JPH04315247A - Method for operating computer system - Google Patents

Abstract

PURPOSE:To present the computer system operating method to avoid an abnormal condition by detecting this abnormal condition of extremely delaying response time to the processing request of an end user beforehand although a computer is operated normally in the manner of the system. CONSTITUTION:A means is provided to predict the working condition of computer system resources and a service condition to the processing request in the future, and a means is provided to display the predicted results to the operator of the compute system or the operation manager so that the operator or the operation manager can change the operation of the system based on the predicted conditions. Therefore, since the operator of the computer system or the operation manager can operate the system while forecasting the working condition of the computer system resource and the service condition to the processing request in the future, a system fault or the abnormal condition in the aspect of service to the end user is detected in advance, and any countermeasure can be taken before the abnormal condition is generated.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for monitoring and operating the operating status of a computer system.

0002

[Prior Art] As a prior art related to the present invention, "System Operation System Forced to Respond to Online Failures," Nikkei Computer, 1989.12.2.18, pp.8
9-99''.

[0003] Conventionally, as described in the above-mentioned literature, three major operational systems have been established to deal with system failures in computer systems such as online systems. ■Establishing a backup system in terms of system configuration, such as duplication of hosts and files. ■Development of failure detection and recovery mechanisms for operational monitoring systems, network management systems, automatic failure recovery tools, etc. ■Backup system from an operational perspective, such as establishing operational procedures in the event of a failure, reducing human operational errors using automated operational tools, and establishing a collaborative system between the development team and the operational team. The above describes how to quickly detect system failures based on the assumption that computer systems are prone to failures.
The aim is to provide quick recovery measures after a failure is detected.

0004

[Problems to be Solved by the Invention] The conventional methods for dealing with system failures are how quickly the fact that a failure has occurred is detected, and how quickly the failure can be recovered when the failure actually occurs, that is, how quickly the failure occurs, This is a follow-up measure. Therefore, it is important to take measures before a failure occurs, and to ensure that end users (those who use the services provided by the computer system)
It does not prevent the effects of system failure.

[0005]Furthermore, conventional countermeasures are for system failures, and do not take into account abnormalities in terms of service to end users.Although the system is operating without any downtime, end users' There has been a problem in that it is not possible to detect abnormalities such as extremely delayed response times to processing requests and to take measures against them.

An object of the present invention is to provide an operating method that prevents a computer system from experiencing a system failure and also prevents the occurrence of an abnormal situation in terms of service for processing requests from end users. .

[0007]

[Means for Solving the Problems] In order to achieve the above object, the present invention provides means for predicting the operating status of a computer system at a future time and the service status (response time) to a processing request, Alternatively, a means for displaying the information to the operations manager is provided, and the operator or operations manager changes the operation based on the predicted situation.

[0008]

[Operation] That is, by using the situation prediction means and the display means, the operator or operation manager of the computer system can obtain information about the operating status of the computer system and the service status for processing requests in the future; Operational changes can be made in anticipation of future situations. Therefore, it is possible to prevent system failures such as system failures, and abnormal situations in terms of service to end users, such as extreme delays in response time to processing requests.

[0009]

Embodiment An embodiment of the present invention will be described below with reference to FIGS. 1 to 21.

FIG. 1 is a flowchart showing a method of operating a computer system according to the present invention. Before explaining this operating method, an overview of the hardware configuration and processing of the computer system handled in this embodiment will be explained.

FIG. 2 shows the hardware configuration of the computer system used in this embodiment. This computer system consists of computer 1 (201), processing request source terminal (20
2) Group, communication control device (203), storage device (204)
group, system console terminal (205), computer 2 (2
06), an intercomputer communication line (207), and an operation monitoring terminal (208). Computer 1 (201) accepts input processing requests, executes processing according to the requests, and notifies processing results to the processing request source. The processing request source terminal (202) is a terminal that inputs a processing request to the computer 1 (201) and outputs a processing result from the computer 1 (201). Computer 1 (201) and processing request source terminal (2
Data communication with 02) is carried out by the communication control device (203).
controlled by The storage device (204) is connected to the computer 1 (20
1) stores data necessary when executing processing according to a processing request. The system console terminal (205) is
This is a terminal for controlling the operation of computer 1 (201). The computer system consisting of the device group (201 to 205) described above corresponds to a computer system such as an online system, and is the system to be operated in this embodiment. Hereinafter, this system will be described as the operation target system. The computer 2 (206) executes a process of predicting the processing request amount and operating status at a future time from the current processing request amount and operating status of the operation target system. Further, the prediction result is output to the operation monitoring terminal (208). Calculator 1 measures the processing demand and operating status of the current operational target system.
(201), and the information is exchanged via the intercomputer communication line (207).

FIG. 1 is a flowchart showing a method of operating the system to be operated described using FIG. 2 according to the present invention. This operation method mainly consists of two steps (steps 106 and 105). One is the service status for end users (those who use the services provided by the operation target system) of the operation target system at a future time, and in this example, system resources such as each device (201, 204) that constitutes the operation target system. This step is a step (step 106) of predicting the utilization rate of and the response time to a processing request. This step (step 106) will be explained in detail later. The other step is the step of changing the operation method of the operation target system based on the results predicted in the previous step (step 106).
Step 105). Changes in the operating method of the target system are performed by the operation manager or computer operator of the target system via the system console terminal (205). Operational changes made by the operation manager or computer operator of the target system include, for example, whether there is an extreme delay in response time to a processing request or a system failure based on the predicted operation status output to the operation monitoring terminal (208). When it is determined that there is a possibility of an abnormality occurring, the processing request source terminal (202
) by lengthening the polling interval to suppress the amount of processing requested to the operation target system, degenerate the functionality of computer 1 (201), and cancel the suppression and degeneracy when the above-mentioned abnormal condition is avoided. The service status prediction process (step 106) will be described in detail below. First, the amount of processing requests input to the operation target system within a certain time interval, and each device (201, 204) that constitutes the operation target system.
(step 101). Here, the utilization rate of resource i is the rate at which resource i is being used per unit time. Based on the measured quantity, the input speed of processing requests to the operation target system at the measurement time and the amount of change from the previous measurement time are determined (step 102). The input speed of processing requests, the utilization rate of each resource, and the response time to processing requests at a future time are predicted from the quantities obtained in the previous step (step 103), and the prediction results are output (
Step 104). As mentioned in the explanation of FIG. 2, the computer 1 (201) in FIG. 2 executes step 101.
The computer 2 (206) in FIG. 2 executes steps 102 to 104.

