JP2014186416A - System, method and device for managing job operation - Google Patents

Abstract

An object of the present invention is to reduce the total operation time of an instance and reduce the operation cost of a job operation management system.
An instance in which a job execution agent operates on a computer. The job management manager 106 includes a job definition database 305 and an instance management control unit 304. The job definition database 305 stores job start times when the job execution agent starts executing each job for a plurality of jobs. The instance management control unit 304 starts the job of the next job in the job execution agent stored in the instance management control unit 304 after the number of jobs executed by the job execution agent running on the instance becomes zero. When the time until the time is longer than the start control time obtained by a predetermined method, the instance is stopped, and the instance stopped before the job start time of the next job is restarted.
[Selection] Figure 3

Description

  The present invention relates to a job operation management system, a job operation management method, and a job management apparatus that manage jobs executed by a job execution agent that operates on an instance running on a computer.

In recent years, the scope of application of job automation by job operation management systems has become large and complex. In a large-scale / complex job operation management system, in order to process a large number of jobs, a method of distributing and executing a job to a job execution agent on a server on which various systems are operating is generally used (for example, Patent Documents). 1).
Cloud computing has begun to be used to operate this job execution agent. In cloud computing, it is possible to process many jobs simultaneously by using many computing resources and performing distributed data processing. Hereinafter, the actual state of one process that uses a computing resource to provide a function is collectively referred to as an “instance”. An example of an instance is a virtual computer, for example.

JP 2004-102449 A

When a large number of jobs are executed in the job operation management system, many resources can be freely used by operating the job execution agent on the instance and increasing the number of job execution agents.
However, in cloud computing, you are charged according to the operating time. When using cloud computing, as the number of instances increases, the operation cost of the job operation management system increases.

  An object of the present invention is to provide a job operation management system, a job operation management method, and a job management apparatus that can reduce the total operation time of an instance and reduce the operation cost.

In order to achieve the above object, the job operation management system of the present invention provides:
A job operation management system having a computer on which an instance on which a job execution agent operates, and a job management device,
The job management device is
Job start time storage means for storing a job start time at which the job execution agent starts execution of each job for a plurality of jobs;
The time from when the number of jobs executed by the job execution agent running on the instance becomes 0 to the job start time of the next job in the job execution agent stored in the job start time storage unit Is an instance stop means for stopping the instance when it is longer than the start control time obtained by a predetermined method;
Instance restarting means for restarting the stopped instance before the job start time of the next job;
Comprising
It is characterized by that.

Preferably, the job operation management system according to the present invention includes:
The job management device comprises a minimum stop time storage means for storing a preset minimum stop time for an instance;
The instance stop means is an estimated start time that is an estimated time required to start an instance, an estimated stop time that is an estimated time required to stop an instance, and a minimum stored in the minimum stop time storage means The start control time is obtained by adding the stop time,
The instance restarting unit restarts the stopped instance at a time earlier than the job start time of the next job by the predicted startup time;
It is characterized by that.

Preferably, the job operation management system according to the present invention includes:
When the instance stop means obtains the start control time, the average time taken to start the instance is used as the predicted start time, and the average time taken to stop the instance is used as the predicted stop time.
The instance restarting means uses an average time taken to start the instance as the predicted start time.
It is characterized by that.

In addition, the job operation management method of the present invention includes:
A computer on which an instance on which the job execution agent operates operates; and a job management apparatus including a job start time storage unit that stores job start times at which the job execution agent starts executing each job for a plurality of jobs. A job operation management method in a job operation management system,
After the number of jobs executed by the job execution agent operating on the instance becomes 0, the job management apparatus stores the next job in the job execution agent stored in the job start time storage unit. An instance stop step for stopping the instance when the time until the job start time is longer than the start control time obtained by a predetermined method;
An instance restarting step in which the job management device restarts the stopped instance before the job start time of the next job;
It is characterized by providing.

In addition, the job management apparatus according to the present invention includes:
Job start time storage means for storing a job start time at which the job execution agent starts executing each job for a plurality of jobs;
After the number of jobs executed by the job execution agent running on the instance running on the computer becomes zero, the job start of the next job in the job execution agent stored in the job start time storage means Instance stopping means for stopping the instance when the time until the time is longer than the start control time obtained by a predetermined method;
Instance restarting means for restarting the stopped instance before the job start time of the next job;
It is characterized by providing.

