JP5472626B2 - Information processing apparatus operation management apparatus, cloud computing system, operation management program, storage medium, and information processing apparatus operation management method - Google Patents

Description

  The present invention relates to an information processing device operation management device, a cloud computing system, an operation management program, a storage medium, and an information processing device that control the number of operating information processing devices connected to a plurality of terminal devices via a network. The operation management method.

  Conventionally, users of companies, individuals, etc. own and manage computer hardware, software, data, etc. themselves. On the other hand, in recent years, companies that provide various services have equipment such as servers for processing, while users only have terminal devices that can connect to the Internet, and connect terminal devices to the Internet. The form of receiving necessary services from service providers is increasing.

  In such a cloud computing system, in order to cope with the peak of the number of used accesses, a plurality of servers are always activated to distribute the load.

  By the way, conventionally, various methods have been proposed in order to distribute the load on the server (see, for example, Patent Document 1). However, in cloud computing, since the service contents and usage purposes are various, it is difficult to identify changes in the usage access amount, so it is difficult to predict the number of servers.

  Therefore, the fixed number of servers is operated based on the maximum value of the access amount, or the approximate number of servers is operated based on the large features of the access amount (daytime and nighttime, weekdays and holidays).

JP 2005-141605 A

  However, in the above-described conventional operation management, in order to keep a necessary number of servers in operation when necessary, it is necessary to operate at the predicted maximum number of operations at that time, and it is difficult to obtain an energy saving effect. Accordingly, the present invention has been made in view of the above circumstances, and an information processing apparatus operation management apparatus and cloud computing capable of further enhancing the energy saving effect by operating only a minimum necessary server at a necessary time. An object is to provide a system, an operation management program, a storage medium, and an information processing apparatus operation management method.

The present invention employs the following means in order to solve the above problems.
An information processing apparatus operation management apparatus according to the present invention is an information processing apparatus operation management apparatus that controls the number of operating information processing apparatuses connected to a plurality of terminal apparatuses via a network, and changes in time series An acquisition unit that acquires management target data of the information processing apparatus and stores it in a predetermined storage unit, and creates a data string that takes out the management target data in a past predetermined time from the storage unit and connects them in time series An alignment unit; a short-term prediction unit that calculates a short-term prediction value of the management target data after a predetermined time based on the data string; and the management target at a future point in time than the fixed time based on the data string A long-term time prediction unit for calculating a long-term prediction value of data; and comparing the current value of the management target data with the short-term prediction value and the long-term prediction value, An estimation unit that calculates the required number of operating units after or a number of units to be added or stopped, and a command unit that starts or stops the information processing apparatus so that the number of operating units after the predetermined time becomes the required operating number, It is characterized by having.

  The cloud computing system according to the present invention includes a plurality of information processing devices connected to a plurality of terminal devices via a network, and the information processing device operation management device according to the present invention. And

  The operation management program according to the present invention causes a computer to function as the information processing apparatus operation management apparatus according to the present invention.

  A computer-readable storage medium according to the present invention stores the operation management program according to the present invention.

An information processing apparatus operation management method according to the present invention is an information processing apparatus operation management method for controlling the number of operating information processing apparatuses connected to a plurality of terminal apparatuses via a network, in a time series. An acquisition step of acquiring management target data of the information processing apparatus that changes;
An alignment step of creating a data string by linking the acquired management object data within a predetermined past time in a time series, and a short term for calculating a short-term prediction value of the management object data after a certain time based on the data string A time prediction step, a long-term time prediction step for calculating a long-term prediction value of the management target data at a future time point based on the data string, a current value of the management target data, and the short-term prediction value And an estimation step of calculating the number of required operating units of the information processing apparatus after the predetermined time, or the number of units to be added or stopped, and the operating number after the predetermined time And a command step for starting or stopping the information processing apparatus so that the number of the information processing apparatuses becomes the number.

  According to the present invention, it is possible to further enhance the energy saving effect by operating only the minimum necessary server at a necessary time.

1 is a schematic diagram illustrating a cloud computing system and an information processing apparatus operation management apparatus according to a first embodiment of the present invention. It is a functional block diagram which shows the information processing apparatus operation management apparatus which concerns on the 1st Embodiment of this invention. It is a flowchart which shows the information processing apparatus operation management method which concerns on the 1st Embodiment of this invention. It is a functional block diagram which shows the information processing apparatus operation management apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the information processing apparatus operation management method which concerns on the 2nd Embodiment of this invention.

(First embodiment)
A first embodiment according to the present invention will be described with reference to FIGS.
As shown in FIG. 1, the cloud computing system 1 according to the present embodiment includes a plurality of application servers (information processing devices) 3 connected to a hub 2, a switchboard 4 that supplies power to these, and an application server 3. And a server operation management apparatus (information processing apparatus operation management apparatus) 5 for controlling start / stop of the server. The cloud computing system 1 is connected to a plurality of terminal devices T via a network N.