If the operation method shown in FIG. 2 is followed, an abnormality in service for the end users of the system to be operated may occur, for example, the system is operating without any system downtime, but the response time to the end users' processing requests is extremely long. It is possible to detect abnormalities in advance, such as delays, and take countermeasures against them.

[0014] Hereinafter, using detailed drawings, computer 1 (20
1) The program configuration and table configuration of computer 2 (206) and the operation flow of each program will be explained.

FIG. 3 shows the computer 1 (201) in FIG.
and shows the program configuration and table configuration of computer 2 (206). Calculator 1 (201) is
System control program (3101), request reception program (3102), processing queue (3103), data processing program (3104), processing result output queue (
3105), result output program (3106), requested input amount count program (3107), requested amount table (3108), monitor program (3109), operating status table 1 (3110), status sending program (31
11). In addition, the computer 2 (206) has a status reception program (3201), a request change amount table (32
02), operation status table 2 (3203), operation status prediction program (3204), utilization rate calculation program (3
205), response time calculation program (3206), waiting rate table (3207), result output buffer (3208)
) and a result output program (3209).

First, the operation of each program in the computer 1 (201) will be briefly explained. System control program (
3101) controls the start and end of programs of the computer 1 (201), manages tasks, manages resources, communicates, and manages data. System control program (3101
) corresponds to a so-called operating system, database/data communication system. The request reception program (3102) is a process requesting terminal (
202) and registers it in the processing queue (3103). Furthermore, each time a processing request is accepted, the request input amount counting program (3107) is notified of the processing request type and the fact that it has been accepted.

The data processing program group (3104) includes:
Takes out the processing request registered in the processing queue (3103), executes data processing according to the request contents,
The processing result is registered in the processing result output queue (3105). The data processing program group (3104) references and updates data stored in the processing data file (3301) in the storage device (204) via the system control program (3101) as necessary. The result notification program (3106) sends a processing result output queue (310
5) The content of the processing result registered in is retrieved and notified to the processing request source terminal (202). The above programs (
3101 to 3106) correspond to a group of programs constituting an online banking system, etc., and process request source terminals (
202), it is possible to execute appropriate processing and respond to various processing requests generated from the processing section 202).

Other program groups (3107, 3109, 3111) and table groups (
3108, 3110) are parts for measuring the service status for the above-mentioned processing request, for example, whether there is a delay in response time. The request input amount count program (3107) is the request reception program (3102).
), the request amount is counted up for each type of processing request. Request amount table (3
108) is a work area for counting up. The detailed operation flow of the requested input amount count program (3107) will be described later using FIG. 9, and the detailed configuration of the requested amount table using FIG. 4.

The monitor program (3109) is located in a subprogram of the system control program (3101), and is a program that obtains the usage status of resources such as the computer 1 (201) and the storage device (204) group at regular time intervals. This corresponds to a so-called software monitor. This program (3109) displays the usage status of the acquired resources, specifically the resource utilization rate, in the operation status table 1 (3110).
). Furthermore, when the output is finished, the status transmission program (3111) is notified to that effect. The detailed configuration of the operating status table 1 (3110) will be described later using FIG. 5.

[0020] When the status sending program (3111) receives the output notification from the monitor program (3109),
This is a program that sends the requested input amount for each processing request type in the requested amount table (3108) and the resource utilization rate in the operating status table 1 (3110) to the computer 2 (206). The detailed operation flow of this program is shown in Figure 10.
This will be described later using FIG. 11.

Next, the operation of each program in the computer 2 (206) will be briefly explained. The program group in the computer 2 (206) receives the service status of the operation target system transmitted from the computer 1 (201), predicts the service status at a future time, and outputs the prediction result.

The status receiving program (3201) receives the request input amount and resource utilization rate for each processing request type sent by the status transmitting program (3111) of the computer 1 (201), and stores the request change amount table (3202). ), set in the operating status table 2 (3203). Further, this program (3201) determines the input speed of the processing request and the amount of change thereof from the received request input amount, and sets it in the request change amount table (3202). The detailed operation flow of the status reception program is shown in FIGS.
3203) will be described later using FIGS. 6 and 7, respectively.

[0023] The operational status prediction program (3204)
This is a program that predicts the input speed of processing requests, resource utilization rate, and response time to processing requests at a future time, and outputs the prediction results to the result output buffer (3208). The detailed operation flow of this program (3204) is as follows:
This will be described later using FIGS. 15 to 17.

The utilization rate calculation program (3205) and response time calculation program (3206) are subprograms of the operational status prediction program (3204) that calculate resource utilization rates and response times to processing requests, respectively. be. The utilization rate calculation in this example is conventionally performed when evaluating the performance of a computer system, which is calculated by the product of the input amount of processing requests per unit time (i.e., input speed) and the time that each processing request occupies a resource. Follow the method from.

