JPH0588917A - Job scheduling system - Google Patents

Abstract

PURPOSE:To easily perform the efficient system operation by preliminarily setting a scheduling update rule to automatically change a job scheduling rule in accordance with the load state of a system. CONSTITUTION:Jobs are inputted by a job input part 11 and are registered by a job queuing part 12, and a scheduling rule is stored by a scheduling control table part 13, and jobs are scheduled in accordance with the scheduling rule by a job scheduling part 14. Scheduled jobs are executed by a job executing part 15, and system load information is periodically outputted by a system load state monitor part 16, and the scheduling update rule is stored by a scheduling update rule storage part 17, and the scheduling rule is changed in accordance with system load information and the scheduling update rule by a scheduling control information update part 18.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a job scheduling method in a computer system that performs batch processing in units of jobs.

[0002]

2. Description of the Related Art Conventionally, a job scheduling method in a computer system that performs batch processing on a job basis uses a fixed scheduling rule regardless of the load state of the system. This scheduling rule is set by the system administrator or the system operator when the system is designed or when the system operation is changed, and the rule is fixedly operated. Then, when the change is made, the system administrator or the system operator gives an instruction.

[0003]

In the above-mentioned conventional method, the system administrator or the system operator monitors the system load state to perform optimal job scheduling according to the system load state. There was a problem that it was necessary to issue a rule change instruction in line with this.

[0004]

According to the job scheduling method of the present invention, in a computer system that performs batch processing on a job basis, a job input section for inputting a job and a job input from this job input section are executed. A job queuing unit that is registered in a predetermined queue according to a predetermined order until the start, and scheduling control information such as the number of concurrently executable jobs for each queue in this job queuing unit. Job scheduling for selecting the next job to be executed from the jobs registered in the job queuing section by referring to the scheduling control table section to be stored and the scheduling control information stored in this scheduling control table section Department and selected in this job scheduling department It shows the job execution unit that executes jobs, the system load status monitoring unit that monitors the load status of the above system and periodically outputs the system load information, and what kind of scheduling is performed according to the load status of the above system. The scheduling update rule storage unit storing the scheduled update rule, the system load information output from the system load state monitoring unit, and the scheduling update rule stored in the scheduling update rule storage unit, the scheduling in the scheduling control table unit. A scheduling control information updating unit for updating control information is provided, and the scheduling rule can be automatically changed according to the load state of the system.

[0005]

According to the present invention, the job scheduling rule can be automatically changed according to the current load state of the system, so that optimum system operation can be easily performed.

[0006]

FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, 11 is a job input unit for inputting a job, and 12 is a job input unit 11 that registers jobs in a predetermined queue in a predetermined order until execution of the job is started. The job queuing unit 13 stores a scheduling control table unit 13 for storing scheduling control information such as the number of concurrently executable jobs for each queue in the job queuing unit 12, and 14 stores it in the scheduling control table unit 13. The job scheduling unit that selects the next job to be executed from the jobs registered in the job queuing unit 12 by referring to the scheduling control information that is present, 15 executes the job selected by the job scheduling unit 14. The job execution unit, which executes 16 monitors the system load condition and periodically A system load state monitoring unit that outputs system load information, 17 is a scheduling update rule storage unit that stores a scheduling update rule that indicates what kind of scheduling should be performed according to the system load state, and 18 is a system load state monitoring unit 16
The scheduling control information updating unit updates the scheduling control information in the scheduling control table unit 13 from the system load information output from the system and the scheduling update rule stored in the scheduling update rule storage unit 17. Then, the scheduling rule can be automatically changed according to the load state of the system.

FIG. 2 is an explanatory diagram showing an example of the job queuing unit 12 in FIG. In FIG. 2, 21
Represents the name of each queue (queue name), 22 is the maximum CPU usage time of jobs that can be registered in that queue, and 23 is the memory usage of jobs that can be registered in that queue. Is the maximum value (maximum memory used size), and 24 is a queue (queue section) for registering each job. The circles in the figure represent individual jobs registered in each queue. Among these, the values of 21, 22, and 23 other than the jobs represented by ◯ in the queue of the queue unit 24 for registering each job are initial values at the time of system design or operation change at the time of system administrator or system operation. It is set by the person. In this example, Q ₁ is C
Queue for registering jobs with low PU usage and memory usage, Q ₂ for queuing jobs with medium CPU usage and medium memory usage, and Q ₃ for CPU usage and memory usage Shows queues for registering jobs with a large usage amount. And FIG.
The job input unit 11 shown in FIG. 1 reads a job image from a file on a card reader or a magnetic disk device, and stores the job image in the job queuing unit 12 according to how much CPU the job uses or how much memory the job requires. Register the job in one of the queues.

FIG. 3 is an explanatory diagram showing an example of the scheduling control table unit 13 in FIG. In FIG. 3, 31 is the name of each queue (queue name), 32 is the initial value of the number of concurrently executable jobs in each queue, 33 is the current value of the number of concurrently executable jobs in each queue, 34
Represents the number of jobs currently being executed among the jobs registered in the queue (currently executed job number).

FIG. 4 is an explanatory diagram showing an example of the scheduling update rule stored in the scheduling update rule storage unit 7 in FIG. In FIG. 4, (a)
Indicates the first condition of the scheduling update rule,
(B), (c), and (d) indicate the following conditions.