  According to the present invention, it is possible to reduce the total operation time of the instance and reduce the operation cost of the job operation management system.

It is a figure which shows an example of a structure of the job operation management system which concerns on embodiment of this invention. It is a figure which shows an example of the internal structure of the information processing apparatus which comprises each server computer, job management manager server, and job execution agent server which comprise a server computer group. It is a figure which shows an example of a structure of a job management manager. It is a figure which shows an example of a structure of the job definition data stored in a job definition database. It is a figure which shows an example of a structure of the agent definition data stored in an agent definition database. It is a figure which shows an example of a structure of the instance statistical data stored in an instance statistical database. It is a figure which shows an example of the instance control start time data which an instance control management part calculates | requires and memorize | stores in the main memory of a job management manager server. It is a figure which shows the instance starting time data which an instance control management part calculates | requires and memorize | stores in the main memory of a job management manager server. It is a figure which shows an example of the job number data in the instance control which an instance management control part calculates | requires and memorize | stores in the main memory of a job management manager server. It is a flowchart which shows an example of the flow of a process in which an instance management control part calculates instance control start time. It is a flowchart which shows an example of the flow of a process in which an instance management control part calculates | requires instance start time. It is a flowchart which shows an example of the flow of a process in which an instance management control part starts an instance. It is a flowchart which shows an example of the flow of a process in which an instance management control part stops an instance. It is a figure which shows an example of control of the instance where the job execution agent A operates.

  Hereinafter, a job operation management system, a job operation management method, and a job management apparatus according to an embodiment of the present invention will be described with reference to the drawings. In all the drawings for explaining the embodiments, common constituent elements are denoted by the same reference numerals, and repeated explanation is omitted.

FIG. 1 shows an example of the configuration of a job operation management system according to an embodiment of the present invention.
Each apparatus constituting the job operation management system according to the embodiment of the present invention is connected to be communicable with each other via a network 100 and an Internet 101 which are communication lines for exchanging information.
The devices constituting the job operation management system are a server computer group 102 composed of one or more server computers, a job management manager server 103 that is a computer that manages job execution, and a computer that executes jobs. A job execution agent server 104 and a terminal 105 used by a user who manages job execution.
In the job management manager server 103, a job management manager 106 operates. The job management manager 106 manages job execution in a plurality of job execution agents. In the server computer group 102, the instance 107 and the instance 108 are realized as virtual computers. In the instance 107 and the instance 108, a job execution agent A 109 and a job execution agent B 110 operate. Further, the job execution agent server 111 operates in the job execution agent server 104. Job execution agent A 109, job execution agent B 110, and job execution agent C 111 execute jobs. Note that FIG. 1 illustrates a job operation management system including one job execution agent server and two instances. The number of job execution agent servers and instances may be arbitrary.

FIG. 2 shows an example of the internal configuration of the information processing apparatus 200 constituting each server computer, job management manager server 103, and job execution agent server 104 constituting the server computer group 102.
The information processing apparatus 200 is a computer. The information processing apparatus 200 includes a main memory 202 including a CPU (Central Processing Unit) 201, a RAM (Random Access Memory), a hard disk 203, and a network interface card 204.
The network interface card 204 exchanges data bidirectionally with other nodes (computers and the like) via the network 100.

FIG. 3 shows an example of the configuration of the job management manager 103.
The hard disk 203 of the job management manager server 103 stores a job management program, a job definition database 305, an agent definition database 306, and an instance statistics database 307.
When the CPU 201 of the job management manager server 103 reads the job management program from the hard disk 203 into the main memory 202 and executes it, the job definition database reference update unit 301, the agent definition database reference update unit 302, and the job execution management unit 303 are executed. And the function of each part with the instance management control part 304 is implement | achieved.

