JP4650504B2 - Process control device, process control method, and process control program - Google Patents

Description

  The present invention relates to a process control device, a process control method, and a process control program.

  As a technology related to the present invention, when a workload class uses up the CPU time given to the workload class while the system shared resource is locked, the system is unconditionally shared without stopping the execution of the process. There has been a CPU distribution device that gives CPU resources over the allotted time until resources are released. The workload class is a CPU allocation unit.

  FIG. 24 is a block diagram showing an example of a CPU distribution device related to the present invention. The CPU distribution device illustrated in FIG. 24 includes a CPU distribution table 202, a target time storage table 203, a process correspondence table 204, a usage time storage table 205, an interval initialization unit 250, a target time calculation unit 210, A process selection unit 220, a process execution unit 240, a process management table 207, an executable process queue storage unit 206, and a resource management table 208 are provided.

  The CPU distribution table 202 stores a CPU distribution ratio for each workload class. The target time storage table 203 stores a target time for each workload class. The process correspondence table 204 stores the workload class number of each process. The usage time storage table 205 stores the CPU usage time for each workload class. The interval initialization unit 250 initializes the CPU usage time stored in the usage time storage table 205 at regular intervals. The resource management table 208 stores a queue of processes waiting for resource release.

  The target time calculation unit 210 sets a target time according to the CPU distribution ratio. The process determination unit 240 executes the executable process stored in the executable process queue storage unit 206 and records the process status in the process management table 207.

  The process selection unit 220 operates as described below. 25 and 26 are flowcharts showing the operation of the process selection means 220 of the CPU distribution device shown in FIG. The process selection means 220 acquires the process number and execution priority of the process queued at the head of the executable process queue storage means 206, and sets them in the variables n and w, respectively (step S201). Then, it is determined whether or not the execution priority order w is higher than the priority order that can be set by the user (step S202).

  If the execution priority w is higher than the priority that can be set by the user, control is passed to the process execution means 240. On the other hand, if the execution priority w is within the range of the priority order that can be set by the user, the process proceeds to step S203.

  In step S203, the workload class number of process number n is acquired from the process correspondence table 204. Then, the CPU target time of the workload class is acquired from the target time storage table 203 (step S204), and the CPU usage time of the workload class is acquired from the usage time storage table 205 (step S205).

  After step S205, the process selection unit 220 determines whether or not the CPU usage time is less than the CPU target time (step S206). When the CPU usage time is less than the CPU target time, control is passed to the process execution means 240. If the CPU usage time is equal to or longer than the CPU target time, the process proceeds to step S207.

  In step S207, the process selection unit 220 determines whether the next process stored in the executable process queue storage unit 206 is queued. If the next process is queued, the process proceeds to step S208. If the next process is not queued, control is passed to the process selection unit 220 (in other words, the process is started from step S201).

  In step S208, the process number and execution priority of the next process stored in the executable process queue storage unit 206 are acquired and set in the variables n and w, respectively. Then, the process after step S202 is repeated.

  With the above operation, if the workload class uses up the CPU time given to the workload class while the system shared resource is locked, the process execution is not aborted and the system shared resource is unconditionally CPU resources are allocated over the allotted time until is released. Due to such an operation, the process of another workload class waits for the system shared resource acquired by another workload class that has used up the CPU time even though the CPU resource is not yet used. It was prevented from falling into a state where it could not be executed. When the system shared resource is locked, the execution priority is always set to be higher than the priority range that can be set by the user. Then, based on the execution priority, it is determined whether or not system shared resources are being acquired.

  As another method, there is a method in which the CPU time used by the system is not subject to allocation to the workload class. In this method, an allocation ratio is designated for each workload class for the remaining time excluding the CPU time used by the system from the entire CPU time. Then, the CPU usage time is accumulated for each workload class, and when the CPU time allocated to each workload class is exceeded, the priority order of processes executed in the workload class is lowered. By controlling in this way, CPU resources are preferentially allocated to workload class processes that have not yet used the allocated CPU resources.

  Japanese Patent Laid-Open No. 2004-26883 allows designation of a CPU resource distribution ratio for each job group classified in advance according to the nature of the job, and shows the CPU resource distribution ratio and the actual distribution allocated to each job. Describes a job schedule processing method that allows a job schedule to be performed based on the above.

  Further, in Patent Document 2, when a job releases a shared resource in the shared resource release request process, it is determined whether or not the job is a temporary CPU allocation guarantee job. In some cases, a job scheduling method for releasing the temporary CPU assignment is described.

  Further, Patent Document 3 discloses a task priority change method for increasing the execution priority of a task in which a delayed processing type transaction operates when a real time processing transaction waits for a resource until the delayed processing type transaction completes processing. Are listed.

JP 62-210546 A (first page) JP 2004-148512 A (paragraphs 0034, 0035) JP 10-40117 A (paragraph 0007)

  In the CPU distribution device shown in FIG. 24, when a workload class is set as a group of large similar functions (for example, when a workload class is divided into batch work, online work, development work, etc.), the system shared resource Even if CPU resources were given over the acquisition interval, the over-CPU time did not deviate significantly from the distribution setting value.

  However, as the performance of a single CPU has been improved, a lot of work has been performed with one OS, and the need to finely divide similar functions into several workload classes has increased. For example, it may be necessary to divide workload classes such as batch business, online business 1, online business 2, online business 3, and development business. If the CPU allocation setting value becomes small, when the CPU resource is given beyond the lock section of the system shared resource, the ratio deviating from the CPU allocation setting value becomes large, thereby destroying the entire CPU allocation. I came. Furthermore, by dividing the similar function into several workload classes, in addition to the case where the system locks shared resources, there are cases where resources (for example, online business shared resources) are shared among the workload classes. It became so. If the shared resource is limited to a specific job rather than a system, it should be a higher priority within the priority range in which the specific job is executed, and a higher execution priority (ie, higher than all jobs). The execution priority is not set higher than the range of priority that can be set by the user. For this reason, a shared resource between workload classes is used even though another workload class has not used up the CPU allocation time because the CPU allocation time has been used up while a shared resource between workload classes has been acquired in a certain workload class. In some cases, the CPU resources cannot be used while waiting.

  Further, in a method in which the CPU time used by the system is not targeted for allocation to the workload class, the CPU distribution balance is not disrupted due to acquisition of system resources. However, since the system part requested and operated by the process executing in the workload class is not the allocation target, the CPU resource used by the computer for each workload class as a result is different from the allocation ratio. There is a problem of becoming something. In addition, when acquiring shared resources between workload classes outside of the system part, the CPU allocation time has been used while acquiring shared resources between workload classes in a certain workload class, and other workloads have been used. There is also a problem that the CPU resource cannot be used while waiting for the shared resource between workload classes even though the class does not use up the CPU allocation time.

  Therefore, the present invention can prevent the CPU resource from being used while waiting for the shared resource even though the CPU allocation time is not used up, regardless of how the workload class is defined. An object of the present invention is to provide a process control device, a process control method, and a process control program that can suppress a deviation from a CPU distribution ratio due to forced execution of a process.

  The process control apparatus of the present invention does not use up the CPU until the CPU target time when there is no process in the executable process queue that is a queue of executable processes that does not use up the CPU until the CPU target time. A process that operates in the workload class and waits for a resource to be released, a forcible execution target specifying means for specifying a resource securing process that secures the resource, and forcibly executing the resource securing process And a forced execution means that, when requested to release a resource in accordance with the execution of a process by the forced execution means, releases the resource, interrupts the process, and waits for the release of the resource. Resource release means for adding to the queue.