Next, the operation of the embodiment shown in FIG. 1 will be described with reference to FIGS. First, the job scheduling unit 14 shown in FIG. 1 next writes a new job in the job queuing unit 12 or finishes a running job, and updates the scheduling control information in the scheduling control table unit 13 at the next timing. The job to be executed is searched by referring to the scheduling control information in this scheduling control table unit 13. When a new job is registered, the current value 33 of the number of concurrently executable jobs shown in FIG. 3 of the scheduling control information in the scheduling control table unit 13 corresponding to the queue in which the job is registered and the current value The number of executed jobs 34 is compared, and if the number of currently executed jobs 34 is smaller than the current value 33 of the number of concurrently executable jobs, the highest priority is given to the jobs registered in the queue. Select the job as the next job to run. Then, when the execution of the job is completed, the value of the number 34 of currently executing jobs of the scheduling control information corresponding to the queue in the job queuing unit 12 which was registered before the execution of the job is set to " If there is a job registered in the queue, the job with the highest priority is selected as the job to be executed next.

Next, when the scheduling control information is updated, the number of jobs currently being executed 34 is smaller than the current value 33 of concurrently executable jobs and the job waiting to be executed is determined by the result of the update. It is checked whether an existing queue is created, and if such a queue is created, the job with the highest priority is selected from the jobs registered in that queue as the job to be executed next. When a job can be selected by the above algorithm, the selected job is passed to the job execution unit 15 and
The value of the currently executed job number 34 in the scheduling control information corresponding to the queue in which the job is registered is increased by "1".

As described above, FIG. 4 is an example of the scheduling update rule stored in the scheduling update rule storage unit 17 in FIG. 1, and this scheduling update rule is used by the system administrator or the system when the system is designed or the system operation is changed. Set or changed by the operator.

The system load condition monitoring unit 16 in FIG. 1 monitors the load condition of the system CPU, memory, etc., and outputs the information to the scheduling control information updating unit 18 regularly and appropriately. In the case of this example, CP within a certain period
The usage rate of U and the usage rate of memory are output. The scheduling control information updating unit 18 in FIG. 1 performs scheduling in the scheduling control table unit 13 according to the scheduling update rule stored in the scheduling update rule storage unit 17 each time the system load information is received from the system load state monitoring unit 16. Update control information.

Now, from the system load state monitoring unit 16 to C
A system with a PU usage rate of 45% and a memory usage rate of 70%
System load information to the scheduling control information updating unit 18.
If passed, the scheduling update rule of Figure 4
According to the scheduling control table of FIG.
Explain how to update the content. Schedule of Figure 4
The first condition (a) of the rule update rule is CPU usage ratio
Since the ratio and memory usage ratio do not hold, the following condition
Examine (b). And the system load status monitor 16
The CPU usage rate of the system load information received from is 45%
Therefore, condition (b) holds. Therefore, next
Check the n-th clause and check the condition (C). Where memory usage
Is 70% and the condition (C) does not hold
Follow the instructions to name each queue in Figure 3 (Queue
Queue Q in name 31₁Current concurrency of
Increase the current value 33 of the number of jobs by 2 to increase from "6" to "8"
To In the case of this example, the job queuing unit 1 of FIG.
Queue of 2₁Queue unit 24 for registering each job of
Queue Q, as shown in₁Waiting for execution
Job scheduling because many jobs are registered
The current value 33 of the number of jobs that can be concurrently executed by the unit 14 is 33.
Queue Q by increasing the value of ₁Registered in
Two jobs are executed in descending order of priority
To be done.

[0015]

As described above, according to the present invention, the job scheduling rule can be automatically changed according to the current system load condition.
By setting the pre-scheduling update rule, the scheduling rule can be automatically adjusted according to the load status of the system, so that there is an effect that the efficient and optimal system operation can be performed.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a job queuing unit in FIG.

FIG. 3 is an explanatory diagram showing an example of a scheduling control table unit in FIG.

FIG. 4 is an explanatory diagram showing an example of a scheduling update rule stored in a scheduling update rule storage unit in FIG.

[Explanation of symbols]

 11 Job Input Unit 12 Job Queuing Unit 13 Scheduling Control Table Unit 14 Job Scheduling Unit 15 Job Execution Unit 16 System Load Status Monitoring Unit 17 Scheduling Update Rule Storage Unit 18 Scheduling Control Information Updating Unit

Claims (1)

[Claims] 1. In a computer system for performing batch processing on a job basis, a job input section for inputting a job and a job input section from this job input section according to a predetermined order until execution of the job is started. A job queuing section that is registered in a specified queue,
A scheduling control table unit that stores scheduling control information such as the number of concurrently executable jobs for each queue in this job queuing unit, and the job queue by referring to the scheduling control information stored in this scheduling control table unit The job scheduling unit that selects the next job to be executed from among the jobs registered in the scheduling unit, the job execution unit that executes the job selected by this job scheduling unit, and the load status of the system is monitored. A system load state monitoring unit that periodically outputs system load information, a scheduling update rule storage unit that stores a scheduling update rule that indicates what kind of scheduling is to be performed according to the load state of the system, and the system load Output from status monitor Has a scheduling control information updating unit that updates the scheduling control information in the scheduling control table unit from the system load information stored in the scheduling update rule storage unit and the scheduling update rule stored in the scheduling update rule storage unit. A job scheduling method characterized in that the scheduling rule can be changed.