The utilization rate calculation program (3205) is a program of this method. This program (
3205) has two execution modes. One is calculation mode and the other is correction mode. In the calculation mode, the utilization rate of each resource is determined from the input speed of the processing request passed as an argument, and the result (utilization rate of each resource) is returned as a return value to the operation status prediction program (3204), which is the starting program. In correction mode, the utilization rate calculation formula is corrected according to the utilization rate correction value passed as an argument.
The operation status prediction program (32
Return to 04).

In addition, the response time calculation in this embodiment follows the conventional method of expressing the computer system as a queuing model and calculating the response time by analyzing that model using queuing theory. The response time calculation program (3206) is a program of the procedure. This program (3206) calculates the response time for each type of processing request from the usage rate of each resource passed as an argument, and returns the result as a return value to the operation status prediction program (3204), which is the starting program. When calculating response time using this method, a standardized waiting rate for each resource utilization rate is required, but the waiting rate table (3207) is a part that stores this waiting rate, and can be used as needed. The response time calculation program (3206) refers to the response time calculation program (3206).

The detailed operation flow or structure of the utilization rate calculation program (3205), response time calculation program (3206), and waiting rate table (3207) will be described later using FIGS. 19, 21, and 8, respectively.

The result output program (3209) outputs the contents of the result output buffer (3208) to the operation monitoring terminal (20).
8).

The program group (3201, 3204, 3205, 3206, 3
209) and table group (3202, 3203, 32
07, 3208), it is possible to receive the service status of the operation target system transmitted from the computer 1 (201), predict the service status at a future time, and output the prediction result.

FIG. 4 shows the request amount table (3
108). This table (3108) is
It consists of a count area designation buffer (400) and a group of request-specific count buffers (410) for the types of processing requests handled by the operation target system. The request-specific count buffer (410) is a buffer for counting up the input amount for each processing request type, and is a buffer for counting up the input amount for each processing request type. Has an area. Count area 1 (411) and count area 2 (412)
), whose value is the requested input amount count program (310
This is a count-up area that is incremented by 7). The area indicating which area (411 or 412) is to be incremented is the count area designation buffer (400). The value of this buffer is “
The value is either "1" or "2", and the value is changed by the status sending program (3111).If the value of the count area specification buffer (400) is "1", the count area 1 (411), "2" ”, count area 2 (4
12) Request input amount count program (3107
) is incremented. Buffer for counting by request (
410) count correction area (413) is count area 1 (411) or count area 2 (412)
This is an area where the value of is temporarily stored. This area (
The specific operation of step 413) will be explained along with the operation flow of the requested input amount counting program (3107).

FIG. 5 shows the operating status table 1 in FIG.
(3110) is a configuration diagram. Book table (3110)
consists of a measurement time buffer (500), a group of resource-specific utilization rate buffers (501) for the resources of the operation target system, and a write flag buffer (502). The utilization rate buffer for each resource (501) is a buffer for storing the utilization rate for each resource, and the monitor program (3109)
) sets its value. Monitor program (3109
) is the write flag buffer (502). The value of this buffer (502) is operated by the status sending program (3111), and if the value is “0”, it is writable, and if it is “1”
It is not possible in this case. The write flag causes the status sending program (3111) to write the operating status table 1 (3110).
) is prevented from being rewritten by the monitor program (3109) while it is being sent to the computer 2 (206). The measurement time buffer (500) is a buffer that stores the time when a value is set in the resource usage rate buffer (501), and the measurement time buffer (500) is a buffer that stores the time when a value is set in the resource usage rate buffer (501).
) is set.

FIG. 6 is a configuration diagram of the required change amount table (3202) in FIG. 3. Book table (3202)
is the request-specific input amount buffer (600) for each processing request type.
It consists of a group. This buffer (600) is a buffer that stores the input status for each type of processing request, and has three areas: an input amount area (601), an input speed area (602), and a change amount area (603). The three area groups (601, 602, 603) are area groups that store the input amount, input speed, and input speed change amount of processing requests, respectively, and the status receiving program (3201)
) is set.

FIG. 7 shows the operating status table 2 in FIG.
(3203) is a configuration diagram. Book table (3203)
consists of a measurement time buffer (700), a previous measurement time buffer (701), and a group of resource-specific utilization rate buffers (703) for the resources that constitute the operation target system. The measurement time buffer (700) and resource utilization buffer (703) are the same as the buffer group (500, 501) of the operating status table 1 (3110), and are set by the status receiving program (3201). The previous measurement time buffer (701) is a buffer that stores the previous measurement time (the value of the measurement time buffer (700) before being rewritten), and is set by the status reception program (3201). Measurement time buffer (700) and previous measurement time buffer (
701), the time interval at which the amount of processing requests is counted up can be determined, and the input speed of processing requests and the amount of change thereof can be determined.

FIG. 8 shows the waiting rate table (3
207). This table (3207) is 2
It has a table structure with a dimensional layout, and each row in the first column is a group of utilization areas (800) that stores utilization rates, and each row from the second column onwards is a group of standardized waiting areas for resources at each utilization rate. This is a group of waiting rate areas (801) that store the waiting rate. In the example of this figure, the second column stores the wait rate of the CPU of computer 1 (201) at each utilization rate, and the third and fourth columns store the wait rate of the storage device (204) group at each utilization rate. It shows. In addition, in the example of this figure, the utilization rate is stored in each row of the first column in increments of 0.05, but the value of this increment value depends on the accuracy of the monitor program (3109) that measures the utilization rate and the waiting rate. The response time is determined according to the required accuracy. Each area (801, 802) of this table (3207) is set in advance on the memory of computer 2 (206). The waiting rate of a resource is determined by referring to the portion corresponding to the queuing model of the resource from a publicly available waiting rate correspondence table of various queuing models at each usage rate.

