JP5440937B2 - Thread number limiting device, thread number limiting method, and thread number limiting program - Google Patents

Description

  The present invention relates to a thread number limiting device, a thread number limiting method, and a thread number limiting program.

  Multithreading is a technique for efficiently executing computer processing. Multi-threading is a technique for generating a plurality of threads with one program and processing the program in parallel with the plurality of threads. A process having such a multi-thread function can improve the processing efficiency by generating a plurality of threads. However, if a thread is generated more than necessary, the processing efficiency decreases due to insufficient memory capacity. Patent Document 1 below discloses a system in which the number of threads is increased when the number of requests waiting for processing in a process is increasing, and the number of threads is decreased when the number of requests waiting for processing is decreasing.

Japanese Patent Laid-Open No. 3-40034

  By the way, the memory capacity allocated to the process when the process is executed varies depending on the resource status (for example, the execution status of other processes) at that time. If the memory capacity that can be used in the process varies, the number of threads that can be created and the processing speed also vary. Therefore, it cannot be said that the processing for each process is optimally controlled simply by increasing or decreasing the number of threads according to the number of requests waiting for processing in the process as in Patent Document 1.

  The present invention has been made to solve the above-described problem, and is capable of efficiently improving the processing for each process according to the resource status, the thread number limiting method, and the thread number limiting program. The purpose is to provide.

  The thread number limiting device according to the present invention includes a first acquisition unit that acquires a first memory capacity that can be used in a process that controls execution of an application, and a first request that corresponds to the first request. A second acquisition unit for acquiring a second memory capacity used by the process during application execution control, and a second application execution control corresponding to the second request received after receiving the first request A third acquisition unit for acquiring a third memory capacity used by the process at times, a fourth acquisition unit for acquiring the number of threads generated by execution of the second application, and the second memory capacity And a first calculation unit that calculates a memory capacity per request using the third memory capacity, and the acquired number of threads and A second calculation unit that calculates the memory capacity per thread using the memory capacity per request, the first memory capacity, the second memory capacity, and the memory capacity per thread. An upper limit value setting unit that calculates and sets an upper limit value of the number of threads in the process.

  According to the thread number limiting method of the present invention, when a process for controlling execution of an application is started, a step of acquiring a first memory capacity usable in the process, and execution control of the first application corresponding to the first request Acquiring a second memory capacity used by the process at times, and a third used by the process during execution control of a second application corresponding to a second request received after receiving the first request A memory capacity per request using the second memory capacity and the third memory capacity, a step of acquiring the number of threads generated by executing the second application; Calculating the number of threads acquired and the number of requests per request. Calculating a memory capacity per thread using a plurality of memory capacity, and using the first memory capacity, the second memory capacity, and the memory capacity per thread, the number of threads in the process Calculating and setting an upper limit value.

  The thread number limiting program of the present invention causes a computer to execute each step included in the thread number limiting method.

  According to the present invention, it is possible to efficiently improve processing for each process according to resource conditions.

It is a block diagram which illustrates functional composition of a thread number limiting device in an embodiment. It is a flowchart which illustrates the process sequence at the time of setting the upper limit of the number of threads. It is a flowchart which illustrates the process sequence at the time of setting the lower limit of the number of threads. It is a figure which illustrates the transition state of the number of threads in the conventional computer apparatus. It is a figure which illustrates the transition state of the number of threads in the thread number restriction | limiting apparatus of embodiment. It is a figure which illustrates the number of excess threads and the number of insufficient threads in the conventional computer apparatus and the thread number limiting device of the embodiment.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of a thread number limiting device, a thread number limiting method, and a thread number limiting program according to the present invention will be described with reference to the accompanying drawings. Here, the thread number limiting device is a device capable of executing a multi-thread program. A multi-thread program is a program written to execute processing in multi-threads.

  With reference to FIG. 1, a functional configuration of the thread number limiting device in the embodiment will be described. The thread number limiting device 1 includes a request transmission unit 11, a thread control unit 12, a memory capacity monitoring unit 13, a thread number monitoring unit 14, a queue monitoring unit 15, and a limit value setting unit 16.

