JP3335488B2 - Performance prediction apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a performance prediction apparatus and method for predicting a change in performance caused by a change in the amount of business given to a computer system or a change in the system configuration.

[0002]

2. Description of the Related Art The present applicant has previously proposed a performance prediction apparatus described in Japanese Patent Application Laid-Open No. 5-324358 as a system for predicting the performance of a computer system. This performance prediction device collects operation information of the computer system at predetermined time intervals, obtains an average response at the time of business load fluctuation based on the collected operation information, and calculates the calculated average response and the performance. Based on information on the processing capacity per unit time based on a given hardware configuration, the system predicts the performance when the workload changes and when the hardware configuration is changed. Performance can be predicted.

[0003]

However, in the above-described apparatus, it is necessary to express the change in the workload and the change in the performance due to each hardware configuration by a coefficient. It is difficult to obtain the above-mentioned coefficient in consideration of the resource waiting state of the job corresponding to the user environment, and it is difficult to predict the performance.

An object of the present invention is to provide a performance predicting apparatus capable of predicting the performance of a computer system in a short time even when the workload of the computer system changes or the hardware configuration is changed.
And a method .

[0005]

In order to achieve the above object, according to the present invention, there are provided means for previously collecting and holding operation information, hardware configuration information, and job execution information of a computer system; Means for holding processing capacity information, and each job based on operation information, hardware configuration information, and job execution information of the computer system.
Use time of each resource used by the job
Means for dividing the resources according to the number and arranging the respective resources to reproduce the operation status of the job; and generating and deleting the job when the change of the work amount is instructed for the reproduced operation status of the job. When the change of the hardware configuration is instructed, a means for changing the use time of the hardware resources with reference to the processing capacity information of the hardware, and a change before and after the change of the workload and the hardware configuration. The present invention is characterized in that it further comprises means for simulating the operation status of the reproduced job later and calculating a response time for each job, and means for displaying a change in the response time before and after the change.

[0006]

In the embodiment , computer system operation information, hardware configuration information, and job execution information collected in advance are input to the job operation status reproduction function unit, and the actual job operation status is reproduced. Change of work volume by user specification,
When a change in the hardware configuration is instructed, the job division information change function unit generates and deletes a job when the change in the workload is instructed, and when the hardware configuration change is instructed, the job division information change function unit executes the job change. The use time of each of the divided resources is changed with reference to the processing capacity information. The simulation function unit arranges the division information of each job on the time axis for each start time, performs a simulation while observing the usage status of each resource, and for each job, uses the usage time of each resource and waits for each resource. The time is obtained, and the result is input to the graph creation function unit, and a prediction result graph is created and displayed.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be specifically described below with reference to the drawings. FIG. 1 shows the configuration of an embodiment of the present invention.
In the figure, 1 is a performance prediction request unit, 2 is a storage unit for computer system operation information and hardware configuration information collected in advance, 3 is a storage unit for job execution information collected in advance, and 4 is a resource within the job response time. A job operation status reproduction function unit for reproducing the use status, a job division information change function unit for changing job division information by changing the workload and a hardware configuration, a storage unit for hardware processing capability information, Is a simulation function unit, 8 is a graph creation function unit, 9 is a prediction result display unit, and 10 is a change designation unit that designates a change in the workload and a change in the hardware configuration. In addition,
FIG. 1 is actually a so-called computer hardware resource.
It is needless to say that the configuration is made by using. For example,
A storage device is used for each of the storage units 2, 3, and 6, and each functional unit is used.
4 to 8 are realized by a processing device (CPU).
You.

FIG. 2 shows an example of operation information and hardware configuration information of the computer system in the storage unit 2. In the figure, a collection start time 101 is a collection start time of the operation information, and a collection end time 102 is a collection end time of the operation information. The CPU model 103 indicates a model of the CPU from which the operation information has been collected. The CPU usage time 104 is between the collection start time 101 and the collection end time 102.
Shows the total time used by the CPU.

The number of disks 105 indicates the number of disks when the operation information was collected. Disc 1 model name 106
Indicates the model name of the first disk device when the operation information is collected (this information is collected for the number of disks). The disk 1 usage time 107 is the collection start time 1
It indicates the total time during which the first disk has been used from 01 to the collection end time 102 (this information is collected for the number of disks).

FIG. 3 shows an example of job execution information in the storage unit 3. The unit of the numbers in the figure is millisecond. The job name 201 indicates the name of the executed job, the job start time 202 indicates the time when the job started, and the job end time 203 indicates the time when the job ended.

The CPU usage time 204 indicates that the job
The total time using U, the disk usage time 205 is
The total number of times the job has used a disk and the number of disk uses 206 indicate the total number of times the job has used a disk.

The paging time 207 is the total time required for the job to perform paging processing, the paging count 208 is the total number of times the job has performed paging processing, and the workstation input / output time 209 is The total number of times the job has performed input / output to / from the workstation, and the total number of times the job has performed input / output to / from the workstation, is the total number of times the job has performed input / output to the workstation.