FIG. 9 is an operational flowchart of the requested input amount count program (3107) in FIG. Below, using this drawing, the required input amount program (3107
) operation flow is explained.

The required input amount program (3107) starts when the computer 1 (201) is powered on, and stops when the power is turned off. After startup, first request amount table (3202)
initialization processing is performed (step 901). The initialization process is performed by ■ all areas of the request-specific count buffer (410) (
411, 412, 413) to "0". ■ Set the value of count area specification buffer (400) to “1”
Set to . After the initialization process is completed, wait until there is a startup request from the request reception program (3102) (
Step 902). As mentioned earlier, the startup request from the request reception program (3102) is
) when a processing request is issued from the processing request source terminal (202), this is a notification to this program (3107) that the request receiving program (3102) has accepted the processing request. When there is an activation request from the request reception program (3102), steps 903, 904, and 905 are executed. In step 903, the value of the count area specification buffer (400) of the request amount table (3108) is referred to, and which count area (411)
or 412) is to be incremented. In step 904, the processing request type is determined. In step 905, the specified count area (
411 or 412) is incremented by "1". After executing steps 903 to 905, the process returns to step 902 and waits.

FIGS. 10 and 11 are operational flowcharts of the status transmission program (3111) in FIG. 3. The operation of the status sending program (3111) will be explained using this drawing. This program (3111) is for computer 1
(201) starts when the power is turned on, and stops when the power is turned off. After startup, set the value of the write flag buffer (502) of the operating status table 1 (3110) to “0” (writable)
(step 1001). Thereafter, the process waits until an activation request is received from the monitor program (3009) (step 1002). Monitor program (30
As described above, the activation request from 09) is a notification of completion of setting the resource utilization rate measured by the monitor program (3109) to the operation status table 1 (3110). When a startup request is received from the monitor program (3009), the value of the time-of-day clock (TOD clock) in the CPU of the computer 2 (206) is referred to, and that value is used as the measurement time in the operating status table 1 (3110). Buffer (50
0) (step 1003). Hereinafter, the value of this buffer will be described as the measurement time of the operating status. After setting the measurement time, the value of the write flag buffer (502) of the operating status table 1 (3110) is set to "1" (write prohibited) (step 1004). Next, the request amount table (3
108) in the count area designation buffer (400) is changed as follows (step 1005). If the value of the count area specification buffer (400) is “1”, “
If it is “2”, change it to “1”.

[0038] The above steps (steps 1003 to 10
05), this program (3111) completes preparations for transmitting data to the status receiving program (3201). After completing the data transmission preparation, the status reception program (320
1) is receivable (step 1006).
). If reception is possible from the status reception program (3201) (step 1007), step 1
008 to step 1010 and step 1011 to step 1012 are executed. If the message indicating that reception is possible is not received in step 1007, the inquiry in step 1006 is retried. However, if the number of retries exceeds k, step 1013 and steps 1011 to 1012 are executed. Here, the number of retries k is arbitrary.

[0039] In step 1008, operation status table 1
(3110) measurement time buffer (500) value (operation status measurement time) and resource utilization rate buffer (501)
) group value (utilization rate of each resource) is sent to the status receiving program (3
201). In step 1009, for each request count buffer in the request amount table (3108), the value in the count area (411 or 412) of the count area specified value before being changed in step 1005 and the count correction area (413) are determined. The value is added and the value (request input amount for each processing request type accepted by the request receiving program (3102) between the previous measurement time and the current measurement time) is sent to the status receiving program (3201). In step 1010, the request amount table (3108
) is set to "0" in the total count correction area (413). In step 1013, the request amount table (31
Step 10 is added to the value of the count correction area (413) for each request count buffer (410) in step 08).
In step 05, the value of the count area (411 or 412) of the count area designation value before being changed is added, and the value is set in the count correction area (413). Step 10
Steps 08 to 1010 are a group of steps that are executed when the message that reception is possible is received in step 1007, and step 1013 is a step that is executed when the message that reception is possible cannot be received during k retries. be. After step 1010 or step 1013 is executed, step 1011 and step 1012 are executed. In step 1011, the request amount table (3108)
The values of all count areas (411 or 412) of the count area specified value before being changed in step 1005 of “
0". In step 1012, the operating status table 1
The value of the write flag buffer (502) of (3110) is set to "0" (writable). After executing step 1012, return to step 1002 and start the monitor program (
3109) until the next activation request is received.

FIGS. 11 to 13 are flowcharts of the operation of the status reception program (3201) in FIG. 3. The operation of the status reception program (3201) will be explained using this drawing. This program (3201) is
(206) starts when the power is turned on, and stops when the power is turned off. After startup, each area group (601,
602, 603) to "0" and all buffer groups (700, 701, 7) in the operating status table 2 (3203).
03) is initialized to “0” (step 1101
). After initial settings, wait until there is an inquiry from the status sending program (3111) (step 1102)
.

When receiving an inquiry from the status sending program (3111), the value of the measurement time buffer (700) of the operating status table 2 (3203) is set in the previous measurement time buffer (701) (step 1103), and the status is sent. Sends a message that reception is possible to the program (3111) (1
104). Then step 1105 and step 1
106 is executed to receive the data sent by the status sending program (3111) and set it in an appropriate table group. In step 1105, the measurement time (value of the measurement time buffer (500) of the operation status table 1 (3110)) and the utilization rate of each resource (the value of the measurement time buffer (500) of the operation status table 1 (3110))
values of the resource-specific utilization rate buffer (501) group) are set in the measurement time buffer (700) and the resource-specific utilization rate buffer (703) group of the operation status table 2 (3203), respectively. In step 1106, the request input amount for each processing request type (count area (4) of each request count buffer (410) of the request amount table (3108)
11 or 412) and the count correction area (413
) is added to the requested change amount table (3202)
The input amount area of the input amount buffer (600) for each request (
601). Status sending program (3111
) receives the data transmitted and sets it in each table, and then calculates the measurement time interval (step 1107).
). The measurement time interval is determined by equation (1) in FIG.