  The process control method of the present invention uses up the CPU until the CPU target time when there is no process in the executable process queue that is a queue of executable processes that does not use up the CPU until the CPU target time. A process that operates in a non-workload class that is waiting for a resource to be released and a resource securing process that secures the resource, forcibly execute the resource securing process, and When a resource release is requested in accordance with execution, the resource is released, the process is suspended, and a release waiting process waiting for the resource release is added to the executable process queue.

  Also, the process control program of the present invention can achieve up to the CPU target time when there is no process in the executable process queue that is a queue of executable processes that has not used up the CPU until the CPU target time. A process that operates in a workload class that does not use up the CPU and that waits for a resource to be released, a forcible execution target specifying process that specifies a resource securing process that secures the resource, and the resource securing process Forcible execution processing to be forcibly executed, and when a resource release request is generated in accordance with execution of a process in the forcible execution processing, the resource is released, the process is interrupted, and the release wait waiting for the resource to be released It is characterized by executing a resource release process that adds a process to the executable process queue. .

  According to the present invention, it is possible to prevent the CPU resource from being in a state where the CPU resource cannot be used while waiting for the shared resource even though the CPU allocation time is not used up, regardless of how the workload class is determined. The deviation from the CPU distribution ratio due to the forced execution of the process can be suppressed.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing an example of a process control apparatus of the present invention. The process control device is realized by the computer 10. The process control apparatus 10 includes a process correspondence table 101, a CPU distribution table 102, a target time storage table 103, a usage time storage table 104, a process management table 105, an executable process queue storage means 106, and a resource management table. 107, forced execution information storage table 108, target time calculation means 110, process selection means 120, forced execution process selection means 130, process execution means 140, interval initialization means 150, resource acquisition means 160, Resource release means 170.

  The process correspondence table 101 is a storage unit that stores the workload class number of each process. FIG. 2 is an explanatory diagram showing the process correspondence table 101. As shown in FIG. 2, the process correspondence table 101 stores the workload class number of each process in the order of the process number. The workload class is a CPU allocation unit, and the workload class number is identification information of each workload class. The process number is identification information of each process.

  The CPU distribution table 102 is a storage unit that stores the CPU distribution ratio of each workload class. FIG. 3 is an explanatory diagram showing the CPU distribution table 102. As shown in FIG. 3, the CPU distribution table 102 stores CPU distribution ratios in the order of workload class numbers. The CPU distribution ratio is, for example, a value expressed as a percentage, but may be expressed as a percentage other than a percentage.

  The target time storage table 103 is storage means for storing the CPU target time of each workload class. The CPU target time is a time allocated to each workload class according to the CPU distribution ratio. FIG. 4 is an explanatory diagram showing the target time storage table 103. As shown in FIG. 4, the target time storage table 103 stores CPU target times in the order of workload class numbers.

  The usage time storage table 104 is storage means for storing the CPU usage time of each workload class. The CPU usage time is the time for each workload class that is actually executed by the CPU. FIG. 5 is an explanatory diagram showing the usage time storage table 104. As shown in FIG. 5, the CPU usage time is stored in the usage time storage table 104 in the order of the workload class number.

  The process management table 105 is a storage unit that stores information indicating the state of each process. FIG. 6 is an explanatory diagram showing the process management table 105. As shown in FIG. 6, the process management table 105 stores a set of an executable flag, a resource waiting flag, an execution interruption flag, an execution priority, and a resource number waiting for resources in the order of process numbers. The executable flag is a flag indicating whether or not execution is possible, and represents that execution is possible in a set state (ON). The resource waiting flag is a flag indicating whether or not the resource (shared resource) is waiting, and indicates that the resource waiting state is set. The execution interruption flag is a flag indicating whether or not there is a request to interrupt the execution, and represents that there is a request to interrupt the execution in the set state. A state where various flags are not set is described as OFF.

  The resources for which a process waits for release include not only system resources but also resources shared among workload classes.

  The resource number is identification information for identifying each resource.

  FIG. 7 is an explanatory diagram showing the executable process queue storage unit 106. The executable process queue storage unit 106 is a storage unit in which executable processes are queued in the order of execution priority. Specifically, the executable process queue storage means 106 stores information on process numbers and execution priorities queued in the order of execution priority.

  FIG. 8 is an explanatory diagram showing the resource management table 107. The resource management table 107 is a storage unit that stores process numbers of tenant processes that have secured resources in the order of resource numbers and queues processes waiting for resources in the order of execution priority. As shown in FIG. 8, the process number of the tenant process and the process number waiting for the resource are stored in the order of resource numbers such as “0” and “1”.

  FIG. 9 is an explanatory diagram showing the forced execution information storage table 108. The forced execution information storage table 108 is a storage unit that stores a resource number and a process number of a process that preferentially delivers a resource when releasing the resource.

  The target time calculation means 110 calculates the CPU time that can be used by each workload class at a fixed interval time based on the CPU distribution ratio for each workload class stored in the CPU distribution table 102, and the target time Store in the storage table 103.

  The process selection unit 120 searches the processes queued in the executable process queue storage unit 106 in order from the top, and identifies the processes that have not used up the CPU until the CPU target time stored in the target time storage table 103. To do. If such a process is specified, the process is selected and control is passed to the process execution means 140. Also, if the process that has not used up the CPU until the CPU target time does not exist until the last queued process, control is passed to the forced execution process selection means 130. The process selection unit 120 determines whether the process has used up the CPU until the CPU target time as follows. That is, the workload class number of the process queued in the executable process queue storage means 106 is acquired from the process correspondence table 101, and the CPU usage time and CPU target time of the workload class number are obtained as the usage time storage table 104. Then, referring to the target time storage table 103 and comparing them, it may be determined whether or not the process has used up the CPU until the CPU target time.

  The forcible execution process selection unit 130 performs the following process when a process that has not used up the CPU until the CPU target time does not exist in the process queued in the executable process queue storage unit 106. The forced execution process selection means 130 refers to the target time storage table 103 and the usage time storage table 104 and searches for a workload class that does not use up the CPU until the CPU target time. Further, the forced execution process selection unit 130 refers to the process correspondence table 101 and searches for a process that operates in the workload class. Then, when the process state stored in the process management table 105 is waiting for a resource, the forced execution process selection unit 130 obtains a target resource number and a process number to which a resource is preferentially delivered when the resource is released. The process number of the process which is stored in the forced execution information storage table 108 and the target resource is secured is acquired from the resource management table 107, and control is passed to the process execution means 140.

  The process execution means 140 is acquired by the forced execution process selection means 130 from the resource management table 107 as a process selected by the process selection means 120 as a process that does not use up the CPU until the CPU target time, or as a process that secures resources. The following processing is performed on the processed process. The process execution unit 140 dequeues these processes from the executable process queue storage unit 106. Then, the process execution unit 140 executes the process until the CPU usage time of the workload class of the process reaches the CPU target time or the execution interruption flag in the process management table 105 is set. The execution time is added to the CPU usage time in the usage time storage table 104 of the workload class of the process. The process execution means 140 deletes the resource number registered in the forced execution information storage table 108 after completion of the process execution.

