JP5565181B2 - Prediction device, prediction method, and prediction program - Google Patents

Description

  The present invention relates to an apparatus, a method, and a program for predicting a processing amount required for a plurality of processing apparatuses in a system including a plurality of processing apparatuses.

  The amount of processing required for a system including multiple processing devices is smaller than the processing amount that can be processed by the processing devices operating in the system, and is required even if some of the operating processing devices are excluded. When the processed processing can be executed, excessive power and machine resources such as CPU and memory are consumed in order to keep some of the processing devices in an operating state. In addition, when a part of a plurality of processing devices is set in a standby state, the required processing amount increases and the required processing amount exceeds the processing capacity of the operating processing device. Is also possible.

  For example, Patent Document 1 collects information related to the operating state of a computer system, records correlation information indicating a correlation between the collected information, and refers to the correlation information, from the collected information, A failure occurring in a service executed in the computer system is detected, a process for recovering the detected failure is generated, the correlation information is referenced, and the generated process is executed to the computer system. There is disclosed a technique for determining an effect and an effect to be performed, and determining necessity, execution order, and execution time of a process for which the effect and the effect are determined.

  Also, for example, in Patent Document 2, when a request for a database search process occurs in the application server such that the response time from the application server exceeds the specified time and the client terminal cannot accept the request from the client terminal An application server processing method is disclosed in which an error response is made without processing an error request.

  Further, for example, in Patent Document 3, load information for each buffer pool such as the number of basic reserved buffers, the current number of buffers, and the number of image holding buffers is acquired from the buffer pool management table at regular time intervals, and the acquired load It is disclosed that the minimum number of buffers required for a buffer pool lacking information is determined, and buffers are added to or deleted from each buffer pool when data is not output. Yes.

  For example, Patent Document 4 discloses that a usage amount of a personal computer by a user is measured through a log file to predict a computer usage tendency.

JP 2008-9842 A JP 2001-325223 A JP 2000-57027 A Japanese National Patent Publication No. 10-510647

  However, when the processing amount required for the system increases, even if the processing device in the standby state is put into operation, it takes time for the processing device to be in operation. Even if it became clear that the quantity exceeded the processing capacity of the active processing equipment group in the system, the situation could not be improved at that time.

  An object of the present invention is to predict that the amount of processing required for a system will exceed the processing capacity of an active processing device group in the system.

  As a first proposal, based on an order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to a request, the first of the plurality of types of processes The estimated time from the execution of the first process to the execution of the second process among the plurality of types of processes is calculated, and when the first process is detected, the calculated time is calculated. A prediction program that causes a computer to predict that the second process is executed after the time indicated by the prediction time is used.

  As a second proposal, based on the order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to the request, the first of the plurality of types of processes The estimated time from the execution of the first process to the execution of the second process among the plurality of types of processes is calculated, and when the first process is detected, the calculated time is calculated. A prediction method including predicting that the second process is executed after the time indicated by the prediction time is used.

  As a third proposal, based on the order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to the request, the first of the plurality of types of processes A calculation means for calculating a predicted time from execution of the first process to execution of the second process of the plurality of types of processes, and when detecting that the first process has been executed, And a prediction unit that predicts that the second process is executed after the time indicated by the calculated prediction time.

  As a fourth plan, a first device that is in operation and a second device that is on standby, and a third device that can communicate with the first device and the second device, in response to a request. Based on the order relationship between the processes of the plurality of types of processes defined in the source code that defines the plurality of types of processes to be executed, the plurality of types of processes are executed, and then the plurality of processes are performed. The calculation means for calculating the predicted time until the second process of the types of processes is executed, and after detecting the execution of the first process, after the time indicated by the calculated predicted time A prediction system comprising: a prediction unit that predicts that the second process is executed; and an output unit that outputs an operation instruction to the second device according to a prediction result of the prediction unit. Use.

  The present invention predicts that the amount of processing required for the system exceeds the processing capacity of the active processing device group in the system.

The example of arrangement | positioning of the apparatus in 1st embodiment is shown. The example of the process performed in 1st embodiment is shown. The model figure of the example of the state transition of the program performed with the processor group 2a-2e is shown. The functional block about the management apparatus 1 is shown. The functional block about the analysis means 12 is shown. The table B1 memorize | stored in the memory | storage means 122 is shown. The table B2 memorize | stored in the memory | storage means 122 is shown. The table B3 memorize | stored in the memory | storage means 13 is shown. The table B4 memorize | stored in the memory | storage means 13 is shown. The functional block about the instruction | indication means 14 is shown. The table B5 memorize | stored in the memory | storage means 142 is shown. The table B6 memorize | stored in the memory | storage means 142 is shown. A table B7 stored in the storage unit 142 is shown. 1 shows a device configuration diagram of a management device 1. FIG. A table B8 stored in the storage unit 13 is shown. The flow of the process performed by the analysis means 12 is shown. The flow of the process performed by the analysis means 12 is shown. The flow of the process performed by the analysis means 12 is shown. The flow of the process performed by the analysis means 12 is shown. A flow of processing performed by the instruction unit 14 is shown. A flow of processing performed by the instruction unit 14 is shown. A flow of processing performed by the instruction unit 14 is shown. A flow of processing performed by the instruction unit 14 is shown. An example of processing amount prediction is shown. The functional block of the instruction | indication means in 2nd Embodiment is shown. The table B10 memorize | stored in the memory | storage means 142 is shown. A flow of processing performed by the instruction unit 14 is shown.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[First Embodiment]
First, an overview of the system of this embodiment will be described.

  FIG. 1 shows an arrangement example of the apparatus of the present embodiment. 1 is a management device, 2a to 2e are processing devices, 3a to 3e are storage devices, 4 is a load distribution device, 5 and 6 are networks, and 7 is a terminal device. Show.

  The management device 1 is a computer capable of transmitting instructions to the processing devices 2a to 2e and the storage devices 3a to 3e based on information acquired from the processing devices 2a to 2e by communication via the network 5. Alternatively, any of the processing devices 2a to 2e may have the function of the management device 1.

  Each of the processing devices 2 a to 2 e is a computer that can serve as a server that executes processing requested by the terminal device 7. For example, as a role played by the processing devices 2a to 2e, a role of providing a Web application in response to an access from the terminal device 7 can be considered. In FIG. 1, as an example, the processing devices 2a to 2c that are in operation are indicated by solid lines, and the standby processing devices 2d and 2e are indicated by dotted lines. Further, in FIG. 1, five processing devices are illustrated, but the number is not limited to this.

  An active processing device is a processing device in a state where processing according to a request from the terminal device 7 can be executed, and a standby processing device executes processing according to a request from the terminal device 7. It is a processing apparatus in a state that takes time to shift to a possible state. The standby processing device is, for example, a processing device that is in a standby state or a processing device that is stopped. The standby state depends on the operational definition, but is, for example, a state in which the main power supply is ON and power supply is partially stopped. A plurality of operational standby states may be defined.

  Each of the storage devices 3a to 3e is a storage device including, for example, an HHD (Hard Disk Drive), an SSD (Solid State Drive), and the like. Similarly to the processing devices 2a to 2e, the operating storage devices 3a to 3c are indicated by solid lines, and the standby storage devices 3d and 3e are indicated by dotted lines.

  The load balancer 4 is a device that performs processing for distributing processing requests from the terminal device 7 to any of the processing devices 2a to 2c in operation.

