JP2006244422A - System parameter automatic tuning method using network - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system parameter automatic tuning method using a network for automatically changing parameters, shortening a time for tuning and preventing an error in setting the parameters. <P>SOLUTION: A combination of a plurality of parameter values assumed when tuning object parameters of an object server 2 is separated from one another and registered, and a tuning result for the object server 2 is returned to an operation server 3. Thus, the tuning condition of the object server 2 is seized by the operation server 3. Then, a check result for the parameter values is reflected to the object server 2. Thus, the parameters can be automatically changed, the time for turning is shortened, and the error in setting the parameters is prevented. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method for automatically tuning system parameters using a network.

  When tuning the target parameters of the system, each of the enormous cases divided by a combination of a plurality of parameter values was manually changed through a person each time.

  For this reason, due to a parameter value setting error or the like, the time required for tuning increases accordingly, and at the same time, the labor cost corresponding to the time required for tuning increases accordingly.

  In addition, by collecting and organizing the system environment and database status, the tuning points and parameters are determined by the fuzzy inference rules that are built for the system, so there is some knowledge about the system to be tuned. For those skilled in the art, a method has been proposed that can enable clear information collection and accurate system performance tuning in a short time (see, for example, Patent Document 1).

In addition, the system parameters that determine the system operating conditions are analyzed based on the data collected by the system operation status data collection means, and the system can be changed efficiently by changing the bottleneck. There has been proposed a method capable of achieving the above (for example, see Patent Document 2).
JP 11-249934 A JP 58-149527 A

  Incidentally, Patent Document 1 described above discloses that parameter management is performed by a server connected through a network and that a parameter value is determined by checking a correlation between parameter values. In some cases, setting the setting conditions in advance is not disclosed.

  Further, Patent Document 2 described above does not disclose checking a correlation between parameter values although parameter values are set based on performance information acquired from a tuning target server.

  Therefore, an object of the present invention is to automatically change system parameters using a network, which can automatically change parameters, reduce the time required for tuning, and prevent parameter setting errors. It is to provide a method.

  In order to solve the above-mentioned problem, the invention according to claim 1 divides cases by combinations of a plurality of parameter values assumed in the tuning of target parameters of a target server to be tuned in a system using a network, and performs tuning. The case classification result is registered in an operation server that performs an operation, a parameter value is set in the target server according to each combination according to an instruction from the operation server, and a tuning result in the target server is System parameters are automatically tuned by returning them to the server.

  According to the invention described in claim 1, by performing case classification based on a combination of a plurality of parameter values assumed in the tuning of the target parameter of the target server, and registering the tuning result in the target server to the operation server The tuning status of the target server can be ascertained by the operation server, and the parameter value check result is reflected in the target server, so that the parameter can be automatically changed and the tuning time can be shortened. It is possible to prevent parameter setting errors.

  The invention according to claim 2 is the invention according to claim 1, wherein the parameter information in the server operating in the verification environment is stored in a file, and the file is placed in the operation server as it is, so that the value of the parameter information is obtained. The parameter value is automatically reflected on the target server.

  According to the second aspect of the present invention, it is possible to further prevent a parameter setting error by automatically reflecting the parameter value in the target server with the value of the parameter information.

  The invention according to claim 3 is the invention according to claim 1, wherein the output information collected by the performance characteristic tool in the verification environment is stored in a file, and the file is placed in the operation server as it is. The value of the target parameter is automatically determined from the parameter value, and the parameter value is automatically reflected on the target server by the determined value.

  According to the third aspect of the present invention, it is possible to further prevent a parameter setting error by automatically reflecting the parameter value on the target server with the determined value.

  The invention described in claim 4 is the invention described in claim 1, wherein a standard tuning pattern is prepared in advance for each application such as a web system or a database system, and a necessary place based on a set value in the selected tuning pattern It is characterized by correcting.

   According to the invention described in claim 4, a standard tuning pattern is prepared in advance for each application, and a necessary portion is corrected based on a set value in the selected tuning pattern, thereby further reducing the time required for tuning. can do.

  The invention according to claim 5 is the problem according to any one of claims 1 to 4, wherein a check is made as to whether or not a plurality of parameters registered and set in the operation server have appropriate parameter values. For certain parameter values, the operation server prompts an error message.

  According to the fifth aspect of the present invention, it is possible to prevent reoccurrence of a parameter setting error by prompting an error message at the operation server for a problematic parameter value.