In the next step (step 1108), the following steps (steps 1109 to 1111) are executed for all processing request types, and the update processing of the request change table (3202) is completed. Determine whether or not. If it is determined that the update processing has been completed for all processing request types, step 1112 is executed. In step 1109, the required input speed is determined according to equation (2) in FIG.

In step 1110, the amount of change in the required input speed is determined according to equation (3) in FIG. Step 11
11, the requested input speed obtained in step 1109 is entered in the input speed area (60) of the requested change amount table (3202).
2), the amount of change in the requested input speed obtained in step 1110 is stored in the amount of change area (
603). In step 1112, the operation status prediction program (3204) is activated. This step (
Step 1112) After execution, return to step 1102 and wait.

FIGS. 15 to 17 are operational flowcharts of the operation status prediction program (3204) in FIG. 3. This program (3204) is a program that predicts the input speed of processing requests, the utilization rate of each resource, and the response time for each processing request at a future time. (3204) in advance. Specifically, the predicted time is expressed as the number of loops (hereinafter written as KK).
and a predicted time interval (hereinafter referred to as ΔT). The prediction time interval (ΔT) is the time interval between predicted times, and the number of loops (KK) is the number of times predicted at ΔT intervals from the measurement time. The values of the number of loops (KK) and the prediction time interval (ΔT) are arbitrary.

The operation of the operation status prediction program (3204) will be explained below using FIGS. 15 to 17. This program (3204) is the status reception program (3201)
It is activated by a activation request from , executes a group of steps (steps 1201 to 1214) described below,
finish. When there is a request to start the status reception program (3201), the request input speed for each processing request type at the measurement time (input speed area (602) of each request input amount buffer (600) of the request change amount table (3202)) The value of the)
Run the utilization rate calculation program (320
5) and calculate the utilization rate of each resource (step 12).
01).

In the next step, according to equation (3) in FIG.
Calculate the correction value of the utilization rate of each resource (step 1202)
. The utilization rate calculation program (3205
) and corrects the utilization rate calculation formula for each resource (step 1203). Next, the response time calculation program (
3206) and calculates the response time for each processing request at the measurement time (step 1204).

In the next step, the utilization rate of each resource at the measurement time (the value of the resource-specific utilization rate buffer (703) group of the operation status table 2 (3203)) and step 1204
The response time for each processing request obtained in step 1 is output to the result output buffer (3208).

Steps 1206 to 1213 are as follows:
This is a group of steps for predicting the request input speed, the utilization rate of each resource, and the response time to each processing request at a future time. First, in step 1206, an internal variable I (a variable valid only within this program (3204)) is initialized to "0". Internal variable I is a variable for controlling the internal loop of this program (3204). Step 120
7, the value of the internal variable I is incremented by "1", and it is determined whether the value of the internal variable I is within KK (the number of loops) (step 1208). If the value of internal variable I is within KK, steps 1209 to 1213 are executed, otherwise step 1214 is executed. In step 1209, the predicted time (TI
). In step 1210, the request input speed for each processing request at the predicted time (TI) is predicted according to equation (2) in FIG.

Next, predict the utilization rate of each resource and the response time to each processing request at the predicted time (TI) (
Step 1211, Step 1212). Step 12
11, the predicted time (TI) obtained in step 1210
With the request input speed for each processing request type as an argument,
The utilization rate calculation program (3205) is activated in the calculation mode, and the utilization rate of each resource at the predicted time (TI) is determined. In step 1212, the response time calculation program (3206) is started using the utilization rate of each resource at the predicted time (TI) obtained in step 1211 as an argument, and the predicted time (TI) is activated.
Find the response time for each processing request in TI).

[0050] In step 1213, steps 1210~
The requested input speed of each processing request, the utilization rate of each resource, and the response time for each processing at the predicted time (TI) obtained in step 1212 are output to the result output buffer (3208). After executing step 1213, step 1
Returning to 207, the execution of steps 1207 to 1213 is repeated. After repeating this loop KK (loop number) times, in step 1214 the result output program (
3209) and starts the operational status prediction program (320
4) ends the execution of the series of steps.

FIG. 19 is an operation flowchart of the utilization rate calculation program (3205) in FIG. 3, which is started in response to a start request from the operation status prediction program (3204). The execution modes of this program (3205) include a calculation mode and a correction mode, and immediately after startup, processing for determining which execution mode the startup request is for is performed (step 1301).

If in calculation mode, step 130
2 to step 1304 are executed. In step 1302, step 1303 is executed for all resources, and it is determined whether the utilization rate has been calculated. After calculating the utilization rate for all resources, return the calculated utilization rate of each resource as a return value to the operation status prediction program (3204), which is the startup program (step 1304).
, ends the execution of this program (3205). In step 1303, the utilization rate is calculated according to the utilization rate calculation formula for each resource shown in FIG. Here, i is one processing request type, and m is the total number of the types. The request input speed of each processing request is determined by the operation status prediction program (3
204) as an argument.

[0053] The time that each processing request occupies a certain resource is calculated separately, and the value is used in this program (
3205) directly. For example, if the resource is the CPU of computer 1 (201), the time occupied by the CPU for each processing request is equal to the number of dynamic steps executed when processing one processing request. This is the value divided by the calculation speed of the CPU. The correction value is a value that cannot be calculated using the first term of the above equation, and can be changed by executing this program (3205) in correction mode.