As will be described later, the job definition database 305 stores job definition data including a job name, a job execution agent name, and a job start time. The job definition database reference update unit 301 updates the job definition database 305 according to an instruction from the terminal 105 and visualizes and displays the job definition data on the display of the terminal 105.
As will be described later, the agent definition database 306 stores agent definition data including a job execution agent name, an instance identification flag, and a minimum instance stop time. The agent definition database reference update unit 302 updates the agent definition database 306 in response to an instruction from the terminal 105, and visualizes and displays the agent definition data on a display or the like of the terminal 105.
The job execution management unit 303 manages job execution of the job execution agent A 109, job execution agent B 110, and job execution agent C 111.
The instance statistical database 307 stores instance statistical data including a job execution agent name 601, a total required start time 602, a start count 603, a total stop required time 604, and a stop count 605. The instance management control unit 304 manages the start and stop of the instance, and updates the instance statistics database 307.
The job management manager 103 is an example of a job management apparatus according to the present invention, and the instance management control unit 304 is an example of an instance stop unit and an instance restart unit according to the present invention.

FIG. 4 shows an example of the configuration of job definition data stored in the job definition database 305.
The job definition data in the job definition database 305 includes a job name 401, a job execution agent name 402, and a job start time 403.
The job name 401 is information for uniquely identifying a job.
The job execution agent name 402 is information for identifying with which job execution agent the job is executed.
The job start time 403 is a time when the job execution management unit 303 starts job execution.
The job definition database 305 is an example of a job start time storage unit of the present invention.

FIG. 5 shows an example of the configuration of agent definition data stored in the agent definition database 306.
The agent definition data in the agent definition database 306 includes a job execution agent name 501, an instance identification flag 502, and a minimum instance stop time 503.
The job execution agent name 501 is information for uniquely identifying the job execution agent.
The instance identification flag 502 indicates whether the job execution agent is operating on a computer that cannot be managed as an instance, such as the job management agent server 104, or is operating on an instance of the server computer group 102, such as the instance 107 and the instance 108. Flag information for identifying whether or not.
For example, when the instance management control unit 304 extracts only jobs that operate on the instance, first, the job execution agent A and the job execution agent B are executed on the instance based on the instance identification flag 502 of the agent definition database 306. Identify that it works. Then, the instance management control unit 304 searches the job definition database 305 and selects jobs that operate on the job execution agent A and the job execution agent B. In the cases shown in FIGS. 4 and 5, JobA, JobB, JobC, JobD, and JobF are extracted as jobs operating on the instance (JobE is excluded).

The minimum stop time 503 of the instance is the minimum time for stopping the instance, and is set in advance. The instance management control unit 304 stops the instance only when the instance can be stopped for the minimum stop time 503 or more. Even if the instance is stopped, if the instance must be restarted before the minimum stop time 503 elapses, the instance management control unit 304 does not stop the instance.
Here, a certain amount of time is required to start and stop the instance, and the start time required to start the instance and the stop time required to stop the instance vary depending on the execution status of other jobs.
Therefore, the instance management control unit 304 adds the predicted start time that is the predicted time required to start the instance, the predicted stop time that is the predicted time required to stop the instance, and the minimum stop time 503 of the instance. Thus, the start control time is obtained. Then, the instance management control unit 304 starts the time until the job execution agent starts executing the next job after the number of jobs executed by the job execution agent running on the instance becomes zero. If it is longer than the control time, the instance is stopped. On the other hand, when the number of jobs executed by the job execution agent reaches 0, the instance management control unit 304 starts execution of the next job before the start control time elapses. Does not stop the instance. By including the minimum instance stop time 503 in the start control time, it is guaranteed that the job can be executed when the job start time is reached.
The agent definition database 306 is an example of the minimum stop time storage unit of the present invention.

FIG. 6 shows an example of the configuration of instance statistical data stored in the instance statistical database 307.
The instance statistical data in the instance statistical database 307 includes a job execution agent name 601, a total required start time 602, a start count 603, a total stop required time 604, and a stop count 605.
The job execution agent name 601 is information for uniquely identifying the job execution agent.
The total required start time 602, the start count 603, the total stop required time 604, and the stop count 605 are statistical information of instances in which the job execution agent having the job execution agent name 601 operates.
The total startup time 602 is the sum of the time required to start an instance. The activation count 603 is the number of times the instance has been activated. By dividing the total startup required time 602 by the startup count 603, the average time required to start the instances can be calculated. The average time required to start an instance is used as the predicted start time when obtaining the above-described start control time.
The total stop time 604 is the sum of the time required to stop the instance. The stop count 605 is the number of times the instance is stopped. By dividing the total required stop time 604 by the stop count 605, the average time required to stop the instance can be calculated. The average time taken to stop the instance is used as the predicted stop time when obtaining the above-described start control time.

