JPH0830560A - Load control system for computer system - Google Patents

Abstract

PURPOSE:To efficiently control the load of a computer especially at the time of high load. CONSTITUTION:A CPU check program 23 having the lowest priority is periodically started in each CPU built in a process monitoring/controlling computer 20 and the contents of a CPU health counter in a system common table 26 are counted up (in each CPU). In order to check the margin of processing capacity, a system state check program 24 to be driven by any CPU and having the highest priority order is started, the load state of each CPU is judged based upon the count value of the CPU health counter and an evaluation value about the normality/abnormality of each CPU is written in the table 26. The evaluation values of respective CPUs are summed based upon the states of respective CPUs and weight coefficients corresponding to the states to find out the state of the whole system. A processing number is determined by a system state transition program 25 and informed to respective programs 21, 22 and processing restriction (e.g. inhibition or the like of data totalization and document preparation) regulated by respective programs 21, 22 is executed, so that the load of the CPU can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load control system in a computer system, and more particularly to a load control system for efficiently controlling the load of a CPU under heavy load.

[0002]

2. Description of the Related Art In recent computer systems, a system (for example, a multiprocessor system) in which a plurality of CPUs are used and each CPU performs an individual task has been used. In such a system, the load of each CPU may vary. In this case, the load of the overloaded CPU is replaced by another lightly loaded CPU, and the load of each processor is changed. A method of equalizing the load is known (for example, Japanese Patent Application Laid-Open No. 3-114242).

[0003]

In the above-mentioned conventional method, the load is assigned only in consideration of the equal load, without considering the execution timing of the program executed by the computer system. However, in an actual computer system, for example, a plant monitoring control computer, some of the programs to be executed are to be executed immediately when a state change (pressure, temperature change, fire, etc.) occurs. In addition, there are some which do not need to be processed in a hurry, such as form creation and tabulation processing.

In the conventional method, the load is simply equalized without considering the timing of executing such a program, and therefore, it is not necessary to perform the processing when the CPU is under heavy load (for example, a form). It was not efficient because it created and aggregated). For this reason, the capacity of the computer is determined based on a simple total of the load amount. Therefore, if the load increases, a large-capacity computer must be used.

The present invention has been made in view of the above points, and an object of the present invention is to enable efficient control of the load on a computer.

[0006]

According to the present invention, there is provided a computer system having a single or a plurality of CPUs, the load state of the CPU being determined based on a running state of a CPU health check program. Is recognized, and the execution contents of the program are controlled according to the load state of the CPU.

[0007]

When the load on the computer is high, the load on the CPU can be reduced by controlling the execution of the program so as not to execute the processing that does not require urgency.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a system configuration of an embodiment of the present invention. Here, a process monitoring computer will be described as an example. This system comprises a process input / output device 10 and a process monitoring control computer 20.

[0009] The process monitoring control computer 20 includes n
The CPUs # 1 to #n and the memory 30 are composed of a plurality (n ≧ 1) of CPUs # 1 to #n, a memory 30, and a bus 40.
Are connected via the bus 40.

[0010] The memory 30 stores the programs and tables shown in FIG. The processing program 21 based on the state change and the processing program 22 based on the time are:
The processing program 21, which is an application program, is a program that executes various processes based on input data (for example, status information such as pressure, temperature, water level, and alarm information such as fire) from the input / output device 10. The processing program 22 is, for example, a program that executes various processes at a certain time, and specifically, is a program that executes, for example, form preparation, data aggregation, and various data processes.

Further, in the memory 30, 23 is each C
A CPU health check program executed for each PU; 26, a system common table in which states of the entire system and each CPU are recorded; 27, a system state transition table that defines a processing method according to a change in the system state;
4 is a system status check program for checking the system status from the status of each CPU; 25 is an application program 21 for processing in accordance with a change in the system status;
22 is a system state transition program to be transmitted to 22.

The CPU health check program 23 comprises:
It is started at the fixed period t1 with the traveling priority being the lowest. The system status check program 24 can be run by any CPU and has the highest priority in running, and is started at a fixed period t2 in order to allow the information processing capability of the process monitoring control computer 20 to have a margin. This activation cycle is longer than that of the CPU health check program 23. That is, t1 ≦ t2.

FIG. 2 shows an example of the configuration of the system common table 26. The table shown in this example includes a system status record 26a representing the system status and each C
It is divided into a CPU status record 26b indicating the status of the PU.

There are one system status record 26a and one CPU status record 26b.
The system state record 26a contains a new value of the system state 2
61, the old value 262 is stored, and the CPU status record 26b is stored.
Stores a CPU status 263, a CPU health counter 264, a CPU health reference value 265, and a CPU status weight coefficient 266 for each CPU.

The system common table 26 is located in an area that can be referred to by each program, but the system state check program 24 and the system state transition program 25 are included.
It cannot be written in other than.

FIG. 3 shows an example of the configuration of the system state transition table 27. The table shown in the example shows the processing to be executed when a record that can be searched by the new value and old value of the system state changes from the old value of the system state to the new value. The process to be executed is a number (process number)
Is remembered. In the processing programs 21 and 22, processing to be executed according to the processing number and processing to be postponed are set in advance.

