at least one of an increase in the size of the message queue and an increase in an upper limit value of the parallel processing number of the thread is performed.   When the control rule determining unit determines that the processing for the processing request message satisfies the quality requirement, the control rule determining unit checks whether the size of the message queue is smaller than a predetermined initial value. The service control apparatus according to claim 4, wherein the size of the message queue is increased, and the upper limit value of the number of parallel processes of the thread is increased when the size of the message queue is not smaller than the initial value.   The thread measures an actual measurement value of processing time from the start of processing to the end of processing for the processing request message in the server, and the control rule determination unit calculates the processing time in the calculated server. The message weight rule is corrected based on the measured value of the processing time so that a difference between the calculated processing time at the server and the measured value of the processing time becomes small. The service control apparatus according to any one of the above.   When the message reception control unit receives the processing request message, the message reception control unit determines whether or not there is a vacancy in the message queue. If the message reception control unit determines that there is no vacancy, the reception process of the processing request message is delayed until the vacancy occurs. The service control device according to any one of claims 1 to 6.   8. The service control apparatus according to claim 7, wherein when the processing request message is stored in the message queue, the message reception control unit transmits a response indicating reception completion to a transmission source of the processing request message. A distributed service control system that exchanges processing request messages between a plurality of servers and operates in cooperation with each other.
The distributed service control system according to claim 1, wherein each of the servers includes the service control apparatus according to claim 1 at an input stage of the processing request message. A service control method for controlling a processing request message to a server using a service control device,
A message receiving step in which the service control apparatus receives a processing request message and stores the received processing request message in a message queue;
The service control device refers to a message weight storage unit that stores a message weight rule that defines a rule for calculating the processing time at the server using the structure of the processing request message, and the processing request message, A message weight calculation step of calculating a processing time at the server for the processing request message based on the message weight rule;
The service control device controls a thread group having a plurality of threads that execute processing by sending the processing request message to a server, and assigns the thread to the processing request message;
The service control device refers to a quality requirement information storage unit that stores quality requirement information related to a quality requirement to be satisfied by the server, and the residence time of the processing request message in the message queue and the calculated server And a control rule determining step for controlling the size of the message queue and the upper limit value of the number of parallel processes of the thread based on the processing time of the message and the quality requirement information.   In the control rule determination step, the service control device, based on the message residence time, the calculated processing time at the server, and the quality requirement, the processing for the processing request message satisfies the quality requirement. The service control method according to claim 10, wherein at least one of reducing the size of the message queue and reducing the upper limit value of the number of parallel processes of the thread is performed when it is determined whether or not it is satisfied.   In the control rule determination step, when the service control device determines that the processing for the processing request message does not satisfy the quality requirement, the service control device reduces the upper limit value of the parallel processing number of the thread and increases the upper limit value of the parallel processing number of the thread. Processing for the processing request message in a state where the upper limit value of the parallel processing number of the thread is reduced based on the processing time in the server after the decrease in value, the residence time of the message, and the quality requirement The service control method according to claim 11, wherein it is determined whether or not a quality requirement is satisfied, and if it is determined that the quality requirement is not satisfied, the size of the message queue is reduced.   In the control rule determination step, the service control device, based on the message residence time, the calculated processing time at the server, and the quality requirement, the processing for the processing request message satisfies the quality requirement. The service control method according to claim 10, wherein at least one of an increase in the size of the message queue and an increase in an upper limit value of the parallel processing number of the thread is performed when it is determined whether or not it is satisfied.   In the control rule determination step, when the service control device determines that the processing for the processing request message satisfies the quality requirement, the service control device checks whether the size of the message queue is smaller than a predetermined initial value. 14. The service control method according to claim 13, wherein when the value is smaller than the value, the size of the message queue is increased, and when the size of the message queue is not smaller than the initial value, the upper limit value of the parallel processing number of the thread is increased. . The service control device measuring an actual measurement value of a processing time from the start of processing to the end of processing for the processing request message in the server in the thread;
The service control device reduces a difference between the calculated processing time at the server and the measured value of the processing time based on the calculated processing time at the server and the measured value of the processing time. The service control method according to claim 10, further comprising a step of correcting the message weight rule.   In the message receiving step, when the service control apparatus receives the processing request message, the service control apparatus determines whether or not there is an empty space in the message queue. The service control method according to any one of claims 10 to 15, wherein the reception process is delayed.   The service control method according to claim 16, wherein in the message reception step, the service control device transmits a response indicating reception completion to a transmission source of the processing request message when the processing request message is stored in the message queue. . A program for causing a computer to execute processing for controlling a processing request message to a server, wherein the computer
A message receiving step of receiving a processing request message and storing the received processing request message in a message queue;
Refer to a message weight storage unit that stores a message weight rule that defines a rule for calculating the processing time at the server using the structure of the processing request message, and to the processing request message and the message weight rule Based on a message weight calculation step of calculating a processing time at the server for the processing request message;
A thread assignment step of controlling a thread group having a plurality of threads that execute processing by sending the processing request message to a server, and assigning the thread to the processing request message;
With reference to a quality requirement information storage unit that stores quality requirement information related to quality requirements to be satisfied by the server, the residence time in the message queue of the processing request message, the calculated processing time in the server, A program for executing a control rule determining step for controlling a size of the message queue and an upper limit value of the number of parallel processes of the thread based on quality requirement information.   In the control rule determination step, based on the message residence time, the calculated processing time at the server, and the quality requirement, it is determined whether or not the processing for the processing request message satisfies the quality requirement. The program according to claim 18, wherein if it is determined that it is not satisfied, at least one of a reduction in the size of the message queue and a reduction in an upper limit value of the number of parallel processes of the thread is performed.   In the control rule determination step, when it is determined that the processing for the processing request message does not satisfy the quality requirement, the upper limit value of the parallel processing number of the thread is reduced and the upper limit value of the parallel processing number of the thread is reduced. Whether processing for the processing request message in a state where the upper limit value of the number of parallel processes of the thread is reduced satisfies the quality requirement based on the processing time in the server, the retention time of the message, and the quality requirement The program according to claim 19, wherein the size of the message queue is reduced when it is determined that the message queue is not satisfied.   In the control rule determination step, based on the message residence time, the calculated processing time at the server, and the quality requirement, it is determined whether or not the processing for the processing request message satisfies the quality requirement. The program according to claim 18, wherein if it is determined that it is satisfied, at least one of an increase in the size of the message queue and an increase in the upper limit value of the number of parallel processes of the thread is performed.   In the control rule determination step, when it is determined that the processing for the processing request message satisfies the quality requirement, it is checked whether or not the size of the message queue is smaller than a predetermined initial value. The program according to claim 21, wherein the size of the message queue is increased, and the upper limit value of the number of parallel processes of the thread is increased when the size of the message queue is not smaller than the initial value. In the computer,
Measuring an actual measurement value of a processing time from the start of processing to the end of processing for the processing request message in the server in the thread;
Based on the calculated processing time at the server and the measured value of the processing time, the message weight is set such that a difference between the calculated processing time at the server and the measured value of the processing time is reduced. The program according to any one of claims 18 to 22, further executing a step of correcting the rule.   In the message receiving step, when the processing request message is received, it is determined whether or not there is a vacancy in the message queue. If it is determined that there is no vacancy, the reception process of the processing request message is delayed until a vacancy occurs. The program according to any one of claims 18 to 23.   The program according to claim 24, wherein, in the message reception step, when a processing request message is stored in the message queue, a response indicating reception completion is transmitted to a transmission source of the processing request message.

Images