  The invention according to claim 6 is the invention according to any one of claims 1 to 4, in which, in addition to the correlation between parameters, an agreement regarding tuning of the commercial introduction system is determined in advance. If a condition that deviates from the agreed items occurs in the tuning after setting the items to be registered in advance, check whether or not the parameter values are appropriate for the plurality of parameters to be registered and set in the operation server. And creating and utilizing a knowledge base for checking whether or not the correlation between the parameters is satisfied.

  According to the sixth aspect of the present invention, the time for automatically changing the parameters and the tuning by creating and utilizing the knowledge base for checking whether or not the correlation between parameters is satisfied. Can be shortened, and parameter setting errors can be further prevented.

  Here, the solving means of the present invention will be described in detail.

  An enormous case classification based on a combination of a plurality of parameter values, which is assumed in the tuning of the target parameter of the system, is registered in advance on the server that performs the tuning operation. Then, according to instructions from the server that performs the tuning operation, parameter values are automatically set on the server that is the tuning target according to each combination, and the tuning result on the server that is the tuning target is set to the server that performs the tuning operation. To return to.

  In addition, for the enormous case-by-case registration by combining multiple parameter values to the tuning server, if the evaluation environment is almost the same as the commercial environment in advance, the following two automatic settings Also prepare.

  The first method is to store the parameter information of the server operating in the verification environment in a file, and place the file on the server that performs tuning operation as it is. Parameter value is reflected.

  The second method is to store the output information (performance statistical information) collected by the performance characteristics tool (performance collection tool such as sar command in UNIX (registered trademark)) in a verification environment in a file, and leave the file as it is By placing it on the tuning server, the target parameter value is automatically determined from the trend of the output information (statistical information), and the parameter value is automatically reflected on the tuning target server with that value. . In this method, it is necessary to create and utilize a knowledge base (knowledge) for automatically determining the value of a target parameter from the tendency of output information (statistical information).

  Here, it is possible to manually correct the value based on the value of the parameter information to be automatically set by the above method. In addition to setting from the beginning, prepare a standard tuning pattern for each application, such as Web / DB, in advance, and based on the setting value in the selected tuning pattern, find the necessary part. It also makes it possible to fix it. By this method, setting correction work is reduced and setting errors are reduced.

Furthermore, in order to prevent setting mistakes, the multiple parameters registered and set on the server that performs tuning operations are checked to see if they are appropriate parameter values or satisfy the correlation between parameters. For certain parameter values, an error message is prompted on the server that performs the tuning operation. In addition to the correlation between parameters, agreements regarding tuning of commercial installation systems (changes / impossibility of each tuning parameter, restrictions on the setting value of each tuning parameter, etc.) are determined in advance. The items to be registered are set, and if there is a possibility of deviating from the agreed items during tuning, an error message is also prompted on the server that performs the tuning operation.
In this method, a knowledge base (knowledge) is created to check whether the parameter values are correct and satisfy the correlation among the parameters registered and set on the server that performs the tuning operation. It is necessary to use it.

  According to the present invention, a case-by-case registration based on a combination of a plurality of parameter values assumed in target parameter tuning of the target server is performed and registered, and a tuning result in the target server is returned to the operation server, so that The tuning status can be grasped by the operation server, and the parameter value check result is reflected in the target server, so that the parameter can be changed automatically, the time required for tuning can be reduced, and the parameter Setting errors can be prevented.

〔Characteristic〕
The present invention provides a method for automatically tuning parameters of servers constituting a system in order to efficiently perform a tuning operation performed before commercial introduction of the system.
Embodiment
The system parameter automatic tuning method using the network of the present embodiment performs the case classification based on a combination of a plurality of parameter values assumed in the tuning of the target parameter of the target server to be the tuning target of the system using the network, Register the case-by-case results on the operation server that performs the tuning operation, set the parameter value on the target server according to each combination according to the instruction from the operation server, and return the tuning result on the target server to the operation server The system parameters are automatically tuned according to the above.

  In addition to the above configuration, the system parameter automatic tuning method using the network of the present embodiment stores parameter information in a server operating in the verification environment in a file, and places the file on the operation server as it is. The parameter value may be automatically reflected on the target server by the value of the parameter information.

  In addition to the above configuration, the system parameter automatic tuning method using the network of this embodiment stores the output information collected by the performance characteristics tool in the verification environment in a file, and places the file on the operation server as it is. Alternatively, the target parameter value may be automatically determined from the tendency of the output information, and the parameter value may be automatically reflected on the target server by the determined value.