  For example, when the instance management control unit 304 tries to calculate the average time required to start and stop the instance in which the job execution agent A operates, the total number of start required times of the job execution agent A is 60 minutes. By dividing, it is possible to calculate the average time required to start the instance in which the job execution agent A is operating as 10 minutes. Similarly, by dividing the total required stop time of the job execution agent A by 30 minutes by the stop count 6, the average time required to stop the instance in which the job execution agent A is operating can be calculated as 5 minutes.

FIG. 7 shows an example of instance control start time data obtained by the instance control management unit 304 and stored in the main memory 202 of the job management manager server 103.
The instance control start time data includes a job name 701, a job execution agent name 702, and an instance control start time 703.
The job name 701 is information for identifying which job is the instance control start time associated with the execution.
The job execution agent name 702 is information for identifying the execution start time of the instance in which the execution agent operates for the job name 701.
The instance control start time 703 is a time when the instance management control unit 304 starts instance control for the job name 701. The instance management control unit 304 creates data at the instance control start time 703. The instance management control unit 304 first obtains the start control time by adding the average time taken to start and stop the instance and the minimum stop time 503 of the instance. Next, the instance management control unit 304 obtains an instance control start time 703 that is earlier than the job start time 403 by the start control time. The instance management control unit 304 stores the instance control start time 703 together with the job name 701 and the job execution agent name 702 in the main memory 202 as instance control start time data.

  For example, when the instance management control unit 304 tries to obtain the instance control start time of Job A, the average time taken to start and stop the instance of the job execution agent A is 10 minutes and 5 minutes from the instance statistical database 307. Ask. Further, 60 minutes that is the minimum stop time 503 of the instance of the job execution agent A is acquired from the agent definition database 306. The start control time (1 hour and 15 minutes in total), which is the sum of the average time taken to start and stop the instance, 15 minutes and the minimum instance stop time of 60 minutes, is advanced from JobA start time 7:00: 45 is obtained as the instance control start time of JobA.

FIG. 8 shows the instance activation time data obtained by the instance control management unit 304 and stored in the main memory 202 of the job management manager server 103.
The instance activation time data includes a job execution agent name 801 and an instance activation time 802.
The job execution agent name 801 is information for identifying the job execution agent operating on the instance in order to uniquely identify the instance.
The instance activation time 802 is a time at which the instance activation is started. When the instance management control unit 304 reaches the instance control start time 703 and the instance is not activated, the instance management control unit 304 first acquires the job start time 403 from the job definition database 305 based on the job name 701. Next, the instance management control unit 304 obtains a time that is earlier than the job start time 403 by the estimated time required to start the instance as the instance start time 802 and stores it in the main memory 202. Here, the instance activation time 802 is the time when the instance is activated when the job start time 403 is reached and the job execution agent can execute the job. Therefore, for example, the instance management control unit 304 may use the average time required for starting an instance as the estimated time required for starting the instance, or may estimate a time slightly longer than the average time required for starting the instance as the estimated time required for starting the instance. It is good.

FIG. 9 shows an example of job count data during instance control that is obtained by the instance management control unit 304 and stored in the main memory 202 of the job management manager server 103.
The job count data during instance control includes a job execution agent name 901 and the job count 902 during instance control.
The job execution agent name 901 is information for identifying the job execution agent operating on the instance in order to uniquely identify the instance.
The number of jobs under instance control 902 is a value indicating how many jobs are subject to instance control for each instance on which the job execution agent indicated by the job execution agent name 901 operates. Used to determine when to perform. The instance management control unit 304 starts an instance when the number of jobs 902 under instance control changes from 0 to 1, and stops the instance when the number of jobs 902 changes from 1 to 0.
For example, when the instance management control unit 304 reaches the instance control start time 703 of Job A, the instance management control unit 304 counts up the number 902 of jobs in the instance control of the job execution agent A existing on the main memory 202. If the number of jobs 902 during instance control before counting up is 0, instance control is started.
When the job execution management unit 303 finishes executing Job A normally, the instance management control unit 304 receives a notification from the job execution management unit 303 that Job A has ended normally, and the number of jobs 902 in the job execution agent A during instance control. Count down. Further, the instance management control unit 304 stops the instance when the number of jobs 902 during instance control after the countdown becomes zero.