  The interval initialization unit 150 operates at regular intervals, and sets the CPU usage time used by each workload class in the usage time storage table 104 to zero.

  When resource acquisition (reservation) is requested from a process, the resource acquisition unit 160 refers to the resource management table 107 and determines whether or not the resource has already been acquired. If the resource has not yet been acquired, the resource acquisition unit 160 stores the process number of the process that requested acquisition of the resource (resource acquisition request process) in the resource management table 107 as the process number of the tenant process. If the resource has already been acquired, the resource acquisition request process is connected to the resource waiting process queue in the resource management table 107, the resource waiting flag in the process management table 105 is set, and the process management table 105 The number of the resource already acquired is set as the resource number waiting for the resource. Note that “connecting a process to a queue” means adding a process to the queue.

  The resource release means 170 releases the resources, and if the number of the released resources matches the resource number of the forced execution information storage table 108, the resource release means 170 executes the process stored in the forced execution information storage table 108 as a resource. Dequeue from the resource waiting process queue in the management table 107. If the resource numbers do not match, the process at the head of the resource waiting process queue in the resource management table 107 is dequeued. Then, the resource release unit 170 sets the dequeued process number to the process number of the tenant process in the resource management table 107. Further, the executable flag in the process management table 105 of the dequeued process is set, the resource waiting flag is reset, and the process is connected to the executable process queue 106. If the released resource matches the resource number of the forced execution information storage table 108, the resource releasing unit 170 finally sets the execution suspension flag and the executable flag in the process management table 105 of the resource release request process, The resource release request process is queued in the executable process queue 106, and control is returned to the process execution means 140.

  The process correspondence table 101, CPU distribution table 102, target time storage table 103, usage time storage table 104, process management table 105, executable process queue storage means 106, resource management table 107, and forced execution information storage table 108 are, for example, This is realized by a storage device included in the computer 10. The target time calculation unit 110, the process selection unit 120, the forced execution process selection unit 130, the process execution unit 140, the interval initialization unit 150, the resource acquisition unit 160, and the resource release unit 170 are, for example, a program (process control program). This is realized by the CPU of the computer 10 that operates in accordance with (1). For example, the CPU of the computer 10 reads the process control program, and according to the program, the target time calculation means 110, the process selection means 120, the forced execution process selection means 130, the process execution means 140, the interval initialization means 150, the resource acquisition means 160 and resource release means 170.

Next, the operation of the present invention will be described.
First, the operation of calculating the CPU target time for each workload class and storing it in the target time storage table 103 will be described. In the CPU distribution table 102, a CPU distribution ratio for each predetermined workload class is set. FIG. 10 is a flowchart illustrating an example of an operation for storing the CPU target time in the target time storage table 103.

  First, the target time calculation unit 110 reads the CPU distribution ratio for each workload class stored in the CPU distribution table 102 (step S111), and calculates the total value of the CPU distribution ratios for all workload classes (step S112). ). As already described, the CPU distribution ratio is, for example, a value expressed as a percentage, but may be expressed as other than a percentage.

  Next, the target time calculation means 110 calculates the CPU target time for each workload class (step S113). The target time calculation means 110 calculates the CPU target time for each workload class by multiplying the “certain interval time” by the ratio of the CPU distribution ratio to the total value obtained in step S112 for each workload class. To do. The “fixed interval time” is a cycle in which the interval initialization unit 150 initializes the CPU usage time stored in the usage time storage table 104 to zero. Assuming that this fixed interval time is I, the total value of CPU distribution ratios of all workload classes calculated in step S112 is S, and the CPU distribution ratio of the workload classes for which CPU target time is calculated is R, the target The time calculation means 110 calculates the CPU target time of each workload class by calculating I × R / S for each workload class. As described above, the target time calculation unit 110 calculates the CPU target time of each workload class by allocating a certain interval time according to the CPU distribution ratio.

  Next, the target time calculation unit 110 stores the calculated CPU target time of each workload class in the target time storage table 103 (step S114). At this time, as shown in FIG. 4, CPU target times are stored in the order of workload class numbers.

  Next, the flow of control from process selection to process execution will be described. FIG. 11 is an explanatory diagram showing the flow of control from process selection to process execution.

  The process selection unit 120 selects an executable process. Then, the process selection unit 120 passes control to the process execution unit 140 when the executable process can be selected, and controls to the forced execution process selection unit 130 when the executable process cannot be selected. give. The operation of the process selection unit 120 will be described later with reference to FIG.

  The forced execution process selection unit 130 selects a process to be forcibly executed. When the process to be forcibly executed can be selected, the forced execution process selection unit 130 transfers control to the process execution unit 140. When the process to be forcibly executed cannot be selected, the process selection unit 120 is selected. Pass control to. The operation of the forced execution process selection unit 130 will be described later with reference to FIGS.

  The process execution unit 140 executes the process selected by the process selection unit 120 or the forced execution process selection unit 130 and passes control to the process selection unit 120. The operation of the process execution unit 140 will be described later with reference to FIGS. 16 and 17.

  Next, the operation of the process selection unit 120 will be described. FIG. 12 is a flowchart illustrating an example of the operation of the process selection unit 120. The process selection unit 120 acquires the process number of the process queued at the head in the executable process queue storage unit 106 (step S121). In other words, the process number having the highest priority among the process numbers stored in the executable process queue storage unit 106 is acquired. Here, the process number acquired by the process selection unit 120 is n.

  Next, the process selection unit 120 acquires the workload class number corresponding to the acquired process number n (that is, the workload class number of the process with the process number n) from the process correspondence table 101 (step S122). When the process proceeds from step S121 to step S122, the process selection unit 120 acquires the workload class number corresponding to the process number acquired in step S121 from the process correspondence table 101. Further, when the process proceeds from step S127 described later to step S122, the process selection unit 120 acquires a workload class number corresponding to the process number acquired in step S127 from the process correspondence table 101.

  After step S122, the process selection unit 120 acquires the CPU target time corresponding to the workload class number acquired in step S122 from the target time storage table 103 (step S123). That is, the CPU target time of the workload class indicated by the workload class number is acquired.

  Next, the process selection unit 120 acquires the CPU usage time corresponding to the workload class number acquired in step S122 from the usage time storage table 104 (step S124). That is, the CPU usage time of the workload class indicated by the workload class number acquired in step S122 is acquired.

  The process selection unit 120 determines whether or not the CPU usage time is shorter than the CPU target time (step S125). If the CPU usage time is shorter than the CPU target time (Y in step S125), control is passed to the process execution means 140. When the CPU usage time is equal to or longer than the CPU target time (N in Step S125), the process selection unit 120 determines whether or not the next process is queued in the executable process queue storage unit 106. Determination is made (step S126).

  When the next process is queued in the executable process queue storage unit 106 (Y in step S126), the process selection unit 120 displays the process number of the next process queued in the executable process queue storage unit 106. Is acquired (step S127), and the process proceeds to step S122. And the process after step S122 is performed about the process number acquired by step S127.

  When the next process is not queued in the executable process queue storage means 106 (N in step S126), there is a process whose CPU usage time is less than the CPU target time until the last queued process. This is the case. In other words, a process that has not used up the CPU usage time until the CPU target time is not queued in the executable process queue storage unit 106. In this case, the process selection unit 120 passes control to the forced execution process selection unit 130.