  In addition to the above configuration, the system parameter automatic tuning method using the network of this embodiment prepares a standard tuning pattern for each application such as the Web system or DB system in advance, and sets the setting value in the selected tuning pattern. You may make it modify a necessary part based on it.

  In addition to the above configuration, the system parameter automatic tuning method using the network of the present embodiment has a problem by checking whether or not a plurality of parameters registered and set on the operation server have appropriate parameter values. Regarding the parameter value, an error message may be prompted on the operation server.

  In the system parameter automatic tuning method using the network of the present embodiment, in addition to the above-described configuration, in addition to the correlation between the parameters, the agreement regarding the tuning of the commercial introduction system is determined in advance. If the items to be registered in advance are set and the condition that deviates from the agreed items occurs in the tuning, check whether or not the appropriate parameter values are set for multiple parameters to be registered and set on the operation server. A knowledge base for checking whether or not the correlation between the two is satisfied may be created and used.

〔Constitution〕
FIG. 1 is a conceptual diagram showing an embodiment of a system parameter automatic tuning method using a network according to the present invention.
In FIG. 1, the tuning target server 2 included in the system 4 before commercial introduction is selected from the tuning operation server 3 by using the network 1, the OS on the tuning target server 2, and the tuning determination information on the application 23. The tuning determination information result is collected by setting the tuning parameter and sending the tuning instruction, and the tuning target server 2 sends the result to the tuning operation server 3.

  The tuning target server 2 previously receives the operation from the tuning operation server 3 on the memory of the tuning target server 2 from the recording medium 21, the operation execution at the tuning target server 2, and the tuning result at the tuning target server 2. The program 22 for receiving and sending the result to the tuning operation server 3 is loaded and started.

  In addition, on the tuning operation server 3, the tuning operation registration from the recording medium 31 in the memory of the tuning operation server 3, the execution instruction by the operation transmission to the tuning target server 2, and the tuning target server 2 The program 32 for receiving the result is loaded and activated.

  The program 22 includes a tuning operation reception unit 221, a tuning operation execution unit 222, a tuning result reception unit 223, a tuning result transmission unit 224, and a result storage file 225.

  The program 32 includes a tuning operation registration unit 321, a tuning operation transmission unit 322, a tuning result reception unit 323, an operation information file 324, a knowledge base (knowledge) 327, and a result storage file 328. If verification is performed in advance in a verification environment similar to the commercial environment, the parameter information file 52 and the performance information file 53 collected by the OS and application 51 are copied from the verification server 5 in the verification environment. The prepared parameter information file 325 and performance information file 326 are prepared.

  Here, with respect to the tuning target server 2, the range of use can be expanded by adding functions to various OSs such as UNIX (registered trademark) and Windows (registered trademark). Further, the tuning operation server 3 is not particularly specified as long as the same function can be realized.

[Operation]
2, 3, and 4 are flowcharts of each processing portion that configures the program 32 that is activated by the tuning operation server. FIGS. 5, 6, 7, and 8 are flowcharts of processing portions constituting the program 22 started up on the tuning target server. Hereinafter, description will be made in order along the flow of processing.

First, the processing of the tuning operation registration unit 321 will be described.
FIG. 2 is a flow of the tuning operation registration unit 321 shown in FIG.
Here, it is a part for registering the setting information for performing the tuning operation.
The operation information registered in the operation information file 324 is described in a text format as shown in FIG.
FIG. 9 is a diagram showing an example of operation information registered in the operation information file 324 shown in FIG. Table 1 is a table showing identification names, forms, items to be specified, and descriptions.

There are the following methods for registering the setting information.
The first method is a format in which a registration screen as shown in FIG. 10 is prepared, necessary items are selected / described in the registration screen, and finally written in the operation information file 324 via the registration screen.
FIG. 10 is a diagram showing an example of a registration screen used in the system shown in FIG. Table 2 is a table showing identification names, forms, designated items, and descriptions.

  The second method is a parameter information file 52 or a performance information file collected by the OS and application 51 from the verification server 5 in the verification environment if the verification is performed in advance in a verification environment similar to the commercial environment. In this method, a parameter information file 325 and a performance information file 326 obtained by copying 53 are used.

Based on the above information, first, a target server to be tuned is designated as setting information (processing 11).
It is confirmed whether or not the parameter information file 325 exists (process 12). If the parameter information file 325 exists (process 12 / Yes), the parameters contained in the parameter information file 325 and their parameter values are temporarily stored. The operation information file 324 is saved (process 13).