On the other hand, if the mode is correction mode, step 1
305 is executed. In step 1305, the correction value of the utilization rate calculation formula for each resource passed as an argument from the operation status prediction program (3204) is added to the current correction value. After correcting the usage rate calculation formula, this program (3205)
Terminates execution.

FIG. 21 is an operation flowchart of the response time calculation program (3206) in FIG. This program (3206) is the operational status prediction program (3206).
204). After startup, the waiting rate for the usage rate of each resource passed as an argument from the operation status prediction program (3204) is determined by referring to the waiting rate table (3207) (step 1401).

For example, the waiting rate at a CPU utilization rate of 0.20 is determined by the CPU waiting rate area group in the row in which "0.20" is stored in the utilization rate area (800) of the waiting rate table (3207). Queue waiting rate area (801)
Referring to , it becomes 0.250. Here, if the utilization rate does not match the value in the utilization rate area (800) group, the values in the waiting rate area (801) of the two rows that store the values closest to the utilization rate are is obtained by linear interpolation. After determining the waiting rate in step 1401, it is determined whether step 1403 has been executed for all processing request types (step 1402).

In step 1403, the response time to the processing request is determined according to the formula shown in FIG. here,
i is one processing request type, J represents a set of resources used when servicing processing request i, and j
represents one resource belonging to set J. The time that each processing request occupies a certain resource is the same as that used in the utilization rate calculation program (3205). The waiting rate for the utilization rate of each resource is the value determined in step 1401. If it is determined in step 1402 that the response time has been calculated for all processing request types, the calculated response time for each processing request is returned as a return value to the operation status prediction program (3204), which is the startup program (step 1404 ), this program (3206)
Terminates execution.

According to the embodiments described above, it is possible to count the amount of processing requests input to the operating system within a certain time interval and measure the utilization rate, which is the operating status of each resource of the operating system (see FIG. 1). Step 101) can be performed, and the requested input speed and the amount of change thereof can be determined from these measured quantities (step 102 in Figure 1). Furthermore, the requested input speed of processing requests at future times, the utilization rate of each resource, and the amount of change thereof can be calculated for each processing request. The response time can be predicted (step 103 in FIG. 1). Furthermore, by displaying the prediction results to the operator or operation manager of the operation target system (step 104 in Figure 1), the operator or operation manager of the operation target system can understand the operating status of the operation target system and processing requests. It is possible to understand how the service situation for the customer is changing, and to make operational changes in anticipation of future situations (step 105 in Figure 1).
) can be done. Therefore, it is possible to prevent the operation target system from falling into an abnormal situation from a system perspective, such as a system failure due to excessive processing request input.

[0059] Furthermore, since the service status (response time to processing requests) provided by the operation target system to the processing requester can be directly evaluated, there is a margin in the resource usage status (utilization rate) of the operation target system. However, it has the effect of preventing an abnormal situation in terms of service to a processing requester, such as an extremely delayed response time to a processing request.

Furthermore, since a computer for processing processing requests and a computer for predicting the operational status at a future time are provided separately, the load required for the prediction processing does not affect the processing of processing requests.

In the above embodiment, the operator or operation manager of the system to be operated constantly monitors the operation monitoring terminal (
208) will be monitored. This work places a large mental and physical burden on the operator. Modifications aimed at reducing the burden of monitoring work are shown below. In this modification, reference values for the utilization rate of each resource and the response time for each processing request are set in advance, and the result output program 3209 in FIG. When outputting the response time (response time to each processing request), comparison processing with a reference value is performed, and if the response time exceeds the reference value, a warning message is output to the operation monitoring terminal 208. The reference value is set to a value smaller than the value at which the operator determines that an abnormal situation exists. The specific value varies depending on the type of processing request, the social importance of the request, and the hardware that implements the system. Hereinafter, this modification will be explained using FIGS. 23 to 25.

FIG. 23 is a diagram showing the software and table configuration of this modification. The difference in the configuration shown in FIG. 3 is that the result output program 3209 uses the warning criteria table 1
501.

This table 1501 is a table that stores the above-mentioned reference values, and its structure is shown in FIG. 24. The warning criteria table 1501 includes resource usage rate warning criteria 1 buffer (1601) and request response time criteria 1 buffer (16
02), resource usage rate warning standard 2 buffer (1603)
and Request Response Time Warning Criteria 2 Buffer (1604)
Consists of. This table group (1601 to 1604) is a buffer that stores the reference value of the utilization rate of each resource and the response time to each processing request for the result output program 3209 to output a warning message. The resource-specific utilization rate warning standard 1 buffer (1601) and the request-specific response time standard 1 buffer (1602) store the reference value for the status data at the measurement time.
03) and per-request response time warning criteria 2 buffers (16
04) is a buffer that stores a reference value for situation data at the predicted time. Note that the values of each buffer in this table (1501) are set in advance on the memory.

FIG. 25 is an operation flowchart of the result output program (3209) in this modification. As explained in the above embodiment, the operational status prediction program (32
This program (3209) started from 04) first stores the situation data at the measurement time into the result output buffer (3208).
) (step 1701). The value of the retrieved status data and warning criteria 1 (resource usage rate warning criteria 1 buffer (1601) and request specific response time criteria 1 buffer (1601)
602) and determines whether the situation data exceeds the warning reference value (step 1702).
. If it exceeds the limit, a warning message is output to the operation monitoring terminal (208) (step 1703), and step 17
Execute 04. If it is determined in step 1702 that the standard is not exceeded, step 1704 is executed. In step 1704, the status data is transferred to the operation monitoring terminal (
208).