  Here, the thread number limiting device 1 physically includes, for example, a CPU (Central Processing Unit), a storage device, and an input / output interface. These elements are connected to each other via a bus. The storage device includes, for example, a ROM (Read Only Memory) and HDD (Hard Disk Drive) that store programs and data processed by the CPU, and a RAM (Random Access Memory) mainly used as various work areas for control processing. ; Memory) and the like. The CPU executes programs stored in the ROM and processes various data expanded in the RAM in accordance with various messages received via the input / output interface. Thereby, the function of each part of the thread number limiting device 1 described in detail below can be realized.

  The request transmission unit 11 issues and transmits a request addressed to its own device 1. The request includes an application execution request. The request transmission unit 11 issues a request for requesting execution of an initial activation application as the first request immediately after the process activation. The request transmission unit 11 issues a request for requesting the execution of the multithread program as the second request immediately after the process is started. The process is generated by executing the program, and executes and controls applications included in the program.

  By providing the request transmission unit 11, it is possible to set an upper limit value of the number of threads to be described later based on a request issued by itself after starting a process until processing a request from a client. Become.

  The thread control unit 12 executes a multi-thread program in response to a request, and controls the number of threads by generating a thread in the process or deleting a thread. The thread control unit 12 limits the number of threads in the process according to an upper limit value or a lower limit value of the number of threads described later.

  When the process is activated, the memory capacity monitoring unit 13 acquires the memory capacity that can be used by the process to be activated and stores it in the monitoring data file. The memory capacity monitoring unit 13 acquires the memory capacity used by the process when executing the application corresponding to the request, and stores the acquired memory capacity in the monitoring data file.

  When the application corresponding to the request is executed, the thread number monitoring unit 14 acquires the number of threads generated in the process by the execution of the application, and stores it in the monitoring data file.

  The thread number monitoring unit 14 collects the number of operating state threads generated in the process at predetermined sampling intervals after the process is started, and accumulates it in the monitoring data file as time-series operating state thread number data. . The predetermined sampling interval can be arbitrarily set by obtaining an interval suitable for changing the setting of the lower limit value of the number of threads in accordance with increase / decrease in the number of requests by experiments or the like. The predetermined sampling interval may be set from a client terminal that transmits a request.

  After the process is started, the queue monitoring unit 15 collects the number of requests waiting for processing at the predetermined sampling interval and accumulates it in the monitoring data file as time-series processing wait state request count data.

  The limit value setting unit 16 calculates an upper limit value and a lower limit value of the number of threads, and sets the upper limit value parameter and the lower limit value parameter of the limit value parameter file, respectively.

  The limit value setting unit 16 calculates the upper limit value of the number of threads, for example, according to the following procedure, and sets the upper limit value parameter.

  First, the limit value setting unit 16 calculates the memory capacity per request using the memory capacity acquired by the memory capacity monitoring unit 13. The memory capacity per request can be calculated using, for example, the following formula 1.

  MR = M2-M1 Equation 1

  MR in Equation 1 is the memory capacity per request, M2 is the memory capacity used by the process when executing the application corresponding to the second request, and M1 is the memory capacity of the application corresponding to the first request. The amount of memory used by the process during execution.

  Subsequently, the limit value setting unit 16 uses the number of threads acquired by the thread number monitoring unit 14 and the memory capacity MR per request to calculate the memory capacity per thread. The memory capacity per thread can be calculated using, for example, the following formula 2. Note that the number of threads and the number of requests are in a proportional relationship, and the number of threads is an integral multiple of the number of requests.

  MT = MR / T Equation 2

  MT in Equation 2 is the memory capacity per thread, MR is the memory capacity per request, and T is the number of threads generated by executing the application corresponding to the second request.

  Subsequently, the limit value setting unit 16 uses the memory capacity acquired by the memory capacity monitoring unit 13 and the memory capacity MT per thread to calculate the upper limit value of the number of threads in the process. The upper limit value of the number of threads can be calculated using, for example, the following formula 3.

  Tmax = {(Mmax−M1) / MT} + 1 + Equation 3

  Tmax in Equation 3 is an upper limit value of the number of threads, and Mmax is a memory capacity usable in the process. M1 is the memory capacity used by the process when executing the application corresponding to the first request, and MT is the memory capacity per thread. Note that 1 is added in the above equation 3 because the upper limit value Tmax of the number of threads is calculated based on the memory capacity obtained by subtracting M1 from Mmax, so that one thread that can be generated in the memory capacity of M1 It is the sum of minutes.

  Subsequently, the limit value setting unit 16 sets the calculated upper limit value Tmax of the number of threads in the upper limit value parameter of the limit value parameter file.

