JPH08314767A - Measuring method for data base operating performance and computer device - Google Patents

Abstract

PURPOSE: To provide a measuring method to acquire quantitative evaluating information for data base operating performance to easily obtain a measure to improve the data base operating performance. CONSTITUTION: A CPU 20 collects all event information (address and time) in an event collection area 101 by an event collecting part 124 that is an OS function while an application program by a multi-task execution part 122 is executed. After the execution of the program was completed, an event edit part 125 edits event edit information on which a resource name, etc., is added from all event information as referring to allocator map information 113, and stores it in an event edit area 102. After that, an event editing part 126 for data base reads out the event information from the event edit area 102 sequentially, and samples the information of a measuring target task, and edits event trace classified by every event factor. Moreover, statistical information is found by tabulating plural parameters for evaluation such as the number of times of occupancy or that of occupancy-waiting, and it is outputted in the form of list or graph.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer system having a database, and more particularly to a method for measuring database operation performance when a plurality of tasks operate the database in parallel.

[0002]

2. Description of the Related Art In an electronic computer system in which a plurality of tasks operate simultaneously, a method of optimizing the processing is a method of improving the processing performance of a program, a method of reducing competition between tasks and reducing a waiting time, and an auxiliary storage device. There is a method to reduce the number of I / O issuance to. In actual adoption, it is necessary to investigate and evaluate the degree of performance improvement by each method described above in advance.

As a system for collecting, editing and displaying information for performance evaluation of application programs, HI
TEST-DEMO Ver. 9 (online function / performance debugging support system; Hitachi, Ltd .; see Software Manual 65-3-3224).

HITEST-DEMO uses event information collected by the operating system (OS) as the OS.
Edit and display in macro trace, event trace, etc. by combining with allocator map information managed by.
With this function, the timing of occupying and releasing shared resources,
It is possible to measure the occurrence of wait due to competition for shared resources with other tasks, the timing of input / output to / from the auxiliary storage device, and the number of transfer bytes. From the analysis of the measurement result, it is possible to evaluate the validity of the processing of the application program and the validity of the input / output occurrence frequency with respect to the auxiliary storage device.

[0005]

However, the above-mentioned HIT
The measurement result by EST-DEMO relates to the processing performance of the entire measurement target application program, and in the measurement target application program, the portion related to database operation and other portions cannot be evaluated separately. Since the database operation performance varies greatly depending on the database structure, scale, access procedures, etc., it makes it difficult to measure and evaluate information on the database operation performance.

In recent years, with the development of databases in control information processing systems, improvement of database operation performance has become a more urgent issue, and simple realization of advance evaluation thereof is required.

A first object of the present invention is to provide a method for measuring the database operation performance when a plurality of tasks operate the database in parallel.

A second object of the present invention is to provide a computer system capable of measuring database operation performance.

[0009]

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel. O
Event information collected by S) is stored in time series, and after execution of all tasks, the event information is sequentially read to extract database resources and database-related event information related to input operations for each task. ,
This is achieved by aggregating multiple evaluation parameters related to database operation performance.

The evaluation parameters are totaled according to the cause of occurrence of the event information. When the cause of occurrence is a database file conflict, the number of times of occupancy and the number of waits for each file name and / or the total time of occupancy and the total waiting time are calculated. Included, if the cause is shared buffer contention, it includes the number of times of occupancy and the number of waits for each buffer name and / or the total occupancy time and the total wait time. If the cause is a database I / O operation, the buffer name And the number of times of operations for each file name and the number of times of input / output of the auxiliary storage device (hereinafter abbreviated as the number of times of I / O).

Further, the evaluation parameter includes index information that quantitatively indicates the database operation performance. In the case of contention of the database file, the index information is a file wait time based on the ratio of the number of times the file is occupied and the number of wait times. The occurrence rate and / or the file waiting time ratio according to the ratio of the total occupied time of the file and the total waiting time, and in the case of contention for the shared buffer, the index information depends on the ratio of the occupied number of the shared buffer and the waiting number. In the case of the database input / output operation, the index information is the number of operations and the number of I / Os, in the case of the database input / output operation, the file waiting rate and / or the file waiting rate based on the ratio of the total occupied time and the total waiting time of the shared buffer. It is characterized in that the hit rate is calculated according to each ratio.