Next, the following steps (steps 1706 to 17) are performed on the situation prediction data at all prediction times.
09) has been executed (step 1705).
). In step 1706, situation prediction data at the predicted time is retrieved from the result output buffer (3208). The value of the retrieved situation prediction data and warning criteria 2 (resource usage rate warning criteria 2 buffer (1603) and request specific response time criteria 2)
It is determined whether the situation prediction data exceeds the warning reference value (step 1707). If it exceeds, the operation monitoring terminal (
208) and outputs a warning message (step 1708).
), step 1709 is executed. If step 17
If it is determined in step 07 that the standard is not exceeded, step 1
Execute 709. In step 1709, the status data is output to the operation monitoring terminal (208), and in step 1705
The process returns to the previous step and continues processing the situation prediction data for the next prediction time. If it is determined in step 1705 that the processing (steps 1706 to 1709) has been executed on the situation prediction data for all predicted times, this program (3
209) ends.

According to the above modification, a warning message is output when a preset reference value is reached, so that the operator or operation manager of the system to be operated always monitors the operation monitoring terminal (208). There is no need to monitor the data, and the burden of monitoring work is reduced.

Furthermore, the operation monitoring terminal (208) is made a terminal with a window function, and the status data is displayed on a window smaller than the screen of the operation monitoring terminal (208) as long as the reference value is not exceeded, and when the reference value is exceeded. When , the window is expanded to its maximum size and the status data is displayed on the expanded window. This results in
In a normal situation (a situation where the reference value is not exceeded), the situation data is displayed inconspicuously on the operation monitoring terminal (208), so that it does not attract unnecessary attention from the operator. Furthermore, in an abnormal situation, in addition to the warning message, the situation data is conspicuously displayed on the operation monitoring terminal (208), so that the operator's attention can be deeply drawn.

Furthermore, the operation monitoring terminal (208) and the system console terminal (205) can be combined into one terminal. FIG. 26 shows its hardware configuration. The operation management terminal (1801) is a terminal that integrates an operation monitoring terminal (208) and a system console terminal (205). Data exchange between this terminal (1801) and computer 1 (201) and computer 2 (206) is performed by the terminal control device (1801).
02). The terminal control device (1802) receives messages and various data output from each computer (201, 206), and sends them to the operation management terminal (1801). In addition, commands, etc. for controlling the operation of computer 1 (201) input from the operation management terminal (1801) are sent to
101) is also performed. As a result, it is possible to monitor the operating status of the system to be operated and to control the operation of the system using a single terminal, thereby improving the work efficiency of the operator.

FIG. 27 is a diagram showing a system configuration for implementing the present invention when there are a plurality of systems to be operated. The operation target system group (1901) is an operation target system defined using FIG. 2, respectively. It is possible to execute processing according to different processing requests for each operation target system (1901) or to execute the same processing. A computer 2 (20
6) is connected via a network (1902). In order to enable the computer 2 (206) to perform the above prediction for each operation target system (1901), an identification number that can uniquely identify the operation target system (1901) is set,
The transmission status program (3111) of each operation target system (1901) adds the given identification number and transmits the information electronically. Furthermore, in the computer 2 (206),
Each table (3202, 3203, 32
By managing 08) and accessing each table (3202, 3203, 3208) corresponding to the transmitted identification number, it is possible to monitor the operation of multiple operational target systems (1901), predict the service status, and operate each system. can be controlled.

In the above embodiment, the utilization rate of each resource and the response time for each processing request are predicted by the utilization rate calculation program (3205) and the response time calculation program (3205).
As mentioned in the explanation of the operation flow in 3206), a mathematical model is used, but it is also possible to describe the operation target system with a simulation model such as GPSS and obtain the desired predicted value by running this simulation model. Can be changed.

[0071]

As described above, according to the present invention, the operator or operation manager of a computer system can perform operations in anticipation of the operating status of the computer system at a future time and the service status for processing requests. It is possible to prevent abnormalities in terms of service to end users, such as system failures and extreme delays in response time to processing requests.

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing a method of operating a computer system according to the present invention.

FIG. 2 is a hardware configuration diagram that enables implementation of the operating method of the present invention.

FIG. 3 is a diagram showing the program configuration and table configuration in each computer in FIG. 2;

FIG. 4 is a detailed configuration diagram of a request amount table in FIG. 3;

FIG. 5 is a detailed configuration diagram of the operating status table 1 in FIG. 3;

FIG. 6 is a detailed configuration diagram of a request change amount table in FIG. 3;

FIG. 7 is a detailed configuration diagram of the operating status table 2 in FIG. 3;

FIG. 8 is a detailed configuration diagram of a waiting rate table in FIG. 3;

FIG. 9 is an operation flowchart of the requested input amount counting program in FIG. 3;

FIG. 10 is a part of an operation flowchart of the status transmission program in FIG. 3;

FIG. 11 is a part of an operation flowchart of the status transmission program in FIG. 3;

FIG. 12 is a part of an operation flowchart of the status reception program in FIG. 3;

FIG. 13 is a part of an operation flowchart of the status receiving program in FIG. 3;

14 is an equation for calculating the measurement time interval, the requested input speed, and the amount of change in the requested input speed calculated by the status receiving program shown in FIGS. 12 and 13. FIG.

FIG. 15 is a part of an operation flowchart of the operation status prediction program in FIG. 3;

FIG. 16 is a part of an operation flowchart of the operation status prediction program in FIG. 3;

FIG. 17 is a part of an operation flowchart of the operation status prediction program in FIG. 3;

FIG. 18 is a formula for calculating a predicted time calculated by the operation status prediction program shown in FIGS. 15 to 17, a requested input speed at the predicted time, and a utilization rate correction value for each resource.