  Next, the operation of the forced execution process selection unit 130 will be described. 13, 14, and 15 are flowcharts illustrating an example of the operation of the forced execution process selection unit 130. In the following description, for example, a workload class whose workload class number is i is simply referred to as a workload class i. The workload class number corresponding to the process number x is the workload class number of the process whose process number is x.

  The forced execution process selection unit 130 sets 0 to the variable i (step S231). Next, the forced execution process selection unit 130 acquires the CPU target time of the workload class i from the target time storage table 103 (step S232), and acquires the CPU usage time of the workload class i from the usage time storage table 104. (Step S233). Then, it is determined whether the CPU usage time is less than the CPU target time (step S234). If the CPU usage time is equal to or longer than the CPU target time (N in step S234), the process proceeds to step S246 (see FIG. 15), and if the CPU usage time is less than the CPU target time (Y in step S234), the process proceeds to step S235. Transition.

  In step S235, the forced execution process selection unit 130 sets 0 to the variable j.

  Next, the forcible execution process selection unit 130 refers to the process correspondence table 101, acquires the workload class number corresponding to the process number j (step S236), and determines whether or not the workload class number is equal to the variable i. Is determined (step S237). That is, it is determined whether or not the CPU usage time is less than the CPU target time and is equal to the focused workload class number i. If the workload class number corresponding to the process number j is not equal to the variable i (N in step S237), the process proceeds to step S244 (see FIG. 14). When the workload class number corresponding to the process number j is equal to the variable i (Y in step S237), j is a process operating in the target workload class whose CPU usage time is less than the CPU target time. Process number. In this case (Y in step S237), the process proceeds to step S238.

  In step S238, the forced execution process selection unit 130 sets the value of j to the variable k.

  Next, the forced execution process selection unit 130 determines whether or not the resource waiting flag for the process number k is on in the process management table 105 (step S239). That is, it is determined whether or not a process operating in the target workload class whose CPU usage time is less than the CPU target time is in a resource waiting state. If the resource wait flag for the process number k stored in the process management table 105 is off (N in step S239), the process proceeds to step S244, and if it is on (Y in step S239), the process proceeds to step S240. To do.

  In step S240, the forced execution process selection unit 130 sets the resource number of the process number k stored in the process management table 105 (resource number waiting for resources; see FIG. 6) in the variable m. That is, the resource number of the resource waiting for the process with the process number k is set to m.

  Next, the forced execution process selection means 130 sets the process number of the tenant process of the resource number m stored in the resource management table 107 to the variable n (step S241). That is, the process number of the process that secures the resource of the target resource number m is set to n (step S241). As described above, the process number of the process that secures the resources waiting for the process operating in the workload class of interest whose CPU usage time is less than the CPU target time is n.

  Next, the forced execution process selection unit 130 refers to the process management table 105 and determines whether the executable flag of the process number n (the executable flag of the process that secures the resource of the resource number m) is on. Is determined (step S242). If the executable flag is on (Y in step S242), the process proceeds to step S248. If it is off (N in step S242), the forced execution process selection unit 130 sets the value of n to the variable k (step S243), and repeats the processing from step S239 onward.

  If it is determined in step S242 that the executable flag is on, n at that time secures resources waiting for processes operating in the target workload class whose CPU usage time is less than the CPU target time. This means the process number of a process in which the executable flag is on.

  In step S244, the forced execution process selection unit 130 determines whether or not the process number j is the last entry (step S244). That is, it is determined whether j has reached the maximum process number. If the process number j is the last entry (Y in step S244), the process proceeds to step S246 (see FIG. 15). If the process number j is not the last entry (N in step S244), the forced execution process selection unit 130 increments the value of j by 1 (step S245), and performs the processing from step S236 onward for that j.

  In step S246, the forced execution process selection unit 130 determines whether the workload class number i is the last entry in the target time storage table 103 (step S246). If the workload class number i is the last entry (Y in step S246), control is passed to the process selection unit 120. If the workload class number i is not the final entry (N in step S246), the value of i is incremented by 1 (step S247), and the processing subsequent to step S232 (see FIG. 13) is performed for that i.

  In addition, when the process proceeds from step S242 to step 248, a resource that is waiting for a process that operates in a focused workload class whose CPU usage time is less than the CPU target time is being secured, and an executable flag is set. The process number (n) of the process that is turned on is specified. The forced execution process selection means 130 stores the resource number m as the resource number of the forced execution information storage table 108 (step S248). Further, the forced execution process selection unit 130 stores the value of k in the forced execution information storage table 108 as the process number (see FIG. 9) of the process that preferentially delivers resources when releasing resources (step S249). Then, control is passed to the process execution means 140.

  In the above operation, the processing in step S239 indicates that the resource wait flag is off means that the process operating in the target workload class whose CPU usage time is less than the CPU target time waits for resources. It means not. In this case, steps S240, S241, S242, S248, and S249 are not performed to transfer control to the process execution means 140, but the process proceeds to step S244. If j is not the final entry, j is incremented by 1 ( Steps S245) and S236 are repeated. If j is the final entry, it is determined whether i is the final entry, i is incremented by 1, and the processing from step S232 is repeated. As described above, when the process operating in the target workload class whose CPU usage time is less than the CPU target time is not waiting for the resource, the process execution unit 140 is forced to execute the process. I won't do that.

  Next, the operation of the process execution unit 140 will be described. 16 and 17 are flowcharts showing an example of the operation of the process execution means 140. The process execution unit 140 acquires the workload class number of the process selected by the process selection unit 120 or the forced execution process selection unit 130, that is, the process of the process number n from the process correspondence table 101, and sets the workload class number. The variable p is set (step S341).

  That is, in step S125 (see FIG. 12) executed by the process selection unit 120, if it is determined that the CPU usage time is less than the CPU target time and control is passed from the process selection unit 120, the process is performed at that time. The workload class number corresponding to the process number n acquired by the selection means 120 is set in the variable p.

  Further, in step S242 (see FIG. 15) executed by the forced execution process selection unit 130, the executable flag of the process of the process number n that secures the resource of the resource number m is on, and steps S248 and S249 (FIG. 16), if the control is passed from the forced execution process selection means 130, the workload class number corresponding to the process number n specified by the forced execution process selection means 130 at that time is set to the variable p. set.

  After step S341, the process execution means 140 acquires the CPU target time of the workload class p from the target time storage table 103, sets it to the variable q (step S342), and sets the CPU usage time of the workload class p as the usage time. Obtained from the storage table 104 and set to the variable r (step S343).

  Then, the process execution unit 140 dequeues n from the process numbers queued in the executable process queue storage unit 106 (step S344), and the executable flag of the process number n stored in the process management table 105 is obtained. Is turned off (step S345). Further, the process execution means 140 turns off the execution interruption flag of the process number n stored in the process management table 105 (step S346). Step S347 to be described next is a step of executing the process. In steps S344 to S346, the process number n of the process to be executed is dequeued from the executable process queue storage means 106, its executable flag and execution interruption are performed. All flags are off.

  After step S346, the process execution unit 140 executes the process with the process number n for the minimum unit (step S347).

  When the forced execution process selection unit 130 performs the processes of steps S242, S248, and S249 and is given control from the forced execution process selection unit 130, the resource of the resource number m is secured by executing step S347 in this way. The process number n is forcibly executed.