Further, it is characterized in that the event trace of the database-related event information and / or the statistical information of the evaluation parameter is edited and output.

[0013]

According to the structure of the present invention, the cause of deterioration of the database operation performance can be easily determined from the event trace and the statistical information, and effective performance improvement measures can be implemented.

That is, in the performance evaluation information on the database file, for example, the reason why the occupation wait occurrence rate of a certain file is high is that the measurement target task occupies the file for a long time and conflicts frequently with other tasks. The cause of the performance deterioration is the task to be measured. In this case, the conflicting time is reduced by shifting the start timing of the task, or the database operation processing is reviewed to shorten the file occupancy time once and improve the parallelism between tasks.

In the performance evaluation information on the shared buffer of the database, for example, the occupied waiting time ratio of the shared buffer is high because the shared buffer is frequently used by a certain file, which is one of the causes of the performance deterioration. Shared buffer configuration. In this case, the contention on the shared buffer can be reduced by dividing the shared buffer used by the file in which the contention frequently occurs into a plurality of divided buffers.

Further, in the performance evaluation information on other database operations, for example, the hit rate of the file operation is low because the data necessary for the number of file operations is not in the shared buffer and the IO to the auxiliary storage device is not present.
This is due to the frequent occurrence of the above, and one cause of the performance deterioration is the lack of the capacity of the shared buffer. In this case, the number of inputs / outputs to / from the auxiliary storage device can be reduced by expanding the capacity of one side of the shared buffer or increasing the number of sides of the shared buffer according to the contents of the database operation processing.

[0017]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a computer device having a database operation performance measuring means according to an embodiment of the present invention.

The main storage device 100 has an event collection area 101, an event editing area 102, and a database event editing area 103, in addition to storage areas for operation programs such as application programs and database management programs (not shown). There is. further,
It is used as a buffer area when accessing the data of the database on the auxiliary storage device 110, and a plurality of buffers B are used.
UF1, 2. ． ． Shared buffer 104 composed of
have.

The auxiliary storage device 110 stores a plurality of database files FAA, FBB ,. ． ． With a database 111 that stores actual data and a data dictionary 1
12 and an allocator map information area 113. The data dictionary 112 stores definition information such as the structure and scale of the database, the relationship between tasks and database files, the relationship between files and shared buffers, and is used when constructing the database and operating the database. In this embodiment, as described later, the OS
Also stores the definition information for associating the relationship between the event information issued by and the database operation. The allocator map information area 113 stores allocation information of the main memory 100, allocation information of the auxiliary storage device 110, task attribute information, and the like, which are managed by the OS.

The central processing unit 120 is an input / output device 1 having a keyboard and a display as functions of the CPU.
An input / output control unit 121 that controls input and output to and from 30, a multitask execution unit 122 that processes a plurality of tasks (here, application programs) in parallel, a database management unit 123, an event information collection unit 124, an event editing unit 125, The database event editing unit 126 and the editing instruction unit 127 are included.

The database operation performance measuring means of the present invention is, in a broad sense, like a box surrounded by a broken line, the event information collecting section 124 which is a well-known function of the OS, and the collected event information and allocator map information for each task. Edit event information such as macro trace or event trace as a whole (equivalent to HITEST-DEMO described above). The same event editing unit 125 as the conventional one, and database related information is extracted from the edited content of the entire event information. It also includes a database event editing unit 126 for obtaining database operation performance evaluation information and an editing instruction unit 127 for instructing activation of event information editing and editing conditions. Of course, the main part of the operation performance measuring means is the database event editing unit 126. The event editing unit 125, the database event editing unit 126, and the editing instruction unit 127 may be integrated.

In the computer system of this embodiment configured as above, when the database operation request is generated by the execution of the application program by the multitask executing unit 122, the database managing unit 123 determines, based on the information in the data dictionary 113, A request for operating a database file is issued to the OS. Upon receiving the database file operation request, the OS operates the database 111 via the shared buffer 104. During execution of the application program, the OS collects various event information (actually, the address of the event information) such as a system call, input / output to / from the auxiliary storage device, and database operation by the event information collecting unit 124 which is a function of the OS. It is collected in the event information collection area 101 together with the occurrence time.