Next, it is confirmed whether or not the performance information file 326 exists (process 14). If the performance information file 326 exists (process 14 / Yes), the performance statistical information described in the performance information file 326 is displayed. Then, using the knowledge base (knowledge) 327, the parameter to be set and its parameter value are automatically determined and stored in the operation information file 324 (process 15).
At this time, if the parameter and the parameter value once saved in the operation information file 324 in the process 13 are the same as the parameters determined in the process 15, the parameter value of the parameter determined in the process 15 is changed. Prioritize.

Thereafter, a parameter to be tuned is selected (process 16).
Here, when the parameter determined in the process 15 is selected and set in the process 16, the parameter value selected and set in the process 16 is prioritized.
If the performance information file 326 does not exist in the process 14 (process 14 / No), the parameter to be tuned is selected (process 16).

  When the parameters to be set and their parameter values are gathered, the mutual relationship between the parameters is checked using the knowledge base (knowledge) 327 (process 17), and when there is a parameter having a problem with the mutual relationship between the parameters (process 18 / In Yes), the parameter name having a problem with the correlation and the details of the reason are output as an error message (process 19). After the error message is output, the process returns to process 16 for input again.

  If there is no problem in the check using the knowledge base (knowledge) 327 (process 18 / No), the item to be confirmed as the tuning result after parameter setting is selected (process 20), and tuning is performed after setting the tuning parameter. The number of hours to be executed is designated (process 21), and the set information is registered and saved in the operation information file 324 (process 22).

In the above series of processes, it is essential to create a knowledge base (knowledge) 327 in advance.
The parameters selected in the process 12 are mainly parameters related to the OS, but depending on the addition of functions, it is possible to select parameters related to the application built in the tuning target server 2.

As a format of the parameter setting value set in the process 12, as shown in FIG. 11, a form in which a specified number is sequentially increased from a certain setting value can be considered.
FIG. 11 is a diagram showing an example of parameter setting values used in the system shown in FIG. Table 3 is a table showing identification names, forms, items to be specified, and descriptions.

In the process 21, the execution time is specified as shown in the example of FIG. 12, and a series of operations “start of specified time−tuning parameter setting−execution of tuning operation−end of specified time” can be performed for the specified number of executions. It can also be considered.
FIG. 12 is a diagram showing an example of parameter setting values used in the system shown in FIG.

Next, processing of the tuning operation transmission unit 322 will be described.
FIG. 3 is a flow of the tuning operation transmission unit 322 shown in FIG.
The setting information registered and stored in the operation information file 324 in the process 15 of the tuning operation registration unit 321 is extracted (process 21), and the setting information is transmitted from the tuning operation server 3 to the tuning target server 2 (process 22).

  As a transmission method in the process 22, transmission by process communication or the like, file transfer by the ftp protocol, or the like can be considered.

Next, processing of the tuning operation receiving unit 221 will be described.
FIG. 5 is a flow of the tuning operation receiving unit 221 shown in FIG.
The setting information is received from the tuning operation server 3 to the tuning target server 2 (process 41), the received setting information is analyzed and extracted (process 42), and the extracted setting information is passed to the tuning operation execution unit (process 43). .

  Note that the content analysis function of the tuning operation receiving unit 221 is added in accordance with the format of the setting value specified in the processing 11 to the processing 14 of the tuning operation registration unit 321.

  Since parameter setting depends on the structure of the OS and the application 23, the setting process needs to be created in accordance with these structures.

Next, processing of the tuning operation execution unit 222 will be described.
FIG. 6 is a flow of the tuning operation execution unit 222 shown in FIG.
After the parameters related to the OS and the application 23 are set in the tuning target server 3 in accordance with the specified values and items, execution as a system after tuning is started (process 51).

  For items to be confirmed as tuning results after parameter setting, the results are automatically stored in the result storage file in the tuning target server 2 from the OS and application 23.

Next, processing of the tuning result receiving unit 223 will be described.
FIG. 7 is a flow of the tuning result receiving unit 223 shown in FIG.
It is checked whether or not the number of hours for executing tuning has been reached after setting the tuning parameter specified in the setting information (step 61). If the number of hours for executing tuning has not been reached (step 61 / No), step 61 is repeated. When the number of times has been reached (process 61 / Yes), the value of the item to be confirmed as the tuning result after parameter setting is extracted from the result storage file in the tuning target server 2 (process 62), and the tuning result transmission unit It passes to 224 (process 63).