  For example, the limit value setting unit 16 calculates the lower limit value of the number of threads and sets it as the lower limit parameter according to the following procedure.

  First, the limit value setting unit 16 receives the time-series processing waiting state request count data collected by the queue monitoring unit 15 during the most recent predetermined time each time a predetermined time elapses after the process starts. Time-series thread number data is generated using the time-series active state thread number data collected by the thread number monitoring unit 14.

  The predetermined time can be arbitrarily set by obtaining an appropriate time by experiment or the like to change the setting of the lower limit value of the number of threads according to the increase or decrease of the number of requests. However, if the predetermined time is too short, it is likely to be affected by sudden changes in the number of requests, and if the predetermined time is too long, the fluctuations in the number of requests are averaged, and the lower limit of the number of threads according to the increase or decrease in the number of requests. It becomes difficult to set. By adjusting this predetermined time and the above-described predetermined sampling interval, it is possible to adjust the degree to which the increase / decrease in the number of requests is reflected in the lower limit value of the number of threads.

  Here, the time-series thread number data is converted, for example, from the processing waiting state request number data into thread number data, and the converted thread number data and the operating state thread number data are added at the same sampling time. It can be generated by doing.

  Subsequently, the limit value setting unit 16 calculates a lower limit value of the number of threads in the process using the generated time-series thread number data. The lower limit value of the number of threads can be calculated using, for example, the following formula 4.

  Tmin = (W1 * X1 + ... + Wn * Xn) / (W1 + ... + Wn) ... Equation 4

  The above equation 4 is an equation for calculating a weighted moving average. Tmin in Equation 4 is a lower limit value of the number of threads, X1 to Xn are time-series thread number data, and W1 to Wn are weights added to the thread number data X1 to Xn, respectively. The weights W1 to Wn increase as the sampling times of the thread number data X1 to Xn are closer to the present time. Thereby, it becomes possible to calculate the lower limit according to the current situation.

  Subsequently, when the lower limit value Tmin of the calculated thread number is larger than the upper limit value Tmax of the thread number set in the upper limit parameter, the limit value setting unit 16 sets the lower limit value Tmin of the calculated thread number. The upper limit value Tmax of the number of threads set in the upper limit value parameter is set.

  Subsequently, the limit value setting unit 16 sets the lower limit value Tmin in which the lower limit value Tmin or the upper limit value Tmax of the calculated number of threads is set as the lower limit value parameter of the limit value parameter file.

  Next, the operation of the thread number limiting device in this embodiment will be described with reference to the drawings. With reference to FIG. 2, the operation when setting the upper limit of the number of threads will be described. FIG. 2 is a flowchart illustrating a processing procedure when setting the upper limit value of the number of threads.

  First, when a process is activated (step S101), the memory capacity monitoring unit 13 acquires a memory capacity Mmax that can be used by the activated process (step S102).

  Subsequently, the request transmission unit 11 issues and transmits a first request addressed to the own apparatus 1 (step S103). The thread limit device 1 receives the first request and executes the initial activation application corresponding to the first request. Thereby, an object at the time of initial activation is generated, but a thread is not generated.

  Subsequently, the memory capacity monitoring unit 13 acquires the memory capacity M1 used by the process when the application for initial activation is executed (step S104).

  Subsequently, the request transmission unit 11 issues and transmits a second request addressed to the own apparatus 1 (step S105). The thread limit device 1 receives the second request and executes the multi-thread program as an application corresponding to the second request. As a result, a thread is generated.

  Subsequently, the memory capacity monitoring unit 13 acquires the memory capacity M2 used by the process when the multi-thread program corresponding to the second request is executed (step S106), and the thread number monitoring unit 14 acquires the second request. The number of threads T generated by executing the multithread program corresponding to the request is acquired (step S107).

  Subsequently, the limit value setting unit 16 calculates the memory capacity MR per request by substituting the memory capacity M1 acquired in step S104 and the memory capacity M2 acquired in step S106 into the above equation 1 (step S104). S108).

  Subsequently, the limit value setting unit 16 substitutes the number of threads T acquired in step S107 and the memory capacity MR per request calculated in step S108 into the above equation 2, and calculates the memory capacity MT per thread. Calculate (step S109).

  Subsequently, the limit value setting unit 16 substitutes the memory capacity Mmax acquired in step S102, the memory capacity M1 acquired in step S104, and the memory capacity MT per thread calculated in step S108 into the above equation 3. Thus, the upper limit value Tmax of the number of threads in the process is calculated (step S110).