  The terminal device 7 is an example of a client-side device that transmits processing requests to the processing devices 2a to 2e. The terminal device 7 may be a computer or the like. Although not shown, the terminal device 7 may be a mobile phone or a portable information terminal (PDA) that can be connected to the network 6 wirelessly. The number of terminal devices 7 is not limited to the number shown in FIG.

  Before explaining the processing of the management apparatus 1 in the present embodiment, an outline of processing performed by the system of the present embodiment will be described with reference to FIGS. 2 and 3.

  FIG. 2 shows an example of processing performed using the system of this embodiment. A1 to A9 in FIG. 2 indicate processing performed in each of the terminal device 7, the processing device 2a, and the storage device 3a in time series.

  A1 indicates that the terminal device 7 inputs an instruction of a reference request (for example, a Web page acquisition request) of data stored in the storage device 3a by the user. A2 indicates that the terminal device 7 transmits data including the reference request to the processing devices 2a to 2e.

  A3 indicates that the processing device 2a receives the data transmitted by the terminal device 7 in A2 and distributed by the load distribution device 4. A4 indicates that data including a read request is transmitted to the storage device 3a in accordance with the data received by the processing device 2a.

  A5 indicates that the storage device 3a receives data including the read request transmitted by the processing device 2a in A4. A6 indicates that the storage device 3a transmits the read data in response to the read request included in the received data to the processing device 2a.

  A7 indicates that the processing device 2a receives the data transmitted by the storage device 3a in A6. A8 indicates that the processing device 2a transmits the data received in A7 to the terminal device 7.

  A9 indicates that the terminal device 7 receives the data transmitted by the processing device 2a in A8. For example, when the input at A1 is a Web page acquisition request, the terminal device 7 performs processing such as switching the screen display based on the data received at A9.

  The processing illustrated in FIG. 2 is merely an example, but the processing devices 2b to 2e can execute the same processing as the processing device 2a, and the storage devices 3b to 3e can also execute the same processing as the storage device 3a.

  In FIG. 2, the storage device 3a performs a data read process requested by a read request during a period from A5 to A6. The time required for this read process varies depending on the content of the read request. For example, the storage device 3a refers to another table based on data read with reference to one data table, rather than reading data with reference to one table stored in the storage device 3a. It is considered that data reading (processing including join etc. in the SQL statement) takes more time for reading processing. Therefore, depending on the content of the read request transmitted by the processing device 2a, it can be predicted whether the read process takes time. Hereinafter, a reading process that takes a particularly long time is referred to as a “heavy process”.

  In FIG. 2, the processing device 2 a performs processing according to the reference request from the terminal device 7 during the period from A3 to A8. This process is executed in response to receiving data including a reference request from the terminal device 7 (A3). Hereinafter, the trigger of processing executed by such a processing apparatus is referred to as an event (E).

  Further, what data is acquired by the terminal device 7 that has generated the event (E) in the processing device 2a (if the event (E) is a Web page reference request from the terminal device 7), which Web page is displayed. Is called state (S). That is, in FIG. 2, a series of processes (A3 to A8) corresponding to the event (E) of the processing apparatus 2a can be said to be a process of changing the state (S) of the terminal apparatus 7 that has generated the event (E). . That is, the program that causes the processing device 2a to execute a series of processes (A3 to A8) is a process in which the state (S) is changed in response to the event (E). The information indicating the state (Sb) before the transition and the information indicating the subsequent state (Sa) can be acquired by the processing of the program corresponding to the event (E).

  When the processing devices 2a to 2e are processing device groups that execute a program that provides a structured Web page group, the event (E) is, for example, a jump instruction by a link between Web pages, and the state ( S) makes a transition by transmitting Web page data to the terminal device 7. In this case, the relationship between the event (E), the state before the transition (Sb), and the state after the transition is determined based on the configuration of the Web page group. Moreover, the character input to the input form in the Web page may be an event (E), and the state before and after the character input may be a state before transition (Sb) and a state after transition (Sa). The relationship between this event (E), the state before transition (Sb), and the state after transition will be described with reference to FIG.

  FIG. 3 is a diagram modeling an example of state transition defined in a program executed by the processing devices 2a to 2e. Each of S1 to S7 indicates a state, and each of E1 to E7 indicates an event.

  In FIG. 3, the arrows of the events (E1 to E7) represent state transitions due to the occurrence of each event. For example, the arrow of the event (E1) in FIG. 3 indicates that the process (A1) corresponding to the event (E1) occurs in any of the processing devices 2a to 2e when the event (E1) occurs when the state (S1) -A8 is performed and the state (S2) is transitioned to.

  The event (E) is input not only by any of the processing devices 2a to 2e receiving data including a Web page acquisition request from the terminal device 7, but also, for example, by a user inputting data on a predetermined Web page. It may be to receive data. In this case, the state (S) is also defined as data input on a predetermined page. In addition, the event (E) may be that a predetermined time has elapsed since the acquisition of the Web page, or a session connection request or disconnection request may be used as the event (E).

Next, the process of the management apparatus 1 in this embodiment is demonstrated.
[Configuration of Management Device 1]
FIG. 4 shows a functional block configuration of the management apparatus 1. 11 indicates communication means, 12 indicates analysis means, 13 indicates storage means, and 14 indicates instruction means.

  The communication unit 11 performs processing for transmitting / receiving data to / from the processing devices 2 a to 2 e via the network 6. Further, the communication unit 11 may perform processing for transmitting / receiving data to / from the storage devices 3a to 3e in addition to the processing devices 2a to 2e.

  As data transmission / reception, for example, reception of a source code (C1) of a program executed by the processing devices 2a to 2e, and transmission of a source code (C2) created by the analysis means 12 adding a code to the source code (C1) , Receiving event notifications from the processing devices 2a to 2e, transmitting instructions from the instruction means 14 to any of the processing devices 2a to 2e, and the like.

  The analysis means 12 is defined in the source code (C1) received by the communication means 11 as to what processing the processor that executes the source code (C1) executes according to an event that has occurred in the processor. Analyzes whether or not In addition, when the event (Ei) occurs in the source code (C1), the analysis unit 12 identifies data (event ID) that identifies the event (Ei) that has occurred, and data that identifies the client that caused the event. Source code (C2) to which a code that specifies that (for example, session ID) is transmitted to the management apparatus 1 as an event notification is added. Furthermore, the analysis means 12 is based on the result of analyzing the source code (C1) for integers i and j within the range of 1 to m (the total number of events defined in the source code (C1) is m). The time (Tij) until the occurrence of another event (Ej) after the occurrence of the event (Ei) and the probability of occurrence (Pij) are estimated and stored in the storage means 13. Specific processing of the analysis unit 12 will be described later with reference to FIG.

  The time (Tij) and the probability (Pij) may be estimated based on design information such as a screen transition diagram used for creating the source code (C1) instead of the source code (C1). Based on either the source code (C1) or the design information, the processing device retains the source code or design information corresponding to the updated program when the program executed by the processing device is updated.

  The storage unit 13 stores the time (Tij) and the probability (Pij) obtained by the analysis unit 12. Further, the time (Tij) or the probability (Pij) is read in response to the read request from the instruction means 14.