Next, processing of the tuning result transmission unit 224 will be described.
FIG. 8 is a flow of the tuning result transmission unit 224 shown in FIG.
The value of the item to be confirmed as the tuning result after the parameter setting passed from the tuning result receiving unit 223 is transmitted from the tuning target server 2 to the tuning operation server 3 (process 71).

  As a transmission method in the processing 71, transmission by process communication or the like, file transfer by the ftp protocol, or the like can be considered.

Finally, the processing of the tuning result receiving unit 323 will be described.
FIG. 4 is a flow of the tuning result receiving unit 323 shown in FIG.
The value of the item to be confirmed as the tuning result transmitted from the tuning target server 2 to the tuning operation server 3 is received (process 31), and the information is stored in the result setting file as it is (process 32).

The values of items to be confirmed as tuning results to be saved in the result setting file are saved in text format. Therefore, it can be confirmed with an editor or the like.
In this way, tuning can be set automatically.

[Effect]
(1) Parameters can be changed automatically without human intervention.
(2) Since tuning is performed automatically, the time required for tuning can be shortened.
(3) When setting the parameter value manually, it automatically checks whether it is an appropriate value, thereby preventing setting errors and reducing labor costs required for tuning. Can do.

It is a conceptual diagram which shows one Example of the automatic tuning method of a system parameter using the network which concerns on this invention. It is a flow of the tuning operation registration part 321 shown in FIG. It is a flow of the tuning operation transmission part 322 shown in FIG. It is a flow of the tuning result receiving part 323 shown in FIG. It is a flow of the tuning operation reception part 221 shown in FIG. It is a flow of the tuning operation execution part 222 shown in FIG. It is a flow of the tuning result receiving part 223 shown in FIG. It is a flow of the tuning result transmission part 224 shown in FIG. It is a figure which shows an example of the operation information registered into the operation information file 324 shown in FIG. It is a figure which shows an example of the screen for registration used for the system shown in FIG. It is a figure which shows an example of the parameter setting value used for the system shown in FIG. It is a figure which shows an example of the parameter setting value used for the system shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Network 2 Tuning object server 3 Tuning operation server 4 System before commercial introduction 5 Verification server 21, 31 Recording medium 22, 32 Program 23, 51 Application 52 Parameter information file 53 Performance information file 221 Tuning operation receiving part 222 Tuning operation execution part 223 323 Tuning result reception unit 224 Tuning result transmission unit 225 Result storage file 321 Tuning operation registration unit 322 Tuning operation transmission unit 324 Operation information file 325 Parameter information file 326 Performance information file 327 Knowledge base (knowledge)
328 Result storage file

Claims (6)

Perform the case classification based on the combination of multiple parameter values assumed in the target parameter tuning of the target server that is the target of tuning of the system using the network,
Register the case classification result on the operation server that performs the tuning operation.
According to the instruction from the operation server, the parameter value is set in the target server according to each combination,
The system parameter is automatically tuned by returning the tuning result in the target server to the operation server.
A system parameter automatic tuning method using a network characterized by the above. Store the parameter information on the server operating in the verification environment in a file,
2. The system parameter automatic using the network according to claim 1, wherein the parameter value is automatically reflected on the target server by the parameter information value by placing the file on the operation server as it is. Tuning method. Store the output information collected by the performance characteristics tool in the verification environment in a file,
By placing the file as it is on the operation server, the target parameter value is automatically determined from the tendency of the output information,
2. The method for automatically tuning system parameters using a network according to claim 1, wherein the parameter value is automatically reflected on the target server with the determined value. Prepare a standard tuning pattern for each application such as web or database in advance.
2. The method for automatically tuning system parameters using a network according to claim 1, wherein necessary portions are corrected based on a set value in a selected tuning pattern. Check whether or not it is an appropriate parameter value for a plurality of parameters registered and set in the operation server,
5. The system parameter automatic tuning method using a network according to claim 1, wherein an error message is urged by the operation server for a problematic parameter value. In addition to the correlation between parameters, if the agreed items for tuning the commercial installation system have been determined in advance, set the items to be registered in advance,
When a condition that deviates from the agreed items occurs during tuning, it is checked whether or not the parameter values are appropriate for a plurality of parameters registered and set in the operation server, and the correlation between the parameters is satisfied. 5. A method for automatically tuning system parameters using a network according to claim 1, wherein a knowledge base for checking whether or not the system is used is created and utilized.

Images