  The server operation management device 5 is a computer system including at least a processor 6, a memory 7, a medium reading device 8, and an input / output interface 9 as a hardware configuration. Each unit is connected to be communicable with each other via the bus 10.

  The processor 6 controls each unit of the server operation management device 5 based on an operating system stored in a ROM or hard disk (not shown) or a program read from the memory 7.

  The memory 7 is used for work of the processor 6 while programs and data are written as necessary.

  The medium reader 8 accepts a computer-readable information storage medium such as a CD-ROM or DVD-ROM and reads a program stored on the medium. The program read in this way is executed by the processor 6.

  As shown in FIG. 2, the server operation management apparatus 5 includes an acquisition unit 11, an alignment unit 12, a short-term time prediction unit 13, a long-term time prediction unit 15, as functional units (program modules) of the operation management program. An estimation unit 16, a calculation unit 17, and a command unit 18 are provided. These functions are controlled by the control unit 20 realized by the processor 6 and the memory 7.

  In each of these functions, an operation management program stored in a computer-readable information storage medium is read into the server operation management apparatus 5 by using the medium reading apparatus 8, and the processor 6 of the server operation management apparatus 5 is mainly used. It is realized by hardware.

  Here, the operation management program is supplied to the server operation management apparatus 5 by the information storage medium, but may be downloaded from a remote place via a data communication network such as the Internet.

  The acquisition unit 11 acquires the CPU load (management target data) Pt of each application server 3 that changes in time series, for example, every minute, and stores it in the CPU load history database (storage unit) 21. The CPU load history database 21 may be provided as a part of the memory.

The aligning unit 12 extracts the CPU load in the past predetermined time (n minutes) from the CPU load history database 21 and creates a data string (P _{t−1 to} P _n ) linked in time series.

  The short-term time prediction unit 13 calculates a short-term predicted value of the CPU load after a certain time (for example, after five minutes as a time required for starting the server) based on the created data string.

  The long-term time predicting unit 15 calculates a long-term predicted value of the CPU load at a future time point, for example, 10 minutes after the current time, based on the created data string, separately from the short-term time predicting unit 13.

  The estimating unit 16 compares the current value of the CPU load with the short-term predicted value and the long-term predicted value, and calculates the required number of operating servers after a certain time of the application server 3.

  The calculation unit 17 calculates the number of application servers 3 to be added or stopped after a certain period of time by subtracting the current operation number from the required operation number.

  The command unit 18 starts or stops the application servers 3 for the calculated number.

  Next, a server operation management method (information processing apparatus operation management method) according to the present embodiment will be described.

  As shown in FIG. 3, the server operation management method includes an acquisition step (S11), an alignment step (S12), a short-term time prediction step (S13), a long-term time prediction step (S14), and an estimation step. (S15), a calculation step (S16), and a command step (S17).

  In the acquisition step (S11), the CPU load current database 21 sequentially acquires the CPU load current value Pt of the CPU load of each application server 3 that changes in time series. The acquisition timing is set to be every minute, but is not limited to this.

In the alignment step (S12), a CPU load that has been traced back by a predetermined time (n minutes) from the current time when prediction is performed is extracted from the CPU load history database 21, and a data string (P _t-1) connected in time series is extracted. ~ _Pn ).

The short-term time prediction step (S13) is based on the data sequence created in the alignment step (S12), for example, by performing a known chaos prediction process described in JP-A-7-270200, and after 5 minutes as the server startup time The short-term predicted value P _{t + 5} of the CPU load is calculated. In addition, the prediction time for this short-term prediction is not limited after 5 minutes.

The long-term time prediction step (S14) is based on the data string created in the alignment step (S12), performs chaos prediction processing in the same manner as the short-term time prediction step (S13), and performs a long-term prediction value of the CPU load after 10 minutes. P _{t + 10} is calculated. Note that the prediction time for this long-term prediction is not limited to 10 minutes later.

In the estimation step (S15), the CPU server current value Pt is compared with the short-term predicted value P _{t + 5} and the long-term predicted value P _{t + 10} to calculate the required number Np of the application server 3 after a certain time. At this time, for the fluctuation ratio of the short-term predicted value P _{t + 5} and the long-term predicted value P _{t + 10} with respect to the current value Pt of the CPU load, a threshold value previously associated with the required operating number Np is set, and based on this, 5 minutes from the present time The required number of operating units Np is calculated. For example, in the case of the system shown in FIG. 1, if P _{t + 5} > 70% of the maximum load and P _{t + 10} > 70% of the maximum load, the required number of operating units Np is 5, and P _{t + 5} > 70 of the maximum load. % And P _{t + 10} > 50% of the maximum load, the required operating number Np is determined by a combination of a plurality of predetermined thresholds such as four required operating numbers Np.