Next, the operation of the above embodiment will be described with reference to FIGS. First, the CPU health check program 23 is set for each CPU so that the traveling priority is set to the lowest and the fixed period t
1 (step 101). The CPU health check program 23 runs on the designated CPU, and when activated, increments the CPU health counter 264 of the corresponding CPU in the CPU status record 26b (step 102). This is repeated for all CPUs (step 103). Since the CPU health check program has the lowest priority, C
When the load on the PU increases, the execution of the health check program is postponed, and the count value of the health counter 264 decreases.

On the other hand, the CPU health check program 2
3, the system status check program 24 is started at the cycle t2, the count value of the CPU health counter 264 is compared with the CPU health reference value 265, and the CP
The U state 263 is determined (FIG. 5, step 201). In short, the number of times the health check program 23 has run during the cycle t2 is compared with the reference value.

If the count value of the CPU health counter 264 ≧ the CPU health reference value 265, it is determined that the CPU is normal, and the CPU is normal in the CPU status 263 provided for each CPU on the system common table 26. An evaluation value (for example, “10”) indicating the fact is written. On the other hand, the count value of the CPU health counter 264 <the CPU
If the health reference value is 265, it is determined that the CPU is overloaded, and an evaluation value (for example, “1”) indicating that the CPU is abnormal is written to the CPU status 263 in the system common table 26.

After writing either CPU normal / abnormal in the CPU state of the system common table 26, the CPU health counter is reset to 0 (step 202).
This process is executed for all CPUs (step 2
03) Then, the evaluation value written in the CPU state 263 is multiplied by the CPU state weighting coefficient 266, and it is summed up for all CPUs to determine the system state (new value) (step 204). Then, it is judged whether or not there is a change in the system state (new value) (step 205), and if there is a change, the latest value is written in the system state (new value) 261 in the system state record 26a, and the conventional ( The data is updated so that the new value) moves to the system state (old value) 262 (step 206).

Next, the system status check program 24
Notifies the system state transition program 25 of the change in the load state of the system (step 207). The system state transition program 25 sets the traveling priority to the highest like the system state check program 24. When activated, the processing to be performed is determined based on the system state transition table 27 with reference to the new and old values of the system state in the system common table 26 (step 208).

Next, the determined processing numbers are notified to the processing programs 21 and 22 (step 209), and the programs 21 and 22 perform the processing determined by the respective programs. For example, a large number of process state changes occur, information is transmitted from the process input / output device 10 to the process monitoring control computer 20, and the information is notified of a state change exceeding the processing capability of the process monitoring control computer 20, and the CPU load is reduced. When it becomes higher, the execution of the state change processing program 21 is prioritized, and the processing program 22 (data tabulation, form creation, etc.) based on time is postponed. When the load on the CPU is further increased, the processing of the processing program 21 due to the state change is also restricted. For example, the output system is only displayed on the screen, and the audio output and the hard copy output are stopped. Further, when the processing program 21 periodically captures state data due to a state change, this processing is stopped when the load is high, and processing such as data capture is performed collectively when returning to the normal CPU state. You can also do so. Then, a consistent state signal can be taken into the process monitoring control computer 20.

In short, when the load is high, the processing contents of the processing programs 21 and 22 are restricted, and the
The load on U decreases after a while, and when the load decreases, the state of each CPU changes due to the above-described processing, and the operations of the processing programs 21 and 22 change accordingly.

Note that programs that are not running when the CPU status changes start up in the system common table 26 at startup.
Read the new value and old value of the system status in, and perform the processing accordingly.

In this way, the CPU load can be easily recognized regardless of the number of CPUs constituting the CPU. That is, in the above system, the number of runs of the CPU health check program within a predetermined time is counted, and the number of runs is compared with a reference value to recognize the load state of the CPU. The load can be checked with a simple configuration. Then, the processing to be executed is determined according to the load state recognized in this way, so that the processing to be performed at a high load can be finely set, thereby reducing the load and causing the entire system to function due to an overload at a high load. Stopping can be prevented, and the safety of the system can be ensured.

[0026]

As described above, according to the present invention,
The load on the computer can be efficiently controlled, and even when the load is high, the processing content can be limited according to the content of the program to prevent the system from stopping.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating a system configuration according to an embodiment of the present invention.

FIG. 2 is a detailed view of a system common table of FIG. 1;

FIG. 3 is a detailed diagram of a system state transition table of FIG.

FIG. 4 is a flowchart illustrating the operation of the embodiment.

FIG. 5 is a flowchart illustrating the operation of the embodiment.

[Explanation of symbols]

 REFERENCE SIGNS LIST 10 process input / output device 20 process monitoring control computer 21 processing program by status change 22 processing program by time 23 CPU health check program 24 system status check program 25 system status transition program 26 system common table 27 system status transition table 30 memory

Claims (5)

[Claims] 1. In a computer system containing a single or a plurality of CPUs, load state recognition means for recognizing a load state of the CPU based on a running state of a CPU health check program, and a program according to the load state. And a program control means for controlling execution of the load control method in a computer system. 2. The load state recognizing means counts the number of times the CPU health check program has run within a reference time, and C compares the number of times of travel with a reference value.
A recognition means for recognizing a load state of the PU is provided.
Load control method described in. 3. The load control method according to claim 1, wherein the recognizing unit recognizes a load state by weighting each CPU. 4. The control device according to claim 1, wherein the control device controls a program according to a change in a load state.
The load control method described in the item. 5. A computer system containing a single or a plurality of CPUs, counting the number of running times of a CPU health check program within a predetermined time,
A CPU load recognition method comprising recognizing a CPU load state by comparing the number of times of travel with a reference value.