  However, step S347 is not only a step for forcibly executing the process of the process number n that secures the resource of the resource number m. When control is passed from the process selection unit 120 and the process proceeds to step S347, a process that does not use the CPU until the CPU target time (process selected by the process selection unit 120) is executed for the minimum unit.

  After step S347, the process execution means 140 adds the time executed in step S347 to the CPU usage time of the workload class p stored in the usage time storage table 104, and the CPU of the workload class p. The usage time is updated (step S348).

  Next, the process execution unit 140 checks whether or not the execution interruption flag for the process number n stored in the process management table 105 is turned on (step S349). If the execution interruption flag of the process number n is on (Y in step S349), the process proceeds to step S354, and if it is off (N in step S349), the process proceeds to step S350.

  In step S350, the process execution unit 140 determines whether or not the value of the variable r is greater than or equal to the value of the variable q (step S350). If the value of the variable r is greater than or equal to the value of the variable q (Y in step S350), the processes in and after step S347 are repeated. That is, if the CPU usage time set in the variable r in step S343 has already exceeded the CPU target time, the process with the process number n is executed until the execution interruption flag is turned on.

  If the value of the variable r is less than the value of the variable q (N in step S350), the process execution unit 140 uses the CPU usage time of the workload class p stored in the usage time storage table 104 (in step S347). It is determined whether or not the CPU usage time to which the executed time is added is less than the CPU target time (step S351). If the CPU usage time is less than the CPU target time, the processing from step S347 is repeated, and if the CPU usage time is equal to or longer than the CPU target time, the process proceeds to step S352.

  In step S352, the process execution unit 140 turns on the execution flag of the process number stored in the process management table 105 (step S352). Next, the process execution unit 140 enqueues the process numbers n into the executable process queue storage unit 106 in order of execution order (step S353). Then, the process execution unit 140 deletes the resource number registered in the forced execution information storage table 108 (step S354).

  When the process execution unit 140 executes the process with the process number n in the above step S347, when the resource release unit 170 requests the resource release, the resource release unit 170 sets an execution interruption flag for the process. Turned on and the process is interrupted. When control is returned from the resource release unit 170 that has released the resource, the process execution unit 140 performs step S348, and since the execution interruption flag is turned on in the subsequent step S349, the process execution unit 140 does not return to step S347. Therefore, the process of process number n is interrupted.

  Next, the operation of the interval initialization unit 150 will be described. FIG. 18 is a flowchart showing an example of the operation of the interval initialization unit 150. When the interval initialization means 150 sets 0 for the CPU usage time of all the workload classes in the usage time storage table 104 (step S151), it waits for a certain interval time to elapse (step S152). When the certain interval time has elapsed, the processing after step S151 is repeated.

  Next, operations of the resource acquisition unit 160 and the resource release unit 170 will be described. The resource acquisition unit 160 and the resource release unit 170 are activated by the process execution unit 140 while the process execution unit 140 is executing the process in step S347. When the process execution unit 140 requests acquisition of a resource in step S347, the resource acquisition unit 160 is activated, the resource number of the resource to be acquired is specified, and the resource is acquired. At this time, the process executed by the process execution unit 140 is referred to as a resource acquisition request process. When the process execution unit 140 requests the resource release in step S347, the resource release unit 170 is activated, the resource number of the resource to be released is designated, and the resource is released. At this time, the process executed by the process execution unit 140 is referred to as a resource release process.

  FIG. 19 is a flowchart illustrating an example of the operation of the resource acquisition unit 160. When the resource acquisition means 160 is designated by the resource number, the resource acquisition means 160 sets the resource number in the variable s (step S161).

  Next, the resource acquisition unit 160 checks whether or not the process number is stored in the process number storage area of the tenant process of the resource number s in the resource management table 107 (step S162). If the process number is not stored in the process number storage area of the tenant process (N in step S162), the process proceeds to step S163. If the process number is already stored (Y in step S162), the process proceeds to step S164.

  In step S163, the resource acquisition unit 160 stores the resource acquisition request process number in the process number storage area of the tenant process of the resource number s in the resource management table 107 (step S163), and ends the resource acquisition unit 160. .

  When the process proceeds to step S164, the resource acquisition unit 160 queues the resource acquisition request processes in the resource waiting process queue of the resource number s in the resource management table 107 in the order of execution priority (step S164). . That is, the process numbers of the resource acquisition request processes are stored in order of execution priority. Next, the resource acquisition unit 160 turns on the execution interruption flag of the resource acquisition request process in the process management table 105 (step S165), and turns on the resource waiting flag of the resource acquisition request process in the process management table 105. Set (step S166). Further, the variable s is set as the resource number of the resource acquisition request process in the process management table 105 (resource number waiting for the resource; see FIG. 6), and control is returned to the process execution means 140 (step S167).

  20 and 21 are flowcharts showing an example of the operation of the resource releasing unit 170. When the resource to be released is designated by the resource number, the resource releasing unit 170 sets the resource number to the variable t (step S171). Next, the resource release unit 170 deletes the process number stored in the process number storage area of the tenant process of the resource number t in the resource management table 107 (step S172). In this way, by deleting the process number stored in the process number storage area of the tenant process, the resource (resource requested to be released) secured in the process executed by the process execution unit 140 is released. To do.

  Next, the resource release unit 170 determines whether or not the resource number stored in the forced execution information storage table 108 matches the variable t (step S173). If they do not match, the process proceeds to step S175, and if they match, the process proceeds to step S174.

  In step S174, the resource release unit 170 sets the “process number of the process to which the resource is preferentially delivered when the source is released (see FIG. 9)” in the forced execution information storage table 108 in the variable u (step S174).

  In step S175, the resource releasing unit 170 checks whether or not there is a process in the resource waiting process queue of the resource number t in the resource management table 107 (step S175). If it does not exist (N in step S175), the resource releasing unit 170 is terminated. If it exists (Y in step S175), the resource releasing unit 170 sets the process number of the first process in the resource waiting process queue of the resource number t in the resource management table 107 in the variable u (step S176). .

  After step S174 and after step S176, the resource releasing unit 170 dequeues the process number u from the resource waiting process queue of the resource number t in the resource management table 107 (step S177). That is, after step S174, “the process number of the process that gives priority to resources when releasing the source” is dequeued, and after step S176, the process number of the first process in the resource waiting process queue of resource number t is dequeued. To do.

  Next, the resource release unit 170 sets a variable u in the process number storage area of the tenant process with the resource number t in the resource management table 107 (step S178).

  Subsequently, the resource releasing unit 170 turns off the resource waiting flag for the process number u in the process management table 105 (step S179), and further turns on the executable flag for the process number u in the process management table 105 (step S179). Step S180). Next, the resource release unit 170 queues the process numbers u in the executable process queue of the executable process queue storage unit 106 in the order of execution priority (step S181).

  Next, the resource release unit 170 determines whether or not the resource number in the forced execution information storage table 108 matches the variable t (step S182). If they do not match, the resource release unit 170 ends. .

  If they match, the resource release means 170 sets the execution flag of the resource release process in the process management table 105 to ON (step S183), and further executes the resource release process executable flag in the process management table 105. Is set to ON (step S184), the resource release processes are queued in the executable process queue storage means 106 in order of execution priority, and control is passed to the process execution means 140 (step S185).