  In the calculation step (S16), the current operating number Na is subtracted from the calculated required operating number Np to calculate the number Nd to be added or stopped.

  The command process (S17) further includes a start process (S171) and a stop process (S172). In the starting step (S171), when the number of added / stopped units Nd calculated in the calculating step (S16) is positive, the application servers 3 corresponding to the number are started. On the other hand, in the stop step (S172), when the added / stopped number Nd is negative, the application servers 3 corresponding to the number are stopped.

  According to this server operation management apparatus 5, operation management program, and server operation management method, for example, hardware that is stopped immediately after increasing the number of operating application servers 3 by taking into account the short-term and long-term prediction results. It is possible to avoid a heavy burden on the server and to start up an extra application server 3 in advance, and a systematic increase / decrease operation of the number of server operations can be performed. Therefore, the energy saving effect can be further enhanced.

(Second Embodiment)
Next, a second embodiment will be described with reference to FIGS.
In addition, the same code | symbol is attached | subjected to the component similar to 1st Embodiment mentioned above, and description is abbreviate | omitted.
The difference between the second embodiment and the first embodiment is that the server operation management apparatus 30 according to the present embodiment has a short-term trend setting unit 31 as a function unit (program module) of the operation management program and a long-term trend. The setting unit 32 is further provided. Another difference is that the estimation unit 33 also serves as the function of the calculation unit 17 in the first embodiment. These functions are controlled by the control unit 35 realized by the processor 6 and the memory 7.

  The short-term trend setting unit 31 calculates a corresponding increase / decrease trend by applying to a plurality of thresholds according to the fluctuation ratio between the current value of the CPU load and the short-term predicted value. Further, the long-term trend setting unit 32 calculates a corresponding increase / decrease trend by applying to a plurality of threshold values according to the fluctuation ratio between the current value of the CPU load and the long-term predicted value.

  The estimation unit 33 calculates the number of application servers 3 to be added or stopped based on the calculated increase / decrease trends.

  The server operation management method according to the present embodiment includes an acquisition step (S21), an alignment step (S22), a short-term time prediction step (S23), a long-term time prediction step (S24), and a short-term trend setting step ( S25), a long-term tendency setting step (S26), an estimation step (S27), and a command step (S28).

  The acquisition step (S21), the alignment step (S22), the short-term time prediction step (S23), and the long-term time prediction step (S24) are the acquisition step (S11) according to the first embodiment, the alignment step (S12), The same processing is performed for the short-term time prediction step (S13) and the long-term time prediction step (S14).

In the short-term trend setting step (S25), an increase / decrease tendency corresponding to a fluctuation ratio between a predetermined current value Pt of the CPU load and a short-term predicted value Pt _{+ 5} is calculated. In calculating the trend, when the difference between the current value Pt and the short-term predicted value Pt _{+ 5} is Pd, for example, when Pd> 30%, an increasing trend (large), when 20% <Pd ≦ 30% Are increasing (medium), if 10% <Pd ≦ 20%, increasing (small), if −20% ≦ Pd <−10%, decreasing (small), −30% When ≦ Pd <−20%, allocation is made as a decreasing trend (middle), and when Pd <−30%, allocation is made as a decreasing trend (large). The classification here is not limited to these.

In the long-term trend setting step (S26), an increase / decrease tendency corresponding to a fluctuation ratio between a predetermined current value Pt of the CPU load and a long-term predicted value Pt _{+ 10} is calculated. In calculating the trend, when the difference between the current value Pt and the long-term predicted value Pt _{+ 10} is Pd, for example, when Pd> 0%, the trend is increasing, and when Pd <0%, the trend is decreasing. Allocate as follows. Note that the increase / decrease setting is not limited to these.

  Here, since the short-term prediction is more accurate than the long-term prediction, the increase / decrease trend in the short-term trend setting step (S25) is set to be subdivided into multiple stages than the increase / decrease trend in the long-term trend setting step (S26). To do.

The estimation step (S27) is calculated in the short-term trend setting step (S25) and the long-term trend setting step (S26) by comparing the current value Pt of the CPU load with the short-term predicted value Pt _{+ 5} and the long-term predicted value Pt _{+ 10} . Based on the respective increasing / decreasing trends, the number Nd of additional / stopped applications server 3 after a certain time is determined.

That is, the fluctuating reasoning is applied to the fluctuation ratio of the short-term predicted value P _{t + 5} and the long-term predicted value P _{t + 10} with respect to the current value Pt of the CPU load to calculate the increase / decrease number Nd with respect to the current operating number. For example, in the case of the system shown in FIG. 1, if the trend is increasing (large) and increasing, the number of additional / stopped Nd is +3, and if the trend is increasing (medium) and increasing, the additional / stopped number Nd The number of added / stopped units Nd is determined by a predetermined combination of threshold values such as +4 units.