  The instructing unit 14 predicts the occurrence timing of a predetermined event based on the event notification received by the communication unit 11 and the time (Tij) and probability (Pij) stored in the storage unit 13, and further generates the event at the predicted timing. When the number of processing devices necessary for processing the event to be executed is not operating, the waiting processing devices 2d and 2e are allowed to transmit an operation instruction to the communication unit 11 in anticipation of the time required for the operation. Alternatively, when the processing of the predicted event can be sufficiently executed even if a part of the processing devices in operation is stopped, the communication unit 11 is instructed to shift or stop the processing devices to the standby state. To send. The details will be described later with reference to FIG.

By using the management device 1 shown in FIG. 4, the processing loads of the processing devices 2a to 2e are based on the programs executed by the processing devices 2a to 2e and the status of access being performed on the processing devices 2a to 2e. Can be activated in anticipation of the activation time of the processing apparatus according to the predicted processing load.
[Configuration of Analysis Unit 12]
FIG. 5 shows functional blocks for the analyzing means 12. 121 denotes a control means, 122 denotes a storage means, 123 denotes an extraction means, 124 denotes a generation means, and 125 denotes a calculation means.

  The control unit 121 determines a processing device in operation based on the table B8 (see FIG. 15) stored in the storage unit 13, and obtains the source code (C1) of the program from the processing device in operation, storage unit 122 performs data writing / reading processing, processing for instructing the communication means 11 to transmit the source code (C2) to the processing devices 2a to 2e, and the like.

  As shown in FIG. 15, the table B8 has a record for each processing device, and has fields for storing information such as the processing device address and the operating state of the processing device. The information included in this table B8 may be set by the administrator, or the management apparatus 1 may automatically detect the processing apparatus and additionally set the information. Further, the management device 1 may periodically inquire and update the processing device registered in each record of the table B8 for the operating state.

  The extraction unit 123 includes a code that specifies that an event occurrence is detected from the source code (C1) read by the control unit 121 from the storage unit 122, and the contents of processing that is executed after the event detected by the code is generated. Extract the code that defines

  The extraction unit 123 generates an event ID of an event detected by the extracted code, and further transitions with data indicating a state before transition (pre-transition state: Sb) by processing executed in accordance with the event. Data indicating a later state (post-transition state: Sa) is extracted.

  The control unit 121 adds a record including the event ID, data indicating the pre-transition state (Sb), and data indicating the post-transition state (Sa) to the table B1 stored in the storage unit 122. For example, if the occurrence of an event is a Web page acquisition request from the terminal device 7, the data indicating the pre-transition state (Sb) may be the address of the Web page provided before receiving the acquisition request. The data indicating the post-transition state (Sa) may be the address of a Web page provided in response to an acquisition request.

  Based on the code extracted by the extraction unit 123 and the event ID generated by the extraction unit 123, the generation unit 124 performs processing for notifying the management apparatus 1 of the event ID and the session ID when the extracted code is executed. Generate the specified code. The session ID is information for identifying the session including the access that generated the event from other sessions. Instead of the session ID, data that can identify the access source accessing the processing apparatus may be used. For example, a login ID for a Web site provided by the processing devices 2a to 2e may be used instead of the session ID. The generated code is added to the source code (C1) stored in the storage unit 122 by the control unit 121.

  The storage unit 122 stores, for each event ID, a table B1 having a record including an event ID and data indicating a pre-transition state (Sb) and a post-transition state (Sa). The storage unit 122 stores setting data used in the processing of the calculation unit 125. The setting data includes, for example, predetermined values indicating how long each state defined in the source code (C1) lasts.

  Based on the table B1 and setting data stored in the storage unit 122, the calculation unit 125 generates an event (Ej) after an event (Ei) has occurred for each of the integers 1 and n of 1 to n. And the probability (Pij) that the event (Ej) occurs after the event (Ei) occurs. The control unit 121 causes the storage unit 13 to store the time (Tij) and the probability (Pij) calculated by the calculation unit 125.

  When the calculation means 125 calculates the time (Tij) and the probability (Pij), the relationship between the state (S) and the event (E) as shown in FIG. 3 is defined in the source code (C1). Will be described as an example.

  FIG. 6 shows a table B1 in which event IDs stored in the storage unit 122 are associated with data indicating the state (Sb) and the state (Sa). The table B1 in FIG. 6 has a record for each of the events (E1 to E7) shown in FIG. 3, and the records with event IDs E1 to E7 are the pre-transition state (Sb) and the post-transition state (Sa), respectively. including.

  The calculation unit 125 extracts the state defined in the program from the data included in the table B1 read from the storage unit 122 by the control unit 121. The calculation unit 125 sets a holding time (Ti) for each of the extracted states (S1 to S7). The holding time (Ti) is an index indicating the time from the transition to the state (Si) to the transition to another state when transitioning to the state (Si). The holding time (Ti) is determined based on setting data stored in the storage unit 122. In the present embodiment, the same value is set for the holding time (Ti) for each state (Si). As will be described later, in the second and third embodiments, processing for updating the holding time (Ti) is performed based on the event notification statistics.

  Further, the calculation means 125 sets the probability (Pi) that the event (Ei) occurs when the pre-transition state (Sb) is reached for each of the events (E1 to E7) defined in the program. The probability (Pi) may be set according to the number of events (number of branches) having the same pre-transition state (Sb). For example, the number of events whose state (Sb) before transition is state (S1) is two (event (E1) and event (E2)) (state (S1) has two branches) ( It is conceivable to set the probability (P1) and the probability (P2) to 0.5, respectively. As for the probability (Pi), similarly to the holding time (Ti), in the second and third embodiments, the probability (Pi) is updated based on the event notification statistics.

  Based on the holding time (Ti) and probability (Pi) calculated by the calculation means 125, the control means 121 stores the probability (E1 to E7) corresponding to each event (E1 to E7) in the table B1 of FIG. Pi) and a holding time (Ti) corresponding to the pre-transition state (Sb) of each event (E1 to E7) is added. FIG. 7 shows the table B2 of FIG. 7 in which the holding time (Ti) and the probability (Pi) are added.

  Furthermore, the calculation means 125 calculates the time (Tij) for the transition from the event (Ei) designated by the control means 121 to the event (Ej) designated by the control means 121 based on the table B2 shown in FIG. And the probability (Pij).

  First, the calculation unit 125 specifies a route that passes from the event (Ei) to the event (Ej) based on the information in the table B2 read from the storage unit 122 by the control unit 121. For example, from event (E6) to event (E3), “event (E6) → status (S6) → event (E7) → status (S1) → event (E1) → (status S2) → event (E3 ) ". This specifying method will be described later using the flow shown in FIG.

  Next, the control unit 121 associates each event (Ek) (including event (Ej)) that occurs after the event (Ei) occurs with the event (Ek) in the table B2 for the identified route. The stored retention time (Tk) and probability (Pk) are read out. The calculation unit 125 calculates the total holding time (Ek) read by the control unit 121 and sets it as the time (Tij) for transition from the event (Ei) to the event (Ej). Further, the calculation unit 125 calculates the product of the probabilities (Pk) read by the control unit 121 and sets the product (Pij) as the probability of transition from the event (Ei) to the event (Ej). For example, the transition time (T63) from the event (E6) to the event (E3) is represented by T6 + T1 + T2, and the probability (P63) is represented by P7 × P1 × P3.