  In the processing of steps S171 to S185, the resource release unit 170 releases the resource requested to be released by deleting the process number stored in the process number storage area of the tenant process in step S172. In addition, by dequeueing the process number u in step S177, the process waiting for the release of the resource is dequeued from the resource waiting process queue. In step S181, the process number u is queued in the executable process queue, so that the process in which the waiting resource is released is queued in the executable process queue.

  In steps S183, S184, and S185, the resource release process is suspended by turning on the execution suspension flag and the executable flag of the resource release process and queuing the resource release process in the executable process queue. After step S185, when the process execution unit 140 that has executed the resource release process returns control from the resource release unit 170, the process execution unit 140 performs steps S348 and S349 (see FIGS. 16 and 17). After confirming that the execution interruption flag of the release process is turned on, the process exits the loop process of steps S347 to S351 and does not return to step S347, so that the interrupted process is not continuously executed.

  Next, a specific example of the operation of the present invention will be described. FIG. 22 is an explanatory diagram showing a specific example of the state of each table. In the following description, “0” is stored as the workload class number corresponding to process number 0 and “1” is stored as the workload class number corresponding to process number 1 in the process correspondence table 101. Take an example. Further, a case where the CPU distribution ratio corresponding to the workload class 0 is defined as 80% and the CPU distribution ratio corresponding to the workload class 1 is defined as 70% in the CPU distribution table 102 is taken as an example. Further, in the example shown below, it is assumed that the interval time for CPU allocation (cycle in which the interval initialization unit 150 initializes the CPU usage time of all workload classes to 0) is 100 milliseconds.

  A specific example of the operation of the target time calculation unit 110 will be described with reference to FIG. The target time calculation means 110 reads the CPU distribution ratio “80” of the workload class 0 and the CPU distribution ratio “20” of the workload class 1 from the CPU distribution table 102 (step S111), and the CPU of each workload class. The total value of the distribution ratio is calculated (step S112). Since the CPU distribution ratios of the workload classes 0 and 1 are 80 and 20, respectively, in this example, 80 and 20 are added to calculate 100 as the calculation result. Next, since the interval time is 100 milliseconds, the CPU target times of the workload classes 0 and 1 are calculated as 100 × 80/100 = 80 milliseconds and 100 × 20/100 = 20 milliseconds, respectively (step) S113). Then, the target time calculation unit 110 sets the CPU target times of the workload classes 0 and 1 as 80 milliseconds and 20 milliseconds, respectively, in the target time storage table 103 (step S114).

  A specific example of the operation of the resource acquisition unit 160 will be described with reference to FIG. Assume that a process having a resource number 0 is requested by a process having a process number 1. That is, it is assumed that the process number of the resource request process is 1. The resource acquisition unit 160 sets the resource number 0 to be acquired to the variable s (step S161). As shown in FIG. 22, it is assumed that process number 0 is stored in the process number storage area of the tenant process with resource number 0 in the resource management table 107. Then, since the process number of the tenant process has already been stored in the resource acquisition unit 160 (Y in step S162), the process 1 is placed in the resource waiting process queue of the resource number 0 in the resource management table 107 in the order of execution priority. Then, queuing is performed (step S164). Next, the resource acquisition unit 160 sets the execution suspension flag and resource wait flag of process number 1 in the process management table 105 to ON (steps S165 and S166), and the resource number of process number 1 in the process management table 105. Resource number 0 is set in (resource number waiting for resource) (step S167). Then, control is returned to the process execution means 140.

  Next, a specific example of the operation of the process selection unit 120 will be described with reference to FIG. In this example, it is assumed that the process with the process number 0 is queued as an executable process in the executable process queue storage unit 106 and there are no other executable processes queued. Further, it is assumed that the CPU usage time of workload class number 0 is 80 milliseconds and the CPU usage time of workload class number 1 is 10 milliseconds.

  In step 121, the process selection unit 120 acquires the process number 0 of the process queued in the executable process queue. Next, the workload class number of process number 0 (in this example, the workload class number is 0) is acquired from the process correspondence table 101 (step S122). Then, the process selection unit 120 acquires the CPU target time 80 milliseconds of the workload class number 0 from the target time storage table 103 (step S123), and obtains the CPU usage time 80 milliseconds of the workload class number 0. And obtained from the usage time storage table 104 (step S124). Then, since the acquired CPU usage time and CPU target time are equal (N in step S125), the process proceeds to step S126, and there is no other executable process queued in the executable process queue storage unit 106. Control is passed to the execution process selection means 130.

  Next, a specific example of the operation of the forced execution process selection unit 130 will be described with reference to FIGS. 13, 14, and 15. The forced execution process selection unit 130 sets 0 to the variable i (step S231). Then, the CPU target time 80 milliseconds of the workload class 0 is acquired from the target time storage table 103 (step S232), and the CPU usage time 80 milliseconds of the workload class 0 is acquired from the usage time storage table 104 (step S233). ). Since the CPU usage time is equal to or longer than the CPU target time (N in Step S234), the process proceeds to Step S246, and the forced execution process selection unit 130 determines whether or not the workload class number 0 is the final entry (Step S234). S246). Since the workload class number is 1 and the workload class number 0 is not the final entry (N in step S246), the process proceeds to step S247, and the forced execution process selection unit 130 increases the value of i by 1, and i = 1 and the process proceeds to step S232 (step S247).

  Subsequently, the forced execution process selection unit 130 acquires the CPU target time 20 milliseconds for the workload class 1 from the target time storage table 103 (step S232), and the CPU usage time for the workload class 1 from the usage time storage table 104. 10 milliseconds is acquired (step S233). Then, since the CPU usage time is less than the CPU target time (Y in step S234), the forced execution process selection unit 130 sets the variable j to 0 (step S235).

  Subsequently, the forced execution process selection unit 130 refers to the process correspondence table 101 and acquires the workload class number 0 of the process number j (0 in this case) (step S236). Since the acquired workload class number 0 is not the value of the variable i (here, 1) (N in step S237), the process proceeds to step S244. In step S244, it is determined whether or not the process number j (= 0) is the last entry in the process correspondence table 101. Here, since process number 0 is not the last entry (N in step S244), the forced execution process selection unit 130 increments the value of j by 1 and sets j = 1, and proceeds to step S236.

  The forced execution process selection unit 130 refers to the process correspondence table 101 and acquires the workload class number 1 of the process number j (here, 1) (step S236). Since the acquired workload class number 1 is equal to 1 which is the value of the variable i (Y in step S237), the forced execution process selection unit 130 sets the value 1 of j to the variable k (step S238).

  Next, it is determined whether or not the resource wait flag of process number 1 stored in the process management table 105 is on (step S239). Since the resource wait flag for process number 1 is on (Y in step S239), the forced execution process selection means 130 is the resource number waiting for resources corresponding to process number 1 stored in the process management table 105. 0 is set to the variable m (step S240). Further, the process number 0 of the tenant process with the resource number 0 stored in the resource management table 107 is set to the variable n (step S241).

  Next, the forced execution process selection unit 130 checks whether or not the execution flag of the process number n (= 0) is turned on in the process management table 105 (step S242). In this example, it is assumed that the execution flag of process number 0 is on in the process management table 105 (Y in step S242). Therefore, n = 0 at this time is the process number of the process that is securing the resources that the process 1 operating in the workload class 1 is waiting for. Subsequently, the process proceeds to step S248, and m (= 0) is stored in the resource number of the forced execution information storage table 108 (step S248). Further, the forced execution process selection means 130 stores the value of k (here, 1) in the forced execution information storage table 108 as the process number (see FIG. 9) of the process that preferentially delivers resources when releasing resources (see FIG. 9). In step S249, control is passed to the process execution means 140.