The other input / output time 211 is the sum of the input / output time performed by the job to peripheral devices other than the disk and the workstation, and the other input / output times 212
Indicates the total number of input / output operations performed by the job with respect to peripheral devices other than disks and workstations.

The file exclusion wait time 213 indicates the total time the job has been waited for by file exclusion, and the file exclusion wait number 214 indicates the total number of times the job has been waited for the file exclusion. The user thinking time 215 indicates the total time the job has been waited for waiting for user input, and the user thinking number 216 indicates the total number of times the job has been waited for waiting for user input.

FIG. 4 shows an example of hardware (CPU) processing capability information stored in the storage unit 6. In this example, the MIPS values for each CPU model are stored. Here, the MIPS value indicates whether the CPU has a processing capability of millions of instructions per second.

FIG. 5 shows an example of hardware processing capability information of the disk device stored in the storage unit 6. The average seek time 402 and the average rotation waiting time 403 for each disk device model name 401 represent the average value of the time required for the disk device to seek and rotate, respectively, in milliseconds. The data transfer time 404 indicates the time required for the disk device to transfer 1 kilobyte of data in milliseconds.

The operation of the present invention from the input of a performance prediction request to the display of a performance prediction result will be described below with reference to FIG. When a performance prediction request is input to the performance prediction request unit 1, computer system operation information and hardware configuration information previously collected and stored in the storage unit 2 (FIG. 2)
Job execution information (FIG. 3) collected in advance and stored in the storage unit 3 is input to the job operation status reproduction function unit 4,
The resource usage status within the job response time is reproduced.

The job operation status reproducing function unit 4 functions as follows. FIG. 6 is a diagram illustrating the reproduction of the job operation status. The job operation status reproduction function unit 4 first obtains the time during which the job has used each resource based on the system operation information and the hardware configuration information. In the example of FIG.
B1 uses the CPU for 30 ms, and uses the disk (DKI /
O) using 40 ms, paging (P) using 7 ms, and workstation (WS I / O) 3891
ms.

Next, an average value of each use time per job is obtained based on the time when the job uses each resource and the number of times. For example, the number of times of paging (P) is seven, and its use time is 7 ms, so that the use time per one time is 1 ms. Note that the number of times of use of the CPU is estimated from the number of times of use of other resources related to the processing of the CPU because collection is difficult. Therefore, the number of times of use of the CPU is the sum of the number of times of use other than the CPU, and in the example of FIG.
The number of times U is used is DK I / O (once), paging (7 times), and WS I / O (2 times), for a total of 10 times. In this example, the usage time of the CPU per time is 3 ms.
Becomes

Then, the use time of each resource used by JOB 1 is divided into the number of times of use. That is, in the example of FIG. 6, the CPU (30 ms) is replaced with ten CPUs (3 ms).
Paging (7 ms) has 7 pagings (1 m
s), the WS I / O (3891 ms) has two WS
(1945 ms) and WS (1946 ms), the user thinking time is divided into two. In the above example, since the processing is performed using integer values, the WS I / O usage time of 3891 ms is not divisible by 2, but is divided as in the above example.

Next, in order to reproduce the actual operation status, the divided resources are arranged, for example, as shown in FIG. That is, first use the CPU for 3 ms, then use paging (P) for 1 ms,. . . WS 1945m
s and 3557 ms as user thinking time. . . Arrange as follows. It should be noted that which resource other than the CPU is to be used, or which of the plurality of disk devices is to be used, is determined from hardware configuration information, computer system operation information, and disk usage time.

By performing the above-described division, the operation status (resource usage status) of the job, which resource is used in which order and to what extent, is reproduced. It is possible to calculate the waiting time and the use time of the resource and predict the performance.

When the reproduction of the execution status of all the jobs is completed by the above-described job operation status reproduction function unit 4, the job division information is input to the job division information change function unit 5. The change function unit 5 receives a change instruction from the change designating unit 10. The change specifying unit 10 instructs a change in the amount of work and a change in the hardware configuration according to the specification of the user.

The job division information change function unit 5 generates and deletes a job when an instruction to change the work volume is issued, and stores a processing capacity information storage unit of the hardware when an instruction to change the hardware configuration is issued. Referring to 6, the usage time of each resource divided by the job operation status reproduction function unit 4 is changed.

FIG. 7 is a diagram for explaining an example of changing the job division information. For example, a case where the change request specified by the user is a change in the CPU configuration will be described. If the current configuration (the configuration before the change) is the model 5 of the hardware processing capability information (CPU) shown in FIG. 4 and the change-designated CPU configuration (the configuration after the change) is the model 3, the model 3 Is 9.6 MIPS, Model 5 is 6.0
Since it is MIPS, model 3 is about 1.6 times faster than model 5.

Therefore, the usage time of the model 3 CPU is
Since it is 0.625 times that of the model 5, the CPU usage time in the division result obtained by the job operation status reproduction function unit 4 is all made 0.625 times. In the example of FIG. 7, the 8 ms usage time of the CPU before the change is changed to 5 ms.
The usage time of the disk, paging, etc. is not changed.