  The command process (S28) further includes a start process (S281) and a stop process (S282), and performs the same processing as the command process (S17) according to the first embodiment.

  This server operation management apparatus 30, operation management program, and server operation management method can achieve the same effects as the server operation management apparatus 5, operation management program, and server operation management method according to the first embodiment. .

The technical scope of the present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention.
For example, although the CPU load of the application server 3 is used as the management target data, the present invention is not limited to this, and the amount of traffic or the amount of hard disk access may be used.

1 Cloud Computing System 3 Application Server (Information Processing Device)
5,30 Server operation management device (information processing device operation management device)
DESCRIPTION OF SYMBOLS 11 Acquisition part 12 Alignment part 13 Short-term time prediction part 15 Long-term time prediction part 16, 33 Estimation part 18 Command part 21 CPU load history database (storage part)
31 Short-term trend setting section 32 Long-term trend setting section

Claims (11)

An information processing apparatus operation management apparatus for controlling the number of operating information processing apparatuses connected to a plurality of terminal apparatuses via a network,
An acquisition unit that acquires management target data of the information processing apparatus that changes in time series and stores the management target data in a predetermined storage unit;
An alignment unit that creates the data string that is extracted from the storage unit in the past predetermined time from the storage unit and connected in time series;
Based on the data string, a short-term prediction unit that calculates a short-term prediction value of the management target data after a predetermined time;
A long-term time predicting unit that calculates a long-term predicted value of the management target data at a time point in the future based on the data sequence;
An estimation unit that calculates the required number of operating units of the information processing apparatus after the certain time, or the number to be added or stopped, by comparing the current value of the management target data with the short-term predicted value and the long-term predicted value. ,
A command unit that starts or stops the information processing apparatus so that the number of operating units after the predetermined time becomes the required number of operating units;
An information processing apparatus operation management apparatus comprising:   The estimation unit calculates the required number of operating units or the number of units to be added or stopped based on a fluctuation ratio of the short-term predicted value and the long-term predicted value with respect to a current value of the management target data. The information processing apparatus operation management apparatus according to 1. A short-term trend setting unit for calculating an increase / decrease trend according to a fluctuation ratio between the current value of the management target data and the short-term predicted value;
A long-term trend setting unit that calculates an increase / decrease trend according to a fluctuation ratio between the current value of the management target data and the long-term prediction value;
With
The information processing apparatus operation management apparatus according to claim 1, wherein the estimation unit calculates the number of the information processing apparatuses to be added or stopped based on each of the increase / decrease trends.   The information processing apparatus operation management apparatus according to claim 1, wherein the predetermined time is an activation time of the information processing apparatus. A plurality of information processing devices connected to a plurality of terminal devices via a network;
Information processing apparatus operation management apparatus according to any one of claims 1 to 4,
A cloud computing system characterized by comprising:   An operation management program for causing a computer to function as the information processing apparatus operation management apparatus according to any one of claims 1 to 4.   A computer-readable storage medium storing the operation management program according to claim 6. An information processing apparatus operation management method for controlling the number of operating information processing apparatuses connected to a plurality of terminal apparatuses via a network,
An acquisition step of acquiring management target data of the information processing apparatus that changes in time series,
An alignment step of creating a data string by concatenating the acquired management target data within a predetermined past time in time series;
A short-term time prediction step of calculating a short-term prediction value of the management target data after a predetermined time based on the data string;
A long-term time predicting step of calculating a long-term predicted value of the management target data at a time point in the future based on the data sequence;
An estimation step of calculating the required number of operating units of the information processing apparatus after the certain time, or the number to be added or stopped, by comparing the current value of the management target data with the short-term predicted value and the long-term predicted value; ,
A command step of starting or stopping the information processing apparatus so that the number of operating units after the predetermined time becomes the required number of operating units;
An information processing apparatus operation management method comprising:   The number of required operating units or the number to be added or stopped is calculated in the estimation step based on a fluctuation ratio of the short-term predicted value and the long-term predicted value with respect to a current value of the management target data. Item 9. The information processing apparatus operation management method according to Item 8. A short-term trend setting step of calculating an increase / decrease trend according to a fluctuation ratio between the current value of the management target data and the short-term predicted value;
A long-term trend setting step for calculating an increase / decrease trend according to a fluctuation ratio between the current value of the management target data and the long-term prediction value;
With
The information processing apparatus operation management method according to claim 8 or 9, wherein, in the estimation step, the number of the information processing apparatuses to be added or stopped is calculated based on each of the increase / decrease trends. The information processing apparatus operation management method according to claim 8, wherein the predetermined time is an activation time of the information processing apparatus.

Images