  As described above, the time (Tij) and the probability (Pij) calculated by the calculation unit 125 are written into the storage unit 13 by the control unit 121. The storage means 13 stores time (Tij) and probability (Pij) in the data formats shown in FIGS. 8 and 9, respectively. The table shown in FIG. 8 is called table B3, and the table shown in FIG. 9 is called table B4.

As described above, based on the source code (C1) received by the communication unit 11, the analysis unit 12 adds a code for transmitting an event notification to the management apparatus 1 to the source code (C1). Is generated, and a prediction time (Tij) and a probability (Pij) for a transition between events defined in the source code (C1) are calculated and stored in the storage unit 13.
[Configuration of Instruction Unit 14]
FIG. 10 shows a functional block configuration of the instruction means 14. 10, reference numeral 141 denotes a control means, 142 denotes a storage means, 143 denotes a generation means, 144 denotes a calculation means, 145 denotes a counting means, 146 denotes a determination means, and 147 denotes a generation means. Indicates.

  The instructing unit 14 predicts the timing at which a predetermined event (Ej) occurs, and performs processing for causing the processing device to output an operation instruction according to the prediction. Assuming that the event causing the above-described “heavy process” is a predetermined event (Ej), the timing at which the “heavy process” occurs can be predicted. Further, not only the event (Ej) but also a plurality of events such as the event (Ej) and the event (Ek) can be predicted to predict the occurrence timing and occurrence probability of each event. By adding together the occurrence timing and occurrence probability of a plurality of events, the processing load can be predicted without being limited to a single process.

  Based on the event ID (Ei) included in the event notification received by the communication unit 11, the control unit 141 determines the time (Tij) from the event (Ei) to the event (Ej), and the event (Ei) to the event ( Ej) is read from the storage unit 13 and stored in the storage unit 142 in association with the session ID (Zi) included in the event notification received by the communication unit 11 and the time (Hi) when the event notification is received. To do. These are stored in the storage unit 142 in the form of a table B5 shown in FIG.

  When data about the same session ID (Zi) is stored when storing in the storage unit 142, the data about the session ID (Zi) stored in the storage unit 142 is used as the predicted time (Tij). , It may be updated at the probability (Pij) and the time (Hi). Further, when the event notification indicates that the session is disconnected, the data regarding the session ID included in the event notification may be deleted from the storage unit 142. Alternatively, a flag may be provided for each record of the table B5 stored in the storage unit 142, and it may be configured to control whether each record is valid by turning the flag ON / OFF. For example, when data about the same session ID (Zi) is received, control may be performed to turn off the flag of a record that has already been stored, or when data indicating session disconnection is received, Control to turn off the flag of the record including the session ID may be performed.

  The table B5 shown in FIG. 11 will be described. The table B5 shown in FIG. 11 includes fields for session ID (Zk), probability (Pij), time (Tij), reception time (Hk), time range ID (Rs), and flag. Each record stores data on the session (Zk) identified by the session ID (Zk). Each record has a probability (Pij), time (Tij), reception time (Hk), and flag set based on the latest event notification.

  Incidentally, in the time range ID (Rs) field of the table B5, any one of the information shown in the table B6 of FIG. 12 is stored based on the time (Tij) and the reception time (Hk). The selection of the time range (Rs) is performed by the generation unit 143 and the calculation unit 144. This table B6 is stored in advance in the storage means 142.

  For example, when the time required for the processing apparatus in the standby state to be in the operating state is 30 seconds, the time range may be divided as shown in the table B6 of FIG. In this case, the instruction given by the management apparatus 1 is an instruction to activate the processing apparatus in the standby state when the processing amount expected in the time range (R3) is large, and the processing amount expected in the time range (R3) is If the time range (R4) is large and the time range (R4) is large, it may be possible to wait for 30 seconds before setting the operation instruction to operate the processing apparatus in the standby state.

  The generation unit 143 outputs data indicating that the predetermined time has passed to the control unit 141 every time the predetermined time has passed. The control unit 141 starts the counting process based on the data acquired from the generation unit 143. The time when the counting process is started is defined as time (H0).

  The calculation means 144 calculates the difference (H0−Hk) between the time (H0) and the reception time (Hk) for each record in the table B5 stored in the storage means 142, and further calculates the predicted time (Tij). A difference (Tij− (H0−Hk)) from the calculated difference is calculated. The control means 141 reads the time range ID (Rs) indicating the time range including the difference (Tij− (H0−Hk)) calculated for each record from the table B6 and stores it in the time range ID field.

  The tabulating unit 145 reads the probability (Pij) for the records having the same time range ID (Rs) for each time range ID (Rs) from R1 to R5 with reference to the table B5, and the read probability (Pij) The sum (Psums) is calculated.

  As an example, a case will be described in which the counting unit 145 performs processing on the table B5 illustrated in FIG. First, a field with a time range ID R1 is extracted by referring to the field for the time range ID in the table B5. In the case of the table B5 shown in FIG. 11, first, a record whose session ID is Z1 is extracted. Next, data indicating P5j is read from the probability field of the record whose session ID is Z1. As long as it is shown in FIG. 11, since the record whose time range ID is R1 is only the record whose session ID is Z1, the total value (Psum1) of the read probabilities (Pij) is P5j.

  Next, the counting unit 145 calculates the total value (Psum2) of the probabilities (Pij) for the records whose time range ID is R2 in the table B5. In that case, as the record whose time range ID is R2, the record with session ID Z3 and the record with Z5 correspond, the probability (P1j) is read from the record with session ID Z3, and the probability (from the record with session ID Z5) P3j) is read out. As long as it is shown in FIG. 11, the total value (Psum) of the read probabilities (Pij) is P1j + P3j. The same process is applied to R3 to R5. The calculated total value (Psums) is stored in the storage unit 142 in the form of the table B7 shown in FIG. 13 in association with the corresponding time range ID (Rs).

  The determination unit 146 refers to the table B7 stored in the storage unit 142, and determines whether the total value (Psums) for each time range (Rs) exceeds the processing capability of the operating processing apparatus. If the determination unit 146 determines that the processing capacity is exceeded, the determination unit 146 reads the time range ID (Rs) for the time range exceeding the processing capacity. At this time, when a plurality of predetermined events (Ej) are set, the total value (Psums) for each is added and determined.

  The generation unit 147 generates data for instructing the processing devices 2a to 2e to operate based on the time range ID (Rs) read by the determination unit 146.

  For example, if it is determined that it takes 30 seconds from the generation of the activation instruction to activate each of the processing devices 2a to 2e, and it is determined that the processing capacity is exceeded in the time ranges R1 and R2, the processing capacity is insufficient. Therefore, the generation unit 147 generates an operation instruction for a processing device that is not operating. If it is determined that the processing capacity is exceeded in the time range R3, it is predicted that the processing capacity of the operating processing apparatus will be insufficient after 30 seconds from the time (H0). An operation instruction for a processing apparatus that is not operating is generated.

  When it is determined that the processing capacity is exceeded in the time range R4 or R5, it is expected that the processing capacity of the processing apparatus in operation is insufficient after 60 seconds or 90 seconds have elapsed from the time (H0). Therefore, when it is determined that the processing capacity is exceeded in the time range R4, the generation unit 147 generates an operation instruction for a processing device that has not been operated after 30 seconds, and performs processing in the time range R5. If it is determined that the capacity is exceeded, an operation instruction is generated for a processing apparatus that has not been operated after 60 seconds.