  Next, a specific example of the operation of the process execution unit 140 will be described with reference to FIGS. 16 and 17. The process execution unit 140 acquires the workload class number (here, 0) of the process number n (here, n = 0) selected by the forced execution process selection unit 130 from the process correspondence table 101 and sets it in the variable p. (Step S341). The process number n selected by the forced execution process selection unit 130 is n at the time when control is passed from the forced execution process selection unit 130, and the process 1 operating in the workload class 1 waits as described above. This means the process number of the process that is securing the reserved resource.

  After step S341, the process execution unit 140 acquires the CPU target time (80 milliseconds here) of the workload class number p (= 0) from the target time storage table 103 and sets it in the variable q (step S342). Then, the CPU usage time (80 milliseconds here) of the workload class number p (= 0) is acquired from the usage time storage table 104 and set in the variable r (step S343). Since n = 0, the process number 0 is dequeued from the executable process queue 106 (step S344). Further, the execution flag and the execution interruption flag of process number 0 stored in the process management table 105 are turned off (steps S345 and S346).

  Then, the process execution means 140 executes the process number 0 for the minimum execution unit (step S347), and the CPU usage time of the workload class p (= 0) in the usage time storage table 104 is the time executed in step S347. The minutes are added (step S348).

  Next, it is determined whether or not the execution interruption flag of process number 0 stored in the process management table 105 is turned on (step S349). Here, it is assumed that the execution interruption flag of process number 0 remains off (N in step S349). Next, the process execution unit 140 compares the value of the variable r (80 milliseconds) with the value of the variable q (80 milliseconds), and since r = q (Y in step S350), the process execution unit 140 proceeds to step S347. Transition. Thereafter, the processes in steps S347 to S350 are repeated until the execution interruption flag of process number 0 is turned on.

  Next, a specific example of the operation of the resource release unit 170 will be described with reference to FIGS. Assume that the process execution means 140 issues a resource number 0 release request from the process while executing the process number 0 in step S347 (see FIG. 16). That is, it is assumed that the process with process number 0 is a resource release process. Then, the resource release means 170 sets the resource number 0 to be released to the variable t (step S171). Then, the process number 0 stored in the process number storage area of the tenant process of the resource number t (here 0) in the resource management table 107 is deleted (step S172), and the variable t is the forced execution information storage table 108. It is determined whether or not it matches the resource number 0 in the middle (step S173). Here, since the value of the variable t is 0 and the two match (Y in step S173), the resource release unit 170 gives the resource preferentially when the source is released in the forced execution information storage table 108. The process number of the process (see FIG. 9) ”is set in the variable u. In this example, since the process number of the process that preferentially delivers resources when the source is released is 1, u = 1 is set (step S174).

  Here, since t = 0 and u = 1, the resource releasing unit 170 dequeues process number 1 from the resource waiting process queue of resource number 0 in the resource management table 107 (step S177). Further, process number 1 is stored in the process number storage area of the tenant process of resource number 0 in the resource management table 107 (step S178).

  Further, the resource releasing unit 170 turns off the resource waiting flag of process number 1 in the process management table 105 (step S179), and turns on the executable flag of process number 1 (step S180). Further, the process number 1 is queued in the executable process queue storage means 106 in the order of execution priority (step S181).

  Since the resource number in the forced execution information storage table 108 matches the variable t (Y in step S182), the resource release unit 170 uses the resource release process (in this example, process 0) in the process management table 105. ) Is set to ON (steps S183 and S184). The resource release processes are queued in the executable process queue storage unit 106 in the order of execution priority (step S185). That is, the process number 0 is queued so as to be in the order of execution priority. Thereafter, control is returned to the process execution means 140.

  The process execution unit 140 performs the processing from step S348 onward when control is returned from the resource release unit 170 after making a release request for the resource number 0 during the process execution in step S347 (see FIG. 16). The CPU usage time of workload class 0 in the usage time storage table 104 is added by the time executed in step S347 (see step S348, FIG. 16). In subsequent step S349, since the execution suspension flag of process number 0 stored in the process management table 105 is turned on, the process proceeds to step S354, and the resource number 0 registered in the forced execution information storage table 108 is set. Delete and pass control to the process selection means 120.

  Next, a specific example of the operation of the interval initialization unit 150 will be described with reference to FIG. The interval initialization unit 150 sets 0 for the CPU usage time of all workload classes in the usage time storage table 104 (step S151), and waits for 100 milliseconds to elapse (step S152). When 100 milliseconds have elapsed, the processing after step S151 is performed again.

Next, the effect of the present invention will be described.
According to the present invention, a process that is securing resources waiting for a process operating in a workload class whose CPU usage time is less than the CPU target time is forcibly executed. Accordingly, it is possible to prevent the CPU resource from being in a state where the CPU resource cannot be used while waiting for the shared resource even though the CPU allocation time is not used up.

  According to the present invention, by determining whether the resource waiting flag in step S239 is on or off, a process operating in the target workload class whose CPU usage time is less than the CPU target time is It is determined whether or not it is in a waiting state. If the process is not in a resource waiting state (N in step S239), control is not passed to the process execution unit 140, and the process execution unit 140 does not forcibly execute the process (step S347). In other words, even when a resource such as a system resource is acquired (secured), if another workload class is not waiting for the resource to be released, the forced execution is not performed. Therefore, even if the workload class is set finely and the CPU allocation ratio of each workload class is small, the CPU allocation as a result of the forced execution of the system resource acquisition interval and the originally determined CPU allocation ratio Deviation can be reduced.

  Further, even when the process that has secured the resource is forcibly executed, when the process requests the release of the resource, the resource release means 170 interrupts the process (steps S183, S184, S185). As a result, since the execution priority does not execute an extra section until it returns to the priority order range that can be set by the user, the CPU allocation as a result of the forced execution of the system resource acquisition section was originally determined. Deviation from the CPU distribution ratio can be reduced.

  In addition, resources that a process waits for release include not only system resources, but also resources shared among workload classes. Other workload classes include not only system resources but also shared resources between workload classes. When waiting for the release of the resource, the process that secures the resource is forcibly executed (step S347). Therefore, even if similar functions are further divided into several workload classes, resources that are shared among the workload classes can be prevented, and a decrease in throughput due to a lock neck of shared resources between workload classes is prevented. can do.

  Next, the outline of the present invention will be described. FIG. 23 is a block diagram showing an outline of the present invention. The process control apparatus of the present invention includes a forced execution target identification unit 501 (for example, forced execution process selection unit 130), a forced execution unit 502 (for example, process execution unit 140), and a resource release unit 503 (for example, resource release unit). 170).

  The forced execution target specifying unit 501 determines that the process selection unit 120 does not use the CPU until the CPU target time exists in the executable process queue that is a queue of executable processes (for example, the process selection unit 120 performs step S121). Process in which the CPU usage time is less than the CPU target time in the process of S127), a process that operates in a workload class that does not use up the CPU until the CPU target time and waits for the release of resources (For example, the process of process number k when Y is determined in step S242) and the resource reservation process that secures the resource (for example, the process of process number n when Y is determined in step S242) are specified. .