FIG. 19 is an operation flowchart of the utilization rate calculation program in FIG. 3;

20 is a formula for calculating the utilization rate of each resource calculated by the utilization rate calculation program shown in FIG. 19. FIG.

FIG. 21 is an operation flowchart of the response time calculation program in FIG. 3;

FIG. 22 is a formula for calculating the response time of a processing request calculated by the response time calculation program shown in FIG. 21;

FIG. 23 is a diagram showing the program configuration and table configuration in each computer in FIG. 2, which is a second embodiment.

FIG. 24 is a detailed configuration diagram of the warning criteria table in FIG. 23;

FIG. 25 is an operation flowchart of the result output program in FIG. 23;

FIG. 26 is a second hardware configuration diagram that enables implementation of the operating method of the present invention.

FIG. 27 shows a hardware configuration that allows multiple computer systems to be managed using the operating method of the present invention.

[Explanation of symbols]

101... Step of measuring the amount of processing requests to the computer system and the operating status of system resources, 102... Step of calculating the input speed of processing requests and the amount of change, 103... Step of inputting processing requests at future time and system resources based on the speed 104... Step of displaying and outputting the prediction results; 105... Step of changing the operating method of the computer system based on the prediction results.

Claims (21)

[Claims] 1. A computer system that performs processing corresponding to an input processing request and outputs the processing result to the processing request source, wherein a service status of the computer system after a current time is predicted, and the prediction is performed. A method of operating a computer system, comprising changing the operating method of the computer system based on the results. 2. The method according to claim 1, wherein the service status of the computer system after the current time is predicted from the history of the service status of the computer system before the current time. Method. 3. The method of operating a computer system according to claim 1, wherein the utilization rate of a group of elemental resources constituting the computer system is taken as the service status of the computer system. 4. The method of operating a computer system according to claim 1, wherein a response time to a processing request of the computer system is taken as a service status of the computer system. 5. The method according to claim 3, comprising an element that predicts an input speed of a processing request amount input to the computer system at a time after the current time, and configures the computer system based on the prediction result. A method of operating a computer system characterized by predicting the utilization rate of a group of resources. 6. The method according to claim 4, wherein the input speed of the processing request amount input to the computer system at a time after the current time is predicted, and based on the prediction result, the processing request amount of the computer system is responded to. A method of operating a computer system characterized by predicting response time. 7. The method according to claim 5 or 6,
A computer system characterized in that the input speed of processing requests input to the computer system at a time after the current time is predicted from the history of the input speed of processing requests input to the computer system before the current time. Operation method. 8. The method according to claim 7, by adding the number of processing requests when the computer system receives the processing request, and dividing the added number by the time interval during which the addition processing continues. . A method of operating a computer system, characterized in that the input speed of processing requests input to the computer system is determined. 9. The method according to claim 5, wherein the input speed of processing requests input to the computer system is calculated by multiplying the input speed of processing requests input to the computer system by the time required to use a group of elemental resources constituting the computer system per processing request. A method of operating a computer system, characterized in that a utilization rate of a group of elemental resources constituting the computer system is determined by a calculated formula. 10. The method according to claim 6, wherein the product of the time for using the elemental resources constituting the computer system per one processing request and the value obtained by adding 1 to the waiting rate for the utilization rate of the elemental resources. A method of operating a computer system, characterized in that the response time to a processing request of the computer system is determined by a calculation formula that sums the number of resource groups of all the computer systems used by the processing request. 11. The method of operating a computer system according to claim 1, wherein a computer physically different from the computers constituting the computer system is provided, and the prediction process is performed by the computer. 12. The method according to claim 11, wherein the computer system and the computer that performs the prediction process are connected via a communication line, and information is exchanged between the computer system and the computer that performs the prediction process. How to operate the featured computer system. 13. The method according to claim 11, wherein a terminal that is physically different from a group of terminals constituting the computer system is connected to the computer that performs the prediction processing, and the predicted service status is outputted. How to operate a computer system. 14. In the method of outputting the service status of the computer system after the predicted current time of the method according to claim 13, a warning message is sent to the terminal when the service status exceeds a preset reference value. A method of operating a computer system characterized by output. 15. The method of operating a computer system according to claim 14, wherein two types of reference values are set: a reference value for the service status at the current time and a reference value for the service status at a future time. 16. The method according to claim 14, wherein the terminal for outputting the predicted service status is configured with a terminal whose display area size can be arbitrarily adjusted, and when the predicted service status is not exceeded, the display area is adjusted. A computer that adjusts the display area to a smaller value and outputs the predicted service status in the display area, and if it exceeds the reference value, adjusts and enlarges the display area and outputs the predicted service status in the display area. How to operate the system. 17. The method of operating a computer system according to claim 11, wherein the predicted service status is output to one terminal of a group of terminals constituting the computer system. 18. The method of operating a computer system according to claim 17, wherein the one terminal is a terminal for an operator who operates the computer system. 19. The method according to claim 1, wherein the service status of the computer system after the current time is predicted based on a model that simulates the operation of the computer system. . [Claim 20] In a computer system that performs processing corresponding to an input processing request and outputs the processing result to the processing request source, a computer system that performs processing corresponding to the input processing request and outputs the processing result to the processing request source, based on a predicted result of the service status of the computer system after the current time. A computer system characterized in that an operation method of changing the operation method of the computer system is adopted. 21. In a computer system that performs processing corresponding to an input processing request and outputs the processing result to the processing request source, a computer that is physically different from the computer system performs processing corresponding to the request and outputs the processing result after the current time. A computer system characterized in that a service status of the computer system is predicted.