  The control unit 141 performs processing for transmitting the operation instruction generated by the generation unit 147 to the communication unit.

When the instructing unit 14 performs the above-described processing, the management device 1 can perform a predetermined time elapsed from the time (H0) based on the access made to the processing devices 2a to 2e at the time (H0). The processing load can be predicted, and the processing apparatus can be additionally operated in consideration of the startup time of the processing apparatus.

For example, the process of the management device 1 described above may be executed without using a probability. In that case, the predetermined process may be executed when the time value has elapsed from the reception time of the event notification. By performing the prediction using the probability, it is possible to consider that it is not necessarily executed after the event notification, and therefore, an effect of reducing the excessive amount of processing can be obtained.

  FIG. 14 shows an example of the device configuration of the management device 1. 1a indicates a processor, 1b indicates a memory, 1c indicates a communication interface, 1d indicates a storage device, and 1e indicates a bus.

  The processor 1 includes a CPU (Central Processing Unit), the memory 1b includes a RAM (Random Access Memory), the communication interface includes a NIC (Network Interface Card), etc., and the storage device 1d includes an HDD (Hard Disk Drive), It includes SSDs (Solid State Drives), disk drives for disks such as CDs and DVDs, and the like.

  The processor 1a controls the management device 1 as a whole. The program temporarily stored in the memory 1b is read, and arithmetic processing is performed according to the read program. The memory 1b temporarily stores at least part of an OS (Operation System) and application programs, and stores data used by the processor 1a for processing. The communication interface 1c performs communication processing with other devices. The storage device 1d stores an OS and application programs, and stores data according to the control of the processor.

  The processing of the functional blocks shown in FIGS. 4, 5, and 10 can also be realized by the device configuration shown in FIG.

  The process of the communication means 11 is implement | achieved when the communication interface 1c performs the process according to control of the processor 1a.

  Processing of each functional block included in the analysis unit 12 is realized as follows.

  The processing of the control means 121 is realized by the processor 1a performing read / write control on the data stored in the memory 1b or the storage device 1d according to the program read from the memory 1b. The processing of the storage unit 122 is realized by the memory 1b or the storage device 1d writing / reading data according to the control of the processor 1a. The processing of the extracting unit 123 is realized by the processor 1a performing processing corresponding to the program read from the memory 1b or the storage device 1d on the data read from the memory 1b or the storage device 1d. The processing of the generation unit 124 is realized by the processor 1a reading data from the memory 1b or the storage device 1d. The processing of the calculation means 125 is to read data (setting data etc.) from the memory 1b or storage device 1d according to the data (event ID etc.) read from the memory 1b or storage device 1d by the processor 1a, This is realized by performing a corresponding calculation process.

  The processing of the storage means 13 is realized when the storage device 1d or the memory 1b writes / reads data according to instructions from the processor.

  The processing of each functional block included in the instruction unit 14 is realized as follows.

The processing of the control means 141 is realized by the processor 1a performing processing according to the program read from the memory 1b for the data stored in the memory 1b or the storage device 1d. The processing of the storage unit 142 is realized by the memory 1b or the storage device 1d storing data in accordance with an instruction from the processor 1a. The processing of the generation unit 143 is realized by the processor 1a determining a predetermined time based on a clock included in the processor 1a. The processing of the calculation unit 144 is realized by the processor 1a reading data from the memory 1b or the storage device 1d and performing arithmetic processing according to the program on the read data. The processing of the counting unit 145 is realized by the processor 1a reading data from the memory 1b or the storage device 1d and performing arithmetic processing according to the program on the read data. The processing of the determination unit 146 is realized by the processor 1a reading data from the memory 1b or the storage device 1d and further reading data from the memory 1b or the storage device 1d based on the read data. The processing of the generation unit 147 is realized by generating data based on data read from the memory 1b or the storage device 1d and time information based on a clock included in the processor 1a.
[Processing of Management Device 1]
FIG. 16 shows a flow of processing for analyzing the source code (C1) and adding a code that defines processing for notifying event occurrence.

  When a start instruction is input to the management device 1 (S101), the control unit 121 causes the communication unit 11 to transmit a source code (C1) acquisition request to any of the processing devices 2a to 2e (S102). When the communication unit 11 receives the source code (C1), the control unit 121 stores the received source code (C1) in the storage unit 122 (S103).

  First, the control means 121 sets a number i specifying a unit for reading a code from the source code (C1) to 0 (S104). The unit for reading out the code may be one line of the source code (C1) or a routine unit defined in the source code (C1). Next, the control means 121 adds 1 to the number i (S105), and checks whether there is a code to be read in the reading unit designated by the number i (S106). When there is no code to be read (S106: NO), a process of calculating a probability value (Pij) and a time value (Tij) is executed (S113). If there is a code to be read (S106: YES), the control means 121 reads the code designated by the number i from the storage means 122 (S107).

  The extraction unit 123 determines whether the code read by the control unit 121 is a code for detecting the occurrence of an event (S108). This S108 is determined based on whether or not the read code is a code prescribed to perform processing in response to external access. If the read code is not a code for detecting an event (S108: NO), the process of S105 is executed.

  When the read code is a code for detecting an event (S108: YES), the extraction unit 123 generates an event ID that can distinguish the detected event from other events (S109). After S109, the control means 121 executes a flow for analyzing the source code (C1) regarding the extracted event (S110).

  Next, when the code for detecting an event is executed, the generation unit 124 generates a code that notifies the management apparatus 1 of an event ID indicating the event (S111). The control unit 121 performs a process of adding the code generated by the generation unit 124 to the source code (C1) stored in the storage unit 122 (S112). After executing S112, the process returns to S105.

  FIG. 17 shows a flow of analyzing the source code (C1) regarding the event extracted in S108.

  When S110 is executed (S201), the control unit 121 stores the source code (C1) in which the code defining the processing to be executed after detecting the event indicated by the event ID generated in S109 is stored in the storage unit 122 (S202).

  The control unit 121 extracts, from the code read in S202, information indicating each of the state before the transition (Sb) and the state after the transition (Sa) by the process executed after the event occurs (S203). Further, the control unit 121 adds a record including information indicating the pre-transition state (Sb) and post-transition state (Sa) extracted in S203 and an event ID to the table B1 of the storage unit 122 (S204). When S204 is executed, the code analysis flow ends and S111 is executed (S205).

  18 and 19 show a flow for calculating the probability value (Pij) and the time value (Tij).

  When S113 is executed (S301), the calculation means 125 sets all types of states (Si: i = 1 to n) in the pre-transition state (Sb) and the post-transition state (Sa) extracted in S203. On the other hand, a holding time (Ti) is set (S302). When the total number of extracted states is n, the holding time (Ti) is set for the states (Si) corresponding to the respective integers i from 1 to n. As described above, the holding time (Ti) indicates the time from the transition to the state (Si) to the transition to another state (S).

  When S302 is executed, the control unit 121 adds a holding time (Ti) corresponding to the pre-transition state (Sb) to each record of the table B1 stored in the storage unit 122 (S303).

  Further, the calculation means 125, for each of the states (Si) extracted in S203, according to the number of the respective states (Si) set as the pre-transition state (Sb) in the table B1 stored in the storage means 122. Then, the probability (Pi) is set (S304). The probability (Pi) set in S304 may be weighted according to an instruction from the administrator.