  The forcible execution unit 502 forcibly executes the resource securing process (for example, performs step S347).

  When the resource release unit 503 is requested to release the resource in accordance with the execution of the process from the forced execution unit, the resource release unit 503 releases the resource (for example, performs step S172) and interrupts the process (for example, step S183 to step S183). (S185 is performed), and a release waiting process waiting for the release of the resource is added to the executable process queue (for example, step S181 is performed).

  With the configuration as described above, it is possible to forcibly execute the resource securing process and prevent a situation in which a resource is kept waiting even though the CPU target time is not used up. Further, since the resource release unit 503 interrupts the process that requested the resource release at the time of resource release, even if the workload class is set finely and the CPU allocation ratio of each workload class becomes small, the system resource acquisition interval It is possible to reduce a deviation between the CPU distribution as a result of the forced execution of the CPU and the originally determined CPU distribution ratio.

  Further, in the above embodiment, the forced execution target specifying unit 501 is provided with a resource securing process on the condition that there is a release waiting process waiting for the release of the resource (provided that the result is Y in step S239). Has been disclosed.

  With such a configuration, even when the workload class is set finely and the CPU distribution ratio of each workload class is small, the CPU distribution as a result of forcibly executing the acquisition section of the system resources and the CPU that was originally determined Deviation from the distribution ratio can be reduced.

  It is also disclosed that the resources include resources that are shared between workload classes.

  In addition, there is provided priority process storage means (for example, forced execution information storage table 108) for storing resource identification information of the secured resource and process identification information of a process that preferentially uses the resource when releasing the resource, The forced execution target specifying unit 501 stores the resource identification information of the resource secured by the resource securing process in the priority process storage unit (for example, performing step S248), and prioritizes the process identification information of the process waiting to be released. Is stored in the priority process storage means as the process identification information of the process to be used automatically (for example, step S249 is performed), and the resource release means 503 stores the resource identification information of the released resources and the priority process storage means. When the identification information matches (for example, Y in step S173), it is recorded in the priority process storage means. By the process identification information is queued in the executable process queue processes shown are (e.g., perform step S181 with respect to u which defines in step S174) configuration is disclosed.

  Further, the forcible execution unit 502 determines that there is a process that does not use up the CPU until the CPU target time in the executable process queue (for example, the process selection unit 120 performs the CPU usage time in steps S121 to S127). If a process of less than time is selected), a configuration for executing that process is disclosed.

  The present invention is preferably applied to process control in which resources are allocated for each workload.

It is a block diagram which shows the example of the process control apparatus of this invention. It is explanatory drawing which shows a process corresponding | compatible table. It is explanatory drawing which shows a CPU distribution table. It is explanatory drawing which shows a target time storage table. It is explanatory drawing which shows a usage time storage table. It is explanatory drawing which shows a process management table. It is explanatory drawing which shows the executable process queue memory | storage means. It is explanatory drawing which shows a resource management table. It is explanatory drawing which shows a forced execution information storage table. It is a flowchart which shows the example of the operation | movement which memorize | stores CPU target time in a target time storage table. It is explanatory drawing which shows the flow of control from process selection to process execution. It is a flowchart which shows the example of operation | movement of a process selection means. It is a flowchart which shows the example of operation | movement of the forced execution process selection means. It is a flowchart which shows the example of operation | movement of the forced execution process selection means. It is a flowchart which shows the example of operation | movement of the forced execution process selection means. It is a flowchart which shows the example of operation | movement of a process execution means. It is a flowchart which shows the example of operation | movement of a process execution means. It is a flowchart which shows the example of operation | movement of an interval initialization means. It is a flowchart which shows the example of operation | movement of a resource acquisition means. It is a flowchart which shows the example of operation | movement of a resource release means. It is a flowchart which shows the example of operation | movement of a resource release means. It is explanatory drawing which shows the specific example of the state of each table. It is a block diagram showing the outline of the present invention. It is a block diagram which shows the example of the CPU distribution apparatus relevant to this invention. It is a flowchart which shows operation | movement of the process selection means of the CPU distribution apparatus shown in FIG. It is a flowchart which shows operation | movement of the process selection means of the CPU distribution apparatus shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Process control apparatus 101 Process correspondence table 102 CPU allocation table 103 Target time storage table 104 Usage time storage table 105 Process management table 106 Executable process queue storage means 107 Resource management table 108 Forced execution information storage table 110 Target time calculation means 120 Process Selection means 130 Forced execution process selection means 140 Process execution means 150 Interval initialization means 160 Resource acquisition means 170 Resource release means 501 Forced execution target identification means 502 Forced execution means 503 Resource release means

Claims (9)

A process that operates in a workload class that does not use up the CPU until the CPU target time when there is no process in the executable process queue that is the queue of executable processes that does not use up the CPU until the CPU target time. A forcible execution target specifying means for specifying a release waiting process waiting for the release of a resource and a resource securing process securing the resource;
Forcibly executing means for forcibly executing the resource securing process;
When the forcible execution means requests the release of a resource along with the execution of the process, the resource is released, and the process is suspended and a release waiting process waiting for the release of the resource is added to the executable process queue. A process control apparatus. The process control device according to claim 1, wherein the forcible execution target specifying unit causes the forcible execution unit to execute the resource securing process on condition that there is a release waiting process waiting for the release of the resource. The process control apparatus according to claim 1, wherein the resource includes a resource shared between workload classes. A priority process storage means for storing resource identification information of a secured resource and process identification information of a process that preferentially uses the resource when releasing the resource;
The forced execution target specifying means stores the resource identification information of the resource secured by the resource securing process in the priority process storage means, and the process identification information of the process waiting to be released is the process identification of the process that preferentially uses the resource. Information is stored in the priority process storage means,
The resource release means can execute the process indicated by the process identification information stored in the priority process storage means when the resource identification information of the released resource matches the resource identification information stored in the priority process storage means. The process control device according to any one of claims 1 to 3, wherein the process control device is queued in a process queue. The process according to any one of claims 1 to 4, wherein the forcible execution means executes the process when there is a process that does not use up the CPU until the CPU target time in the executable process queue. Control device. A process that operates in a workload class that does not use up the CPU until the CPU target time when there is no process in the executable process queue that is the queue of executable processes that does not use up the CPU until the CPU target time. Identify the release-waiting process that waits for the release of the resource and the resource reservation process that secures the resource,
Forcing the resource securing process to run;
When a resource is requested to be released in accordance with the execution of a process, the process releases the resource, interrupts the process, and adds a waiting process for waiting for the release of the resource to an executable process queue. Control method. The process control method according to claim 6, wherein the resource securing process is executed on condition that there is a release waiting process waiting for the release of the resource. On the computer,
A process that operates in a workload class that does not use up the CPU until the CPU target time when there is no process in the executable process queue that is the queue of executable processes that does not use up the CPU until the CPU target time. A forced execution target identification process that identifies a release waiting process that waits for the release of a resource and a resource allocation process that secures the resource,
A forced execution process for forcibly executing the resource securing process; and
When a resource release request is generated during execution of a process in forced execution processing, the resource is released, and the process is suspended and a release waiting process waiting for the resource release is added to the executable process queue. Process control program to execute processing. In the computer,
The process control program according to claim 8, wherein the resource securing process is forcibly executed by the forcible execution process on condition that there is a release waiting process waiting for the release of the resource.

Images