After executing the application program, the event editing section 125 is activated. Event editorial department 12
5 reads all the event information stored in the event information collection area 101 in time series, refers to the allocation information of the allocator map information area 113 corresponding to the address of the event information, the attribute information of the task, etc. Create a time-series collection of event editing information by adding event types, resource addresses, etc. further,
Edit macros and races, event traces, etc. according to resources. These event editing information collections are
The data is stored in the event editing area 102 and is output to the display of the input / output device 130 or a printer in response to a request.

FIG. 2 shows an example of macro traces, events and races edited by the event editing section 123. In the figure, a macro trace 201 is a collection of event editing information relating to occupancy and release of shared resources, particularly files, and includes a task number (TNO) and a task name (TN).
AME), event occurrence time (TIME), event type (ACT), target shared resource address (TOP
AD ... LASTED, event execution result (RESU)
LT) and the like are listed. Event type (AC
T) includes occupancy (RSERV) and release (FRER). If the event can be occupied (RSERVE-I
N) if the occupancy time cannot be occupied (RESER)
VED) collects the name and task number of the occupied task.

The event trace 202 is a collection of event editing information relating to occupancy, release, I / O operation, etc. for the buffer, and includes task number, task name, event occurrence time, event type (CODE), and parameter at event issue ( PARA) is collected. Here, as the event type (CODE), the database operation start (d
open), occupancy (rserv) of shared resource, release (free), data reference of database (dbre)
ad), data read from auxiliary storage (rfil
w) etc.

The database event editing unit 126
After the processing of the event editing unit 125, the editing instruction unit 127 for setting the editing request and the event information editing conditions such as the task to be operated and the event factor is activated, and from the event editing information collection stored in the event editing area 102,
The event information related to the database is extracted, the event trace is edited for each factor, and some parameters are aggregated to obtain the evaluation information of the database operation performance, which is edited and output in an easy-to-read format.

In the event editing information collection by the event editing section 125, information related to the database 111 and other information are mixed. The database management information stored in the data dictionary 112 is used to extract the event information related to the database from this.

FIG. 3 is a flow chart for explaining the operation of the database event editing section which is the main part of the database operation performance measuring function in the computer system of this embodiment.

The database event editing unit 126
It is activated after the event editing unit 125 is executed. The task for which the operational performance is measured is set in advance, and when there are a plurality of tasks, the tasks are sequentially executed.

First, the edit event information generated and collected by the execution of the application program is sequentially read from the time series of the event edit information collection stored in the event edit area 102 (s201). Next, the task name or number of the event information is used to check whether the information is related to the target task (s202). If not, the next event information is read.

If it is the event information of the task to be measured,
Each event factor is determined (s203), and the event information is distributed to file contention related, buffer contention related, or other. For example, if the event information is in the macro trace of the event edit information collection, it is classified into file conflict, and if it is in the event trace, it is classified into buffer conflict or other from that CODE. Note that you can set any one of the event factors and only
The following processing may be performed.

Next, when the event factor is file conflict related event information, it is determined whether the event information is for a database file based on the file definition information of the data dictionary 112 registered when the database is constructed. Yes (s204). If it is an event for a database file, file event totaling processing is performed (s205). That is, it is registered as database event information, and parameters such as file occupancy time, file occupancy count, file occupancy wait time, and file occupancy wait count are totaled and stored in the database event collection buffer 103.

When the event factor is the event information related to buffer contention, it is determined whether the event information is for the shared buffer 104 of the database or not based on the buffer definition information of the data dictionary 112 (s20).
6). If the event is related to the database shared buffer, the shared buffer event is aggregated (s20).
7). That is, while registering as database event information, parameters such as the shared buffer occupancy time, the shared buffer occupancy count, the shared buffer occupancy wait time, and the shared buffer occupancy wait count are totaled and stored in the database event editing area 103.

When the event factor is other, the data dictionary 1 registered at the time of database construction
The file and buffer definition information 12 is referred to, and it is determined from the address of the event information whether or not the operation is a database data operation (s208). If the event information is a data operation, the operation event totaling process 208
Is performed (s209). That is, while registering as database event information, parameters such as the operation type (reference, addition, update, etc.) of data, the number of issuances thereof, the number of input / output times of the auxiliary storage device, and the like are stored in the database event editing area 103. To do.