FIG. 10 is a flowchart illustrating an example of a flow of processing in which the instance management control unit 304 obtains the instance control start time 703.
The instance management control unit 304 obtains a list of job execution agent names 501 operating on the instance from the agent definition database 306 at the same time as the startup (S11).
Based on the list of job execution agent names 501 acquired in step S11, the instance management control unit 304 uses the job name as job definition data of the job executed by the job execution agent operating on the instance from the job definition database 305. 401, the job execution agent name 402, and the job start time 403 are acquired (S12).
The instance management control unit 304 refers to the instance statistics database 307 based on the list of job execution agent names 501 acquired in step S11, and calculates the average time required to start and stop the instance of each job execution agent as the total required time. Is divided by the number of times (S13).
The instance management control unit 304 calculates the start control time by adding the average time required to start and stop the instance and the minimum stop time 503 of the instance. Then, the instance management control unit 304 obtains, as the instance control start time 703, a time that is earlier than the job start time 403 by the start control time. The instance control start time data is saved (step 1004).

FIG. 11 is a flowchart illustrating an example of a flow of processing in which the instance management control unit 304 obtains the instance start time 802.
When the instance management control unit 304 detects that the instance control start time 703 has been reached (S21), the instance management control unit 304 determines whether the number of jobs 902 under instance control is 0 (S22).
When the number of jobs 902 under instance control is 0 (S22: YES), the instance management control unit 304 obtains an average time for instance startup from the instance statistics database 307, and averages for instance startup from the job start time 403 The time earlier by the time is obtained as the instance activation time 802 (S23). On the other hand, when the number of jobs 902 under instance control is not 0 (S22: NO), the instance management control unit 304 does not obtain the instance activation time 802.
Next, the instance management control unit 304 counts up the number of jobs 902 under instance control (S24).

FIG. 12 is a flowchart illustrating an example of a process flow in which the instance management control unit 304 activates an instance.
When the instance management control unit 304 detects that the instance activation time 802 has been reached (S31), it executes an instance activation instruction (S32).
Next, after confirming that the instance has been started, the instance management control unit 304 updates the instance statistics database 307 by adding the time required for start-up to the total start time 602 of the instance statistics database 307, and counts up the number of start-ups 603 (S33).

FIG. 13 is a flowchart illustrating an example of a process flow in which the instance management control unit 304 stops an instance.
When the instance management control unit 304 receives a notification that the job has been normally completed from the job execution management unit 303 (S41), the instance management control unit 304 counts down the number of jobs 902 under instance control (S42).
Next, the instance management control unit 304 determines whether or not the number of jobs 902 under instance control is 0 (S43). If it is not 0 (S43: NO), the instance stop command is not executed. The process ends.
When the number of jobs 902 under instance control is 0 (S43: YES), the instance management control unit 304 executes an instance stop command (S44).
After confirming that the instance has been stopped, the instance management control unit 304 adds the time required for the stop to the total stop time 604 in the instance statistics database 307 and updates it, and counts up the stop count 605 (S45). ).