The job division information changed as described above is input to the simulation function unit 7. FIG. 8 is a diagram illustrating a situation where the division information of each job is arranged on the time axis for each start time at the time of simulation. As shown in FIG. 8, the simulation function unit 7 arranges the division information of each job on the time axis for each start time, performs a simulation while observing the usage status of each resource, and performs each resource for each job. And the waiting time of each resource.

In the example of FIG. 8, first, JOB1 operates,
A CPU is used, followed by a DK or the like. JO
B2, while the CPU is being used by JOB1,
The CPU is put in a wait state, and then paging is performed using the CPU. In this way, when a resource requested to be used by a certain job is not being used, the resource is used, and during that time, the resource usage time is added. Also,
When the resource is in use, a wait queue is created to put the job in a waiting state and wait for the resource to become free. The time is added as a resource waiting time. Thus, the use time of each resource and the wait time of each resource are obtained for each job.

The simulated result is input to the graph creation function unit 8, and a prediction result graph is created. 9 and 10 are detailed graphs of the prediction result graph, respectively.
Shows a summary graph. In FIG. 9, for each job, a graph is created by pairing the current (before change) response time (wait time + use time + user thinking time) and the predicted (after change) response time. Based on this graph, the performance (response time) of the job before and after the change is compared and determined. FIG.
The outline graph of 0 is a graph for viewing the overall change of the response time. In this example, the average value of the response time of the job being executed in a certain time zone is represented by a line graph.

The created graph is displayed on the prediction result display unit 9, and the user judges the performance of the job from the display result. Simulation is performed by changing the amount and hardware configuration.

[0031]

As described above, according to the present invention , even a person other than an expert can easily and quickly obtain a computer system.
System performance changes and hardware configuration changes.
It becomes possible to predict.

[Brief description of the drawings]

FIG. 1 shows a configuration of an embodiment of the present invention.

FIG. 2 shows an example of operation information and hardware configuration information of a computer system.

FIG. 3 shows an example of job execution information.

FIG. 4 shows an example of hardware (CPU) processing capability information.

FIG. 5 shows an example of hardware processing capability information of a disk device.

FIG. 6 is a diagram illustrating reproduction of a job operation status.

FIG. 7 is a diagram illustrating an example of changing job division information.

FIG. 8 is a diagram illustrating a situation in which division information of each job is arranged on a time axis for each start time during a simulation.

FIG. 9 shows a detailed graph of a prediction result graph.

FIG. 10 shows a summary graph of a prediction result graph.

[Explanation of symbols]

 1 Performance prediction request unit 2 System operation information and hardware configuration information storage unit 3 Job execution information storage unit 4 Job operation status reproduction function unit 5 Job division information change function unit 6 Hardware processing capacity information storage unit 7 Simulation function unit 8 Graph Creation function section 9 Prediction result display section 10 Change designation section

────────────────────────────────────────────────── ─── Continued on the front page Examiner Takeshi Sakaniwa (56) References JP-A-5-324358 (JP, A) JP-A-6-95931 (JP, A) JP-A 5-334102 (JP, A) JP-A-61-294559 (JP, A) JP-A-4-186445 (JP, A) JP-A-2-202541 (JP, A) JP-A-60-11948 (JP, A) JP-A-4-24841 (JP JP, A) JP-A-6-59939 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 11/34

Claims (2)

(57) [Claims] 1. A performance prediction apparatus for predicting a change in job response time caused by a change in the amount of work given to a computer system or a change in a system configuration, the operation information and the hardware configuration information of the computer system collected in advance. , And means for storing job execution information,
Based on the operation information, hardware configuration information, and job execution information of the computer system ,
Use time of each resource used by the job according to the number of uses
Means for reproducing the operation status of the job by arranging the resources, dividing the resources, and storing the processing capability information of the hardware; and instructing the change of the workload for the reproduced operation status of the job. Means for generating and deleting a job when the change of the hardware configuration is instructed, and changing the usage time of the hardware resources by referring to the processing capacity information of the hardware; and Means for simulating the operation status of the reproduced job before and after the hardware configuration is changed, calculating a response time for each job, and displaying the change in the response time before and after the change. A performance prediction device characterized by the following. 2. Processing of computer hardware resources
Using the apparatus and storage device, the computer business the amount of change given to the system, or a performance prediction method for predicting a change in the response time of job caused by changes in system configuration, operation information of the computer system, hardware The configuration information and job execution information are stored in a storage device, and the processing device is provided for each job based on the operation information of the computer system, hardware configuration information, and job execution information stored in the storage device. , Each used by the job
The resource usage time is divided according to the number of uses, and each resource is distributed.
The job status is reproduced in a row, and when the operation status of the reproduced job is instructed to change the workload, the job is generated and deleted, and the hardware configuration is changed. Changes the usage time of the hardware resources with reference to the processing capacity information of the hardware held in advance, and operates the reproduced job before and after the change in the workload and the hardware configuration. And calculating a response time for each job, and displaying a change in the response time before and after the change.