The counting process (s
A part of 205, s207, s209) may be performed before the display editing process (s211) described later.

FIG. 4 shows an example of database management information in which the data dictionary 112 is stored. The file definition information is the file name of the database file,
It manages addresses, record length, number of records, etc. The buffer definition information manages the buffer name, address, buffer size, number of buffer files, etc. of the shared buffer of the database.

In step s204, in order to extract the event information related to the database file,
Each address information of the macro trace 201 is compared with each address of the file definition information 401, and only the event information of the defined address is selected. Further, in step s206, in order to extract the event information related to the database shared buffer, the event trace 20
2 address information is compared with the buffer definition information 402,
Select only the event information of the defined address. Further, in step s208, in order to extract the event information regarding the data operation of the database, the address information of the event trace 202 is compared with the addresses of the file definition information 401 and the buffer definition information 402, and only the defined event information is obtained. Select.

After each processing for each event factor, it is judged whether all the target event information of the event editing information stored in the event editing area 102 has been processed (s210), and the processing is completed. If not, the processing from step s201 is repeated.

If the processing of all the event information in the target event editing information collection is completed (the total number of event information is managed in the header part of the event editing area 102),
The process moves to the display editing process (s211). Here, event trace information for each factor and aggregated information of parameters stored in the database event editing area 103 are taken out, and the aggregated information is edited and output in the form of a statistical list or statistical graph in an easy-to-read expression. This editing is also performed by referring to information such as related task names and file names stored in the data dictionary 112.

FIG. 5 shows a display example of the performance evaluation information regarding the database file in the form of a list.
It was edited in 11. The file contention event trace information 501 is a time series of event information related to file contention for each measurement task registered in step s205, such as file names and timings of file occupancy and occupancy release, occupancy waiting file names and occupying task names, etc. Is displayed.

The file contention statistical information 502 is statistical information for each operation target file of the measurement task (TA1 in this example), and the total number of occupied times and occupied time, the name of the occupied task, the number of occupied wait times, the occupied wait occurrence rate. , The total occupied waiting time, the waiting time ratio, and other parameters.

Here, the occupancy wait occurrence rate is a ratio of the number of occupancy wait times to the number of occupancy times for each file expressed as a percentage (%) for each occupying task. The larger this ratio, the worse the performance of the target task. . The waiting time ratio is the ratio of the occupied wait time to the total occupied time and occupied wait time for each file, expressed as a percentage (%) for each occupying task. The higher this ratio, the better the performance of the target task. bad.

FIG. 6 shows a display example of the performance evaluation information regarding the database file in the form of a graph.
It was edited in 11. The file conflict event trace information 601 is a graphical representation of the list information 501 of FIG. 5 along the time axis, and the status of each file occupied by the measurement task and the wait status can be recognized at a glance and is easy to see. The file contention statistical information 602 is a bar graph showing the number of occupied times (solid line) and the number of waits (dotted line) for each file in the list information 502 of FIG. 5, and at first glance the target task with poor performance and its factors. I understand.
Similarly, the total occupied time and the total waiting time for each file may be displayed in a bar graph, or both the number of times and the time may be displayed in a graph.

In the performance evaluation information relating to the database file, the reason why the occupation wait occurrence rate of a certain file is high is that the measurement target task occupies the relevant file for a long time and competition with other tasks frequently occurs. , It can be evaluated that the performance of the measurement target task is poor. Therefore, the conflicting time can be reduced by shifting the start timing of the task, or the operation of the database can be reviewed to shorten the file occupancy time once and improve the parallelism between tasks.

FIG. 7 shows a display example of the performance evaluation information regarding the database shared buffer in the form of a list. The shared buffer contention event trace information 701 is a time series of event information related to buffer contention for each measurement task registered in step s207, and occupies the shared buffer, release timing, buffer name, operation file name, and waits for occupancy. This is a display of each situation such as the occupied file name in.

The shared buffer contention statistical information 702 is statistical information for each operation-target shared buffer of the measurement task (TA1), and is the total number of occupied times and occupied time for each operation file,
It displays parameters such as the name of the task being occupied, the number of waits for occupation, the occurrence rate of occupation waits, the total occupation wait time, and the wait rate.

