JP4031947B2 - Query optimization processing device, query optimization processing method, program for causing computer to execute the method, and recording medium storing program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a query optimization processing device that performs query optimization processing in a heterogeneous database integration system using a wrapper architecture, a query optimization processing method, a program that causes a computer to execute the method, and a recording medium that records the program.
[0002]
[Prior art]
The query optimization process is a process for responding to a data operation request expressed in a query language such as SQL in a relational database in as little time as possible. Many methods of optimization processing are known so far. Many of them use a method in which a plurality of access plan candidates for executing a data operation request are listed and the one with the smallest cost (answer time) estimate is selected. For this cost estimation, statistical information such as the number of data and the value selection rate is used.
[0003]
As a reference for the optimization processing, there is [1] “An Overview of Query Optimization in Relational Systems” (Surajit Chaudhuri, PODS '98, p. 34-43).
[0004]
A method for integrating heterogeneous databases using a wrapper architecture is disclosed in Japanese Patent Application Laid-Open No. 2000-112982, “Heterogeneous Database Integration System”. In this system, data managed by each of one or more database servers is managed as one or more tables, and the integrated system is regarded as one relational database. The query query request given to the integrated system is decomposed for each table managed by each database server, and information on the estimated time cost and number of results required to process the query from the wrapper module corresponding to each database server And an optimal access plan is generated according to the result.
[0005]
[Problems to be solved by the invention]
However, optimization of query processing in a heterogeneous database integration system requires interaction through a wrapper module, and optimization itself may take time that cannot be ignored. That is, when the same query as the previously optimized query is requested, the time required for the optimization process can be shortened by reusing the previous result. However, there is a possibility that the access plan obtained by the previous optimization result due to the update operation or the like is not already in an optimal state, and in this case, the response time of the database server to the query becomes long. At the same time, the problem arises that it is impossible to know that an inappropriate access plan is being used.
[0006]
In order to solve the above-described problems caused by the conventional technology, the present invention reuses an access plan that is optimal as much as possible while reducing the number of times of interaction with the wrapper module, and reuses an access plan that is already in a non-optimal state. It is an object of the present invention to provide a query optimization processing apparatus and method capable of performing a prevented optimization process, and a program for causing a computer to execute the method.
[0007]
[Means for Solving the Problems]
In order to solve the above-described problems and achieve the object, a query optimization processing apparatus according to the present invention optimizes a search for a query request input to a heterogeneous database integration system using a wrapper architecture. , A query request input means for receiving a query request input from a user, one or more database servers each managing one or more table data, and a search server provided corresponding to each database server. A wrapper module that estimates the time cost and the estimated number of results when executing the query request, and converts the query request from the query request input means into an operator tree structure according to the estimate obtained from the wrapper module Generate an access plan An access plan generating means, an access plan executing means for issuing an inquiry request to each database server according to the access plan generated by the access plan generating means, or executing a search process instead of the database server; The access plan corresponding to the query given by the query request input means and the history of the access plan generated by the access plan generation means is stored if the query has a semantically equivalent query to the newly given query. The access plan holding means for passing the