  When S304 is executed, the control unit 121 adds the probability (Pi) corresponding to the pre-transition state (Sb) to each record of the table B2 stored in the storage unit 122 (S305). A table obtained by adding the holding time (Ti) and the probability (Pi) in S303 and S305 to the table B1 is a table B2 shown in FIG. The order of S302 and S303 and S304 and S305 may be interchanged.

  When S305 is executed, a flow for calculating a time value (Tij) and a probability value (Pij) between events (event (Ei) to event (Ej)) is executed (S306, S401). Assuming that the total number of extracted events (number of records in table B1 and table B2) is m, the analysis unit 12 will generate an event after an event Ei has occurred for integers i and j of 1 to m. A time value (Tij) that predicts the time until Ej occurs and a probability value (Pij) that predicts the probability that event (Ej) will occur after event (Ei) occurs are calculated.

  The control means 121 sets the number i indicating the event (Ei) to 0 (S402), and adds 1 to the number i (S403). It is determined whether the number i is less than or equal to m (S404). If the number i is larger (S404: NO), this flow ends (S413).

  When the number i is less than or equal to m (S404: YES), the control means 121 sets the number j to 0 (S405) and adds 1 to the number j (S406). The control means 121 determines whether the number j is less than or equal to m (S407). If the number j is greater than m (S407: NO), the process of S403 is executed.

  When the number j is less than or equal to m (S407: YES), the control unit 121 determines whether the number i is different from the number j (S408). When the number i and the number j are the same (S408: NO), the process of S406 is executed.

  When the number i is different from the number j (S408: YES), the flow for extracting the event (E) that occurs after the event (Ei) occurs until the event (Ej) occurs is executed (S409). When there are a plurality of event patterns that occur from the occurrence of the event (Ei) to the occurrence of the event (Ej), the events that pass for each pattern are extracted. Each extracted event pattern is used as route information.

  The control unit 121 reads out the holding time (T) corresponding to each event (E) that passes from the event (Ei) to the event (Ej) for each piece of route information extracted in S409, and calculates the calculation unit 125. Calculates the sum of the read retention times (T) for each route information extracted in S409, and sets the average of the sums for each route information as Tij (S410).

  Further, the control means 121 reads, from the table B1, the probability (P) corresponding to each event (E) that passes from the event (Ei) to the event (Ej) for each of the route information extracted in S409, The calculating unit 125 calculates the total of the read probabilities for each route information, and sets the sum of the totals of the respective route information as Pij (S411). S410 and S411 may be executed by switching the order.

  The control unit 121 stores the time value Tij calculated in S410 in the table B4 of the storage unit 13, and stores the probability value Pij calculated in S411 in the table B3 of the storage unit 13 (S412). When the flow of S412 is executed, the flow returns to the flow of S406.

  FIG. 20 shows a flow for extracting an event (E) that elapses from the occurrence of the event (Ei) to the occurrence of the event (Ej).

  When S409 is executed (S501), the control unit 121 stores the event ID (Ei) in the storage unit 122 as the first elapsed point of the route information between the event (Ei) and the event ID (Ej). (S503).

  Further, the control unit 121 acquires information indicating the pre-transition state (Sb) stored in the record including the event (Ej) of the table B1 in the storage unit 122, and sets this information as (Sy) (S504).

  Next, the control unit 121 determines whether there is any route information stored in the storage unit 122 that has not been determined (S505). If it is determined that there is no route information that has not been determined in S505 (S505: NO), this flow ends (S513).

  When there is route information that has not been confirmed in S505 (S505: YES), the control unit 121 selects any of the route information that has not been confirmed, and among the selected route information, an event ID indicating the last elapsed point. Is read (S506). The read event ID is (Ex).

  The control unit 121 reads information indicating the post-transition state (Sa) included in the same record as the event ID (Ex) read in S506 in the table B1 of the storage unit 122, and sets the read information as (Sx). (S507).

  Next, the control unit 121 determines whether the information (Sx) read in S507 and the information (Sy) acquired in S504 are different (S508). When the information (Sx) and the information (Sy) are the same (S508: NO), the event (Ej) is added to the route information (S511). In this route information, it is assumed that events occurring from the event (Ei) to the event (Ej) can be listed, and the route information is determined (S512). When S512 is executed, the process of S505 is executed again.

  When the information (Sx) is different from the information (Sy) (S508: YES), the control unit 121 refers to the table B1 stored in the storage unit 122 and sets the same information as the information (Sx) to the pre-transition state (Sb). The event ID of the record is extracted (S509).

  The extracted event ID (E) is added to the end of the route information selected in S505 (S510). When there are a plurality of event IDs extracted in S509, the control unit 121 duplicates the route information selected in S505, and adds each of the event IDs extracted at the end of each route information.

  FIG. 21 shows a flow for executing the processing of the instruction unit 14. When the flow starts, the control unit 141 determines whether a predetermined time has elapsed (S602) and whether notification data has been received (S603). When the control unit 141 detects that the predetermined time has elapsed (S602: YES), the control unit 141 clears the measurement timer for measuring the predetermined time and starts measuring the predetermined time again (S605). After the process of S605, a statistics / instruction flow (which will be described later with reference to FIG. 23) is started (S606).

  When the control unit 141 detects the reception of the notification data (S603: YES), the control unit 141 starts an accumulation flow (described later with reference to FIG. 22) (S604).

  FIG. 22 shows a flow for accumulating notification data received from the processing devices 2a to 2e.

  When S604 is executed (S701), the control unit 141 acquires the reception time (Hk) (S702). In addition, the control unit 141 acquires an event ID and a session ID from the data received by the communication unit 11 (S703). Further, the control unit 141 refers to the table B3 and the table B4 stored in the storage unit 13, and reads the time value (Tij) and the probability value (Pij) corresponding to the acquired event ID (Ei) (S704).

  When S704 is executed, the control unit 141 refers to the table B5 stored in the storage unit 142, and determines whether there is a record including the session ID acquired in S703 (S705). When there is no record including the session ID acquired in S703 in the table B5 (S705: NO), the time value (Tij) and probability value (Pij) acquired in S704, and the session ID and event ID acquired in S703 are displayed. The included record is added to the table B5 (S708). When the process of S708 is executed, the process of S602 is executed.

  When a record including the session ID acquired in S703 exists in the table B5 (S705: NO), the record including the acquired session ID includes the time value (Tij) and the probability value (Pij) acquired in S704, and S703. The record is updated to the record including the event ID acquired in (S706). When the process of S706 is executed, the process of S602 is executed.

  The processing of S705, S706, and S707 is an additional flow, which can improve the accuracy of statistics.

  When S606 is executed (S801), the control unit 141 acquires information indicating the time (H0) at that time (S802). Further, the number of records in the table B5 stored in the storage unit 142 is assumed to be s at time (H0).

  Next, the control unit 141 reads the time value (Tij) and the reception time (Hk) for each record in the table B5 of the storage unit 142 (for an integer r of 1 to s) (S806). Based on the time value (Tij) read out by the control unit 141, the reception time (Hk), and the time (H0), the calculation unit 144 satisfies the time value (Ur = Tij− (H0−Hk)). A value corresponding to (Ur) is calculated (S807).