Here, the occupancy wait occurrence rate is a ratio of the occupancy wait occurrence with respect to the number of times of occupancy of each shared buffer, expressed as a percentage (%) for each operation file and occupied task. It can be evaluated that the task performance is poor. The waiting time ratio is the occupied time for each shared buffer, and the ratio of the occupied waiting time to the total occupied waiting time is expressed as a percentage (%) for each operation file and occupying task. The larger this ratio, the more the target. The processing performance of the task decreases.

FIG. 8 is a display example of the performance evaluation information regarding the database shared buffer in a graph format. The shared buffer contention event trace information 801 is a graphical display of the list information 701 of FIG. 7 along the time axis, and the occupancy of each buffer and the status of occupancy waiting in the measurement task can be grasped visually, making it easy to see. . The shared buffer contention statistical information 802 is the list information 702 of FIG.
Of these, a bar graph shows the number of occupied times or total occupied time (solid line) and the number of waiting times or total wait time (dotted line) for each shared buffer, and the occupation waiting status of the shared buffer can be seen at a glance.

In the performance evaluation information on the database shared buffer, for example, the occupied waiting time ratio of a certain shared buffer is high because a certain file is frequently used by the certain shared buffer, which causes deterioration of the configuration of the shared buffer. It can be evaluated as having a cause. Therefore, the buffer in which the contention frequently occurs is redivided into a plurality of buffers to reduce the occupied waiting time of the shared buffer.

FIG. 9 shows a display example of the performance evaluation information regarding the database input / output operation in the form of a list. The input / output event trace information 901 is a time series of input / output information relating to database operation for each measurement task registered in step s209, and is used for file occupancy, occupancy release, shared buffer occupancy, occupancy release, and data operation type (see, (Update), event contents such as I / O, and the timing thereof are shown. I / O indicates the occurrence of input / output to / from the auxiliary storage device.

The input / output statistical information 902 is the measurement task (T
It is statistical information for each operation target buffer and operation target file in A1), and displays parameters such as the number of file operations, the number of I / O operations, and the hit rate. here,
The hit ratio is a ratio (%) of the ratio of necessary data already read into the shared buffer to the number of file operations, that is, the ratio of no I / O. The lower the hit rate, the worse the buffer usage efficiency and performance.

FIG. 10 is a display example of the performance evaluation information regarding the database input / output operation in a graph format. The input / output event trace information 1001 is a graphical display of the list information 901 in FIG. 9 along the time axis, and shows the timing of file occupancy and occupancy release in the measurement task.
It shows the relationship between the occupation of the shared buffer, the timing of releasing the occupation, the data operation and the input / output generation timing of the auxiliary storage device. In the figure, the time relationship of occupancy / release of the shared buffer BUF1, data reference / update, and input / output of the auxiliary storage device is graphed for a certain period during which the file AAA is occupied.

The input / output statistic information 1002 is included in the list information 902 of FIG. 9 and includes the operation target file in the measurement task (TA1), the number of file operations for each shared buffer used (solid line), and the auxiliary storage device. The number of outputs (dotted line) and the hit rate (dotted line) are displayed.
From this, BUF when the file BBB is occupied
You can see at a glance how bad the hit rate is.

In the performance evaluation information relating to the database input / output operation, for example, the hit rate for the file operation is lowered because the necessary data is often not in the shared buffer and I / O to the auxiliary storage device frequently occurs. Therefore, one cause of the performance deterioration can be evaluated as the lack of capacity of the shared buffer. Therefore, the hit rate can be increased and the processing performance can be improved by increasing the capacity of one side of the shared buffer or expanding the number of sides of the shared buffer according to the contents of the input / output operation of the database.

As described above, according to the present embodiment, it is possible to extract only the event information relating to the database operation performance from the OS event information and acquire the event trace necessary for the database performance evaluation. In addition, a plurality of parameters can be aggregated from the event trace to obtain statistical information as an index for performance evaluation. These database operation event traces and statistical information are displayed in easy-to-read form in the form of lists and graphs to facilitate performance evaluation or investigation of the factors of performance deterioration. In this way, the index for objectively evaluating the factors of database operation performance and performance deterioration is provided to facilitate the improvement measures of the application program and the database for improving the processing performance.