FIG. 14 shows an example of control of the instance 107 on which the job execution agent A 109 operates.
Here, it is assumed that each execution time of JobA 1401, JobC 1402, and JobF 1403 is 1 hour, and the end times thereof are 8:00, 10:00, and 13:00.
At the instance control start time 5:45 of JobA 1401, the number of jobs 902 in the instance control before counting up is zero. For this reason, the instance management control unit 304 obtains 6:50, which is 10 minutes earlier than the job start time 7:00 of the job A 1401 as an average start time for starting the job execution agent A, as the instance start time, and starts the instance.
At the instance control start time 7:45 of JobC 1402, the number of jobs 902 during instance control before counting up is 1, so the instance management control unit 304 does not obtain the instance activation time. Job A 1401 ends at the end time 8:00, but the instance management control unit 304 does not stop the instance because the number of jobs 902 under instance control after the countdown is 1. When JobC 1402 ends at the end time 10:00, the number of jobs 902 under instance control after the countdown becomes 0, so the instance management control unit 304 stops the instance, and the instance stops at 10:05.
Further, when the instance control start time 10:45 of JobF 1403 is reached, the number of jobs 902 during instance control before counting up is zero. For this reason, the instance management control unit 304 obtains 11:50, which is 10 minutes earlier than the job start time 10:00 of the job start time of JobF 1403, as an instance start time, and starts the instance again. . When JobF 1403 ends at the end time 13:00, the number of jobs 902 under instance control after the countdown becomes 0, so the instance management control unit 304 again executes the instance stop, and the instance stops at 13:05. .

  As described above, according to the present invention, the instance is stopped when the job is not executed. For this reason, the total operation time of the instance can be reduced, and the operation cost of the job operation management system can be reduced.

100 Network 101 Internet 102 Server Computer Group 103 Job Management Manager Server 104 Job Execution Agent Server 105 Terminal 106 Job Management Manager 107, 108 Instance 109 Job Execution Agent A
110 Job execution agent B
111 Job execution agent C
301 Job definition database reference update unit 302 Agent definition database reference update unit 303 Job execution management unit 304 Instance management control unit 305 Job definition database 306 Agent definition database 307 Instance statistics database 503 Minimum instance stop time 602 Total required startup time 603 Start Number of times 604 Total stop time 605 Number of times of stop 703 Instance control start time 802 Instance start time 902 Number of jobs under instance control

Claims (5)

A job operation management system having a computer on which an instance on which a job execution agent operates, and a job management device,
The job management device is
Job start time storage means for storing a job start time at which the job execution agent starts execution of each job for a plurality of jobs;
The time from when the number of jobs executed by the job execution agent running on the instance becomes 0 to the job start time of the next job in the job execution agent stored in the job start time storage unit Is an instance stop means for stopping the instance when it is longer than the start control time obtained by a predetermined method;
Instance restarting means for restarting the stopped instance before the job start time of the next job;
Comprising
A job operation management system characterized by that. The job management device comprises a minimum stop time storage means for storing a preset minimum stop time for an instance;
The instance stop means is an estimated start time that is an estimated time required to start an instance, an estimated stop time that is an estimated time required to stop an instance, and a minimum stored in the minimum stop time storage means The start control time is obtained by adding the stop time,
The instance restarting unit restarts the stopped instance at a time earlier than the job start time of the next job by the predicted startup time;
The job operation management system according to claim 1. When the instance stop means obtains the start control time, the average time taken to start the instance is used as the predicted start time, and the average time taken to stop the instance is used as the predicted stop time.
The instance restarting means uses an average time taken to start the instance as the predicted start time.
The job operation management system according to claim 2. A computer on which an instance on which the job execution agent operates operates; and a job management apparatus including a job start time storage unit that stores job start times at which the job execution agent starts executing each job for a plurality of jobs. A job operation management method in a job operation management system,
After the number of jobs executed by the job execution agent operating on the instance becomes 0, the job management apparatus stores the next job in the job execution agent stored in the job start time storage unit. An instance stop step for stopping the instance when the time until the job start time is longer than the start control time obtained by a predetermined method;
An instance restarting step in which the job management device restarts the stopped instance before the job start time of the next job;
A job operation management method comprising: Job start time storage means for storing a job start time at which the job execution agent starts executing each job for a plurality of jobs;
After the number of jobs executed by the job execution agent running on the instance running on the computer becomes zero, the job start of the next job in the job execution agent stored in the job start time storage means Instance stopping means for stopping the instance when the time until the time is longer than the start control time obtained by a predetermined method;
Instance restarting means for restarting the stopped instance before the job start time of the next job;
A job management apparatus comprising:

Images