  Subsequently, the limit value setting unit 16 sets the upper limit value Tmax of the number of threads calculated in step S110 as an upper limit parameter (step S111).

  With reference to FIG. 3, the operation when setting the lower limit of the number of threads will be described. FIG. 3 is a flowchart illustrating a processing procedure when setting the lower limit value of the number of threads.

  First, it is determined whether or not a predetermined time has elapsed since the process was started (step S201). When this determination is YES (step S201; YES), the limit value setting unit 16 acquires time-series processing waiting state request number data collected by the queue monitoring unit 15 (step S202), and monitors the number of threads. The time-series operation state thread number data collected by the unit 14 is acquired (step S203).

  Subsequently, the limit value setting unit 16 generates time-series thread number data X1 to Xn using the processing wait state request number data acquired in step S202 and the operating state thread number data acquired in step S203. (Step S204).

  Subsequently, the limit value setting unit 16 calculates the lower limit value Tmin of the number of threads in the process by substituting the time-series thread number data X1 to Xn generated in step S204 into the equation 4 (step S205). .

  Subsequently, the limit value setting unit 16 determines whether or not the lower limit value Tmin of the number of threads calculated in step S205 is larger than the upper limit value Tmax of the number of threads set in the upper limit parameter (step S206). ). If this determination is NO (step S206; NO), the process proceeds to step S208 described later.

  On the other hand, if it is determined in step S206 that the lower limit value Tmin of the number of threads is larger than the upper limit value Tmax of the number of threads (step S206; YES), the limit value setting unit 16 calculates the thread calculated in step S205. The upper limit value Tmax of the number of threads set in the upper limit parameter is set to the lower limit value Tmin of the number (step S207).

  Subsequently, the limit value setting unit 16 sets the lower limit value Tmin of the number of threads calculated in step S205 or the lower limit value Tmin of the number of threads set with the upper limit value Tmax in step S207 as the lower limit value parameter (step S208). .

  As described above, according to the thread number limiting device 1 in the present embodiment, the memory capacity MT per thread is calculated based on the memory capacity MR per request, and the memory capacity MT per thread and each process are calculated. The upper limit value Tmax of the number of threads can be calculated based on the memory capacity Mmax that can be used for the above, and the calculated upper limit value Tmax of the number of threads can be set as the upper limit value parameter. As a result, the number of threads operating for each process can be limited to be equal to or less than the upper limit value Tmax of the number of threads, so that wasteful consumption of memory capacity can be suppressed for each process, and resources can be effectively used. Can be used.

  Further, according to the thread number limiting device 1 in the present embodiment, the time series thread number data X1 to Xn are generated based on the collected time series processing waiting state request number data and time series operation state thread number data. The lower limit value Tmin of the thread number can be calculated based on the time-series thread number data X1 to Xn, and the calculated lower limit value Tmin of the thread number can be set as the lower limit value parameter. As a result, the number of threads operating for each process can be limited so as to be equal to or greater than the lower limit value Tmin of the number of threads, so the minimum necessary number of threads can be secured according to the processing status of each process. The processing time can be shortened.

  Here, with reference to FIG. 4 to FIG. 6, the difference in processing efficiency between the conventional computer device that fixes the number of threads and the thread number limiting device 1 of the present embodiment will be specifically described. The data indicating the transition of the number of active threads shown in FIGS. 4 and 5 is simulated data in which the number of requests reaches a peak around 6:00. The simulated data shown in FIGS. 4 and 5 is data when operating in one request and one thread. Therefore, the transition of the number of operating threads can be replaced with the transition of the number of requests.

  FIG. 4 is a diagram showing a transition state of the number of threads in a conventional computer apparatus. In the conventional computer apparatus shown in FIG. 4, the past number of threads is fixed to “5”, which is an average value. In contrast, the number of requests (number of simulated data threads) varies greatly between “0” and “15” depending on the time zone.

  FIG. 5 is a diagram illustrating a transition state of the number of threads in the thread number limiting device 1 according to the present embodiment. As described above, the thread number limiting device 1 of this embodiment shown in FIG. 5 sets the upper limit value of the thread number according to the resource status for each process, and sets the lower limit value of the thread number according to the request number. ing. Thereby, the thread number limiting device 1 of the present embodiment increases or decreases the number of threads following the number of requests (number of simulated data threads).