As described above, one embodiment of the database operation performance measuring method of the present invention has been applied to a general computer system having a multitask executing means and a database managing means. However, it goes without saying that the invention can also be applied to a dedicated processing device such as a program debugging device.

[0058]

According to the database operation performance measuring method of the present invention, the event trace related to the database such as the contention of the database file or the shared buffer or the database input / output operation is edited to edit the event trace whose contents can be easily understood. In addition, since multiple parameters required for performance evaluation can be aggregated and provided as statistical information in a list or graph format, it is possible to obtain quantitative evaluation information regarding database operation performance. There is an effect that the deterioration factor can be quantitatively and easily performed. This facilitates measures for improving database operation performance.

According to the computer system equipped with the database operation performance measuring method of the present invention, the evaluation information for evaluating the database operation performance and the performance deterioration factor can be obtained.
There is an effect that the processing performance can be improved by optimizing the application program or the database configuration.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a computer device including a database operation performance measuring unit according to an embodiment of the present invention.

FIG. 2 is a list showing an example of an overall event information collection collected by an OS and edited by conventional event editing means.

FIG. 3 is a flowchart for explaining the operation of the database performance measuring means according to the embodiment.

FIG. 4 is a data table showing an example of database management information defined in a data dictionary.

FIG. 5 is a display example of a list format of event trace information and statistical information related to database file conflict.

FIG. 6 is a display example of a list format of event trace information and statistical information related to database file conflict.

FIG. 7 is a display example of a list format of event trace information and statistical information related to database shared buffer contention.

FIG. 8 is a display example of event trace information and statistical information related to database shared buffer contention in a graph format.

FIG. 9 is a display example of a list format of event trace information and statistical information related to database input / output operations.

FIG. 10 is a display example in a graph format of event trace information and statistical information related to database input / output operations.

[Explanation of symbols]

100 ... Main storage device, 101 ... Event collection area, 1
02 ... event editing area, 103 ... database event editing area, 104 ... shared buffer, 110 ... auxiliary storage device, 111 ... database, 112 ... data dictionary, 113 ... allocator map information area,
Reference numeral 120 ... Central processing unit (CPU), 121 ... Input / output control unit, 122 ... Multitask execution unit, 123 ... Database management unit, 124 ... Event information collecting unit, 125 ... Event editing unit, 126 ... Database event editing unit,
127 ... Edit instruction unit, 130 ... Input / output device, 140 ... System bus.

Front page continuation (72) Inventor Kohei Seki 5-2-1 Omika-cho, Hitachi-shi, Ibaraki Hitachi Process Computer Engineering Co., Ltd. (72) Masao Sueki 5-2-1 Omika-cho, Hitachi-shi, Ibaraki Incorporated company Hitachi Ltd. Omika factory (72) Inventor Shinki Sasaki 5-2-1, Omika-cho, Hitachi-shi, Ibaraki Hitachi Process Computer Engineering Co., Ltd. (72) Inventor Takashi Owaki Gomika-cho, Ibaraki 2-1-1, Ltd. Omika Factory, Hitachi Ltd.

Claims (14)