  The control unit 141 reads the time range ID (R) of the time range including the calculated time value (Us) with reference to the time range shown in the table B6 stored in the storage unit 142 (S808). The control unit 141 stores the read time range (R) in each record of the table B5 (S809). The flow from S804 to S809 is repeatedly executed for an integer r of 1 to s.

  Next, the control unit 142 reads the probability value (Pij) and the time range (Rq) for each record of the table B5 stored in the storage unit 142 (for the integer r of 1 to s) (S813). The aggregation unit 145 performs a process of adding the probability value (Pij) read by the control unit 141 to the statistical value (Psumq) corresponding to the time range (Rq) read by the control unit 141 (SPsumq). The flow from S811 to S814 is repeatedly executed for the integer r of 1 to s.

  Next, the integer r is set to 0 again (S810), 1 is added to the integer r (S811), the magnitudes of the integer r and the integer s are compared (S812), and the integer r is greater than the integer s (S812: If NO), the process of S815 is executed.

  When the integer r is a number equal to or less than the integer s (S812: YES), the control unit 121 determines the probability value (Pij) and the time range ID (Rq) from the r-th record in the table B5 stored in the storage unit 142. ) Is read (S813). The read probability value (Pij) is added to the total value (Psumq) corresponding to the read time range ID (Rq) (S814).

  Next, the determination unit 146 determines the instruction content to the processing devices 2a to 2e according to the total value (Psumq) for the integer q of 1 to p and the operating status of the processing devices 2a to 2e (S815). The generation unit 147 generates instruction data to be transmitted to the processing device according to the instruction content determined by the determination unit 146 (S816). When S816 is executed, the process of S602 is executed (S817).

  An example of the determination performed by the determination unit 146 in S815 will be described with reference to FIG. In FIG. 24, let M be the processing amount that can be processed by one processing apparatus.

  For example, among the processing devices 2a to 2e, two processing devices are operating at the time (H0), and the standby processing takes 30 seconds until the two processing devices are in an operating state. Suppose that it is in a dormant state that takes 60 seconds for the processing device to become operational.

  In this case, assuming that the total value (Psum) calculated by the statistics / instruction flow shown in FIG. 23 is as shown in FIG. 24, it is expected that, for example, three or more processing devices are required after 30 to 60 seconds. For this reason, it is possible to take measures such as generating operation instructions for the two processing devices in the standby state. Furthermore, since it is expected that four or more processing devices will be required after 60 to 90 seconds, an operation instruction is generated for the processing device in the idle state at almost the same timing as the operation instruction to the processing device in the standby state. The correspondence such as is considered.

  The above description of the determination based on the total value (Psum) is merely an example, and other determinations can be made according to the likelihood of the processing amount.

According to the first embodiment described above, the processing amount can be estimated based on the contents defined in the source code (C1) executed by the processing device and the event notification received from the processing device.
[Second Embodiment]
In the second embodiment, a test operation period is provided, and a time value (Tij) and a probability (Pij) between events are determined based on a result of totaling event notifications received by the management apparatus 1 during the test operation period.

  Each of the analysis unit 12 and the storage unit 13 of the second embodiment may be a functional block similar to that of the first embodiment.

  FIG. 25 shows a functional block of the instruction unit 14 in the second embodiment. The instruction unit 14 in the second embodiment has a configuration in which a totaling unit 148 is added to the instruction unit 14 of the first embodiment. The counting unit 148 totals the event notification, creates a table B2 based on the counting result, and stores it in the storage unit 122. The event notification aggregation processing will be described with reference to the table B10 shown in FIG. 26 and the flow shown in FIG.

  The storage unit 142 further stores a table B10 shown in FIG. The table B10 has a record for each event ID (Ei), and includes fields for event ID, pre-transition state (Sb), post-transition state (Sa), number of occurrences, number of occurrences, time accumulation value, average time, and average probability. Have The values stored in the fields of the number of occurrences, the number of occurrences, the time accumulation value, the average time, and the average probability in the table B10 of FIG.

  In the event ID, pre-transition state (Sb), and post-transition state (Sa) fields, information having the same contents as B2 in FIG. 7 is stored. The number of hits indicates the number of times that the pre-transition state (Sb) stored in the same record is reached within the test period. The number of occurrences indicates the number of times the event indicated by the event ID stored in the same record has occurred within the test period. The time accumulated value indicates a value obtained by accumulating the time during which the pre-transition state (Sb) corresponding to the event is held before each event occurs, every time the event occurs.

  A value obtained by dividing the accumulated time value by the number of event occurrences is used as an average time, and is used as an index indicating an average of a period in which the pre-transition state (Sb) is held until the event occurs. Further, a value obtained by dividing the number of event occurrences by the number of corresponding Sb is used as an average probability, and is used as a probability that an event will occur from the state before transition (Sb).

  FIG. 27 shows a flow for calculating the average time and average probability based on the table B10 shown in FIG. The average time and average probability are calculated by executing the flow shown in FIG. 27 between the processing of S705 and the processing of S706 in the flow of FIG.

  First, before starting the test operation, the information of the table B2 is copied to the event ID, pre-transition state (Sb), and post-transition state (Sa) fields of the table B10. Furthermore, 1 is stored in the field of the number of times of each record.

  When it is determined YES in S705 (S901), the control unit 141 reads a record including the same session ID as the session ID included in the event notification from the table B5, and extracts the reception time (Hk) from the read table ( S902).

  Next, the counting unit 148 calculates the difference between the reception time included in the event notification and the reception time (Hk) extracted in S902 (S903). The counting unit 148 adds the difference calculated in S903 to the accumulated time value of the record included in the table B10 and including the same ID as the event notification (S904).

  Further, the counting unit 148 adds 1 to the occurrence number field of the record (updated in S904) that matches the event notification and the event ID among the records in the table B10 (S905). After the processing of S905, the counting unit 148 divides the number of occurrences of the record having the event notification and the event ID matching by the number of times corresponding to the record, and stores the calculated value in the average probability field (S906). Further, the counting unit 148 divides the cumulative time value of the record by the number of occurrences, and stores the calculated value in the average time field (S907).

  The aggregation unit 148 reads the post-transition state (Sa) of the record whose value has been updated in S904 to S907, and reads the record including the read state as the pre-transition state (Sb) from the table B10. The counting unit 148 adds 1 to the value of the corresponding number of times field of each read record (S908). After step S908, step S706 is executed (S909).

  After the management apparatus 1 executes the above-described flow during the test operation period, the aggregation unit 148 stores the average time stored in each record of the table B10 in the holding time (Ti) of the table B2, and stores each record in the table B10. Is stored in the probability (Pi) of the table B2. Thereafter, the analysis means 12 calculates the time value (Tij) and the probability value (Pij) in the same procedure as in the first embodiment based on the holding time (Ti) and the probability (Pi) stored in the table B2. calculate.

  As in the second embodiment, not only the result of analyzing the program but also the statistics of event notification during operation are used to reflect the tendency of clients accessing the processing device group in the prediction of the processing amount. I can do it.

  The processing performed by the counting unit 148 can be realized by the apparatus shown in FIG. The processor 1a stores the table B2, the table B5, and the table B10 in the memory 1b or the storage device 1d, reads the information stored in the table B2, the table B5, and the table B10, and performs an operation according to the program on the read information. The processing performed by the counting means 148 is realized by storing information in the tables B2 and B10.

  Further, the time value (Tij) and the probability value (Pij) between events are set and operated in advance, and the time value (Tij) and the probability value (Pij) are determined based on the total result of the event notification received during the operation. May be updated. Even when the time and probability between events change during the period in which the embodiment is operated, the accuracy of estimation is expected to be improved by updating the time value (Tij) and the probability value (Pij).

  The embodiment of the present invention is not limited to the above-described embodiment, and may take other embodiments.

DESCRIPTION OF SYMBOLS 1 Management apparatus 2a-2e Processing apparatus 3a-3e Storage apparatus 4 Load distribution apparatus 5,6 Network 7 Terminal apparatus 11 Communication means 12 Analysis means 13 Storage means 14 Instruction means 121 Control means 122 Storage means 123 Extraction means 124 Generation means 125 Calculation Means 141 Control means 142 Storage means 143 Generation means 144 Calculation means 145 Aggregation means 146 Determination means 147 Generation means 1a Processor 1b Memory 1c Communication interface 1d Storage device 1e Bus 148 Aggregation means

Claims (14)

Based on an order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to the request, the first process of the plurality of types of processes is executed. A predicted time until the execution of the second process among the plurality of types of processes and the probability of executing the second process after executing the first process ,
In addition to the execution of the first process, a notification code that specifies notification of an identifier for identifying the first process is added to the source code,
Storing the calculated predicted time and the probability in the storage unit in association with the identifier;
Detecting that the first process has been executed by obtaining the notification of the identifier performed based on the source code to which the notification code is added;
When it is detected that the first process has been executed, the prediction time and the probability stored in the storage unit associated with the same identifier as the identifier included in the notification are read out,
A prediction program for causing a computer to predict that the second process is executed with the read probability after the read prediction time . In the computer,
Obtaining the source code from a processing device that performs processing based on the source code;
Calculate the predicted time based on the obtained source code.
The prediction program according to claim 1, wherein: In the computer,
According to a prediction result that the second process is executed, the process can be executed based on the source code, and an operation instruction is output to a waiting processing apparatus.
The prediction program according to claim 1, wherein the prediction program is executed. In the computer,
According to the prediction result that the second process is executed, the process can be executed based on the source code, and a standby instruction is output to a processing apparatus that is in operation.
The prediction program according to claim 1, wherein the prediction program is executed. In the computer,
When a plurality of notifications are acquired, each of notifications in which the estimated time has elapsed since the notification was acquired is included in a predetermined time range, and stored in the storage unit in association with the identifier included in each notification. Read the probability
Predicting that the second process is executed in the predetermined time range with the probability of adding the read probabilities
The prediction program according to claim 1, wherein the prediction program is executed. In the computer,
Adding to the source code a code specifying that notification of an identifier for identifying each of the plurality of types of processing is performed in conjunction with execution of each processing;
For each of the plurality of types of processes, the estimated time from the execution of each process to the execution of the second process and the probability of executing the second process after executing each process are calculated. And
Storing the calculated predicted time and probability for each process in the storage means in association with an identifier for identifying each process.
The prediction program according to claim 1, wherein the prediction program is executed. The notification further includes an access identifier that identifies an access that has requested a process identified by the identifier included in the notification;
In the computer,
The access identifier included in the notification is stored in the second storage unit in association with the identifier included in the notification and the predicted time and the probability stored in the storage unit in association with the identifier included in the notification. ,
In the case where there are a plurality of prediction times and probabilities associated with the same access identifier in the prediction times and probabilities stored in the second storage means, the prediction time associated with the access identifier included in the last acquired notification and Aggregate multiple prediction times and probabilities into probabilities,
Predicting that the second process is executed based on the prediction time and the probability stored in the second storage unit.
The prediction program according to claim 1, wherein the prediction program is executed. In the computer,
When there are a plurality of prediction times and probabilities associated with the same access identifier in the prediction times and probabilities stored in the second storage means, the difference between the prediction times associated with the same identifier and the association with the same identifier Storing the probability included in the notification acquired earlier in the third storage means,
Based on the difference between the time values stored in the third storage means, each of the plurality of types of processing is executed, and then another processing of the plurality of types of processing is executed. Calculate the average time to
Based on the probabilities stored in the third storage means, for each of the plurality of types of processing, a transition probability in which another processing among the plurality of types of processing is executed after each processing is executed is calculated. And
Updating the time value to the calculated average and updating the probability to the calculated transition probability.
The prediction program according to claim 7, wherein: The predicted time is calculated according to the number of processes to be executed between the execution of the first process and the execution of the second process.
The prediction program according to claim 1, wherein: Calculating the predicted time based on design information related to creation of the source code.
The prediction program according to claim 1, wherein: The probability is the number of processes executed between the execution of the first process and the execution of the second process, and the execution of the second process after the execution of the first process. For each of the processes to be executed in accordance with the number of branches defined in the source code
The prediction program according to claim 1, wherein: Based on an order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to the request, the first process of the plurality of types of processes is executed. A predicted time until the execution of the second process among the plurality of types of processes and the probability of executing the second process after executing the first process,
In addition to the execution of the first process, a notification code that specifies notification of an identifier for identifying the first process is added to the source code,
Storing the calculated predicted time and the probability in the storage unit in association with the identifier;
Detecting that the first process has been executed by obtaining the notification of the identifier performed based on the source code to which the notification code is added;
When it is detected that the first process has been executed, the prediction time and the probability stored in the storage unit associated with the same identifier as the identifier included in the notification are read out,
Predicting that the second process is executed with the read probability after the read prediction time.
The prediction method characterized by including. Based on an order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes executed in response to the request, the first process of the plurality of types of processes is executed. Calculating means for calculating a predicted time until execution of the second process among the plurality of types of processes and a probability of executing the second process after executing the first process;
Adding means for adding to the source code a notification code that prescribes notification of an identifier for identifying the first process in conjunction with the execution of the first process;
Storage means for storing the calculated predicted time and the probability in association with the identifier,
Detecting means for detecting that the first process has been executed by acquiring the notification of the identifier performed based on the source code to which the notification code is added;
A reading means for reading out the predicted time and the probability stored in the storage means in association with the same identifier as the identifier included in the notification when it is detected that the first process has been executed;
Prediction means for predicting that the second process is executed with the read probability after the read prediction time;
The prediction apparatus characterized by including. A prediction system comprising a first device in operation and a second device in standby, and a third device capable of communicating with the first device and the second device,
The third device includes:
Based on the order relationship between the processes of the plurality of types of processes defined in the source code that defines a plurality of types of processes to be executed in response to the request on the first device, The predicted time from the execution of the first process to the execution of the second process among the plurality of types of processes and the probability of executing the second process after the execution of the first process are calculated. A calculation means;
Adding means for adding to the source code a notification code that prescribes notification of an identifier for identifying the first process in conjunction with the execution of the first process;
Storage means for storing the calculated predicted time and the probability in association with the identifier,
Detecting means for detecting that the first process has been executed by acquiring the notification of the identifier performed based on the source code to which the notification code is added;
A reading means for reading out the predicted time and the probability stored in the storage means in association with the same identifier as the identifier included in the notification when it is detected that the first process has been executed;
Prediction means for predicting that the second process is executed with the read probability after the read prediction time;
Output means for outputting an operation instruction to the second device in accordance with a prediction result of the prediction means;
The prediction system characterized by including.

Images