  FIG. 6 is a diagram showing the number of excess threads and the number of insufficient threads respectively generated in the conventional computer device and the thread number limiting device 1 of the present embodiment as a result of executing the processing in the state shown in FIGS. 4 and 5. . As shown in FIG. 6, in the conventional computer device, the number of excess threads and the number of insufficient threads are both over 120, whereas in the thread number limiting device 1 of this embodiment, the number of excess threads is slightly less than 40, The numbers have been reduced to over 70 respectively. By reducing the number of excess threads, it is possible to reduce unnecessary use of memory capacity, and it is possible to shorten the processing time of requests by reducing the number of insufficient threads. That is, the thread number limiting device 1 of this embodiment can improve the processing efficiency for each process as compared with a conventional computer device.

  Note that the above-described embodiment is merely an example, and does not exclude various modifications and technical applications that are not explicitly described in the embodiment. That is, the present invention can be implemented by being modified into various forms without departing from the spirit of the present invention.

  For example, when the lower limit value of the number of threads is calculated, the limit value setting unit 16 in the above-described embodiment generates time series thread number data, and the lower limit value of the thread number based on the time series thread number data. However, the present invention is not limited to this. For example, time-series request number data may be generated, and a lower limit value of the number of threads may be calculated based on the time-series request number data. In this case, the number of active state threads is converted into the number of requests, and the request count data after conversion and the number of requests waiting to be processed are added at the same sampling time, so time-series requests Numerical data can be generated. Also, the generated time-series request count data is converted into time-series thread count data, and the converted time-series thread count data is substituted into the above equation 4 to calculate the lower limit value of the thread count. be able to.

  DESCRIPTION OF SYMBOLS 1 ... Thread number limiting apparatus, 11 ... Request transmission part, 12 ... Thread control part, 13 ... Memory capacity monitoring part, 14 ... Thread number monitoring part, 15 ... Queue monitoring part, 16 ... Limit value setting part

Claims (7)

A first acquisition unit that acquires a first memory capacity that can be used by the process when starting a process that controls the execution of the application;
A second acquisition unit that acquires a second memory capacity used by the process during execution control of the first application corresponding to the first request;
A third acquisition unit for acquiring a third memory capacity used by the process during execution control of a second application corresponding to a second request received after receiving the first request;
A fourth acquisition unit that acquires the number of threads generated by execution of the second application;
A first calculator that calculates a memory capacity per request using the second memory capacity and the third memory capacity;
A second calculation unit that calculates a memory capacity per thread using the acquired number of threads and the memory capacity per request;
An upper limit setting unit configured to calculate and set an upper limit of the number of threads in the process using the first memory capacity, the second memory capacity, and the memory capacity per thread;
A device for limiting the number of threads.   The thread number limiting device according to claim 1, further comprising a thread number upper limit control unit that limits the number of threads according to the upper limit value of the number of threads set by the upper limit value setting unit. A collection unit that collects the number of requests waiting to be processed and the number of threads in the active state at predetermined sampling intervals after the process is started,
A generation unit that generates time-series request count data using the number of requests waiting to be processed and the number of threads in the operation state collected by the collection unit at predetermined time intervals after starting the process;
A lower limit setting unit configured to calculate and set a lower limit value of the number of threads in the process using the generated time-series request number data;
The thread number limiting device according to claim 1, further comprising:   The thread number limiting device according to claim 3, further comprising a thread number lower limit control unit that limits the number of threads according to a lower limit value of the number of threads set by the lower limit value setting unit.   The thread number limiting device according to any one of claims 1 to 4, further comprising a transmission unit that transmits the first request and the second request. Obtaining a first memory capacity that can be used by the process at the time of starting a process that controls execution of the application;
Obtaining a second memory capacity used by the process during execution control of the first application corresponding to the first request;
Obtaining a third memory capacity used by the process during execution control of a second application corresponding to a second request received after receiving the first request;
Obtaining the number of threads generated by execution of the second application;
Calculating a memory capacity per request using the second memory capacity and the third memory capacity;
Calculating the memory capacity per thread using the acquired number of threads and the memory capacity per request;
Calculating and setting an upper limit value of the number of threads in the process using the first memory capacity, the second memory capacity, and the memory capacity per thread;
A method for limiting the number of threads.   A thread number limiting program for causing a computer to execute each step according to claim 6.