[Claims] 1. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, wherein event information collected by an operating system (hereinafter abbreviated as OS) during execution of the tasks. Are stored in time series, after the execution of all tasks, the event information is read in sequence, database related event information related to database resources and input operations is extracted for each task, and multiple evaluation parameters related to database operation performance are extracted. A method for measuring database operation performance, which comprises: 2. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, wherein event information collected by an operating system (hereinafter abbreviated as OS) during execution of the tasks. Are stored in chronological order, and after the execution of all tasks, the collected event information is
The allocator map information managed by the OS is referenced to edit the edit event information to which a task name or the like is added, and the edit event information is read out to obtain database-related event information related to database resources or operations for each task. A method for measuring database operation performance, which comprises extracting and aggregating a plurality of evaluation parameters relating to database operation performance. 3. The database operation performance measuring method according to claim 1, wherein the database-related event information is extracted by referring to a data dictionary that stores a database structure. 4. The evaluation parameter according to claim 1, 2 or 3, wherein the plurality of evaluation parameters are aggregated according to the cause of occurrence of the event information, and when the cause of occurrence is a database file conflict,
Includes the number of times of occupancy and the number of waits for each file name, and / or the total time of occupancy and the total waiting time. If the occurrence factor is a database input / output operation, the total wait time is included, and the number of operations for each buffer name and file name and the number of input / output times for auxiliary storage (hereinafter abbreviated as I / O times) are included. A method for measuring database operation performance. 5. The evaluation parameter according to claim 4, wherein the evaluation parameter includes index information that quantitatively indicates database operation performance, and in the case of contention of the database file, the index information is the number of times the file is occupied and the number of waits. The ratio of the file waiting occurrence rate and / or the file waiting time ratio by the ratio of the total occupied time of the file and the total waiting time, and in the case of contention for the shared buffer, the index information is the number of times the shared buffer is occupied. In the case of the database input / output operation, the index information is the number of operations and the buffer wait occurrence rate according to the ratio of the number of wait times and / or the file waiting time ratio according to the ratio of the total occupied time of the shared buffer and the total wait time. The hit ratio is calculated by the ratio of the I / O counts. Constant method. 6. The database operation performance according to claim 1, wherein the event trace of the database related event information and / or the statistical information of the evaluation parameter is edited and output. Measuring method. 7. The event trace and / or the statistical information according to claim 6,
A method for measuring database operation performance, characterized by editing in a list or graph format and outputting. 8. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, comprising: event information issued and collected by an OS; and a data dictionary storing a database structure. Based on the definition information, for the measurement target task, edit the status of file occupancy and release due to database operations and the file release wait status that occurs due to file contention between tasks as an event trace. A method for measuring database operation performance, characterized in that the number of waits and / or the occupied time and the waiting time are obtained and edited and displayed as statistical information. 9. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, comprising: event information issued and collected by an OS; and a data dictionary storing a database structure. Based on the definition information, edit the occupancy and release status of the shared buffer due to database operations and the shared buffer release wait status that occurs due to shared buffer contention between tasks for the measurement target task, and edit it as an event trace. A method for measuring database operation performance, characterized in that the shared buffer occupancy count and wait count for each file and / or the shared buffer occupancy time and wait time are obtained, and edited and displayed as statistical information. 10. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, comprising: event information issued and collected by an OS; and a data dictionary storing a database structure. Based on the definition information, for the measurement target task, the operation status of files and shared buffers by the database input / output operation and the input / output status (hereinafter abbreviated as I / O) with respect to the auxiliary storage device are edited as event trace information. A method of measuring database operation performance, characterized in that the number of reference processes, the number of additional update processes, and the number of IOs to the auxiliary storage device are calculated for each shared buffer, and the statistical information is edited and displayed. 11. A method for obtaining evaluation information of database operation performance when a plurality of tasks operate a database in parallel, wherein event information collected by an operating system (hereinafter abbreviated as OS) during execution of the tasks. Are stored in chronological order, and after the execution of all tasks, the collected event information is
The allocator map information managed by the OS is referenced to edit the edit event information to which the task name and the like are added. Regarding the information of the measurement target task in the edit event information, file contention or buffer contention that is the event occurrence factor Or, divide it into other categories, extract the database related event information by referring to the data dictionary that stores the database structure for each cause, edit the event trace, and compile multiple evaluation parameters related to database operation performance. Editing statistical information including a wait occurrence rate and / or a waiting rate of files and buffers obtained from the evaluation parameter and the plurality of parameters, and a hit rate of a database input / output operation, and the event trace and / or the statistical information. Display in list or graph format A method for measuring database operation performance, characterized by: 12. Central processing unit (hereinafter abbreviated as CPU)
In a computer device having a main memory, an auxiliary storage device for storing a database, and an input / output device having a display, and in which a plurality of tasks can operate the database in parallel, an operating system (hereinafter, Event information collected by OS)
An event editing unit that refers to the allocator map information managed by the OS after execution of all tasks and edits and stores the edited event information with the task name and the like, and the task to be measured by reading the edited event information. Database-related event information related to database resources and operations is extracted for each, and event traces are edited, as well as a database event editing means that edits statistical information by aggregating multiple evaluation parameters related to database operation performance. A computer device provided with database operation performance measuring means. 13. The computer apparatus according to claim 12, further comprising edit instructing means for instructing activation of the database operation performance measuring means. 14. The computer apparatus according to claim 13, wherein the edit instruction unit sets a cause of event information, and the event trace and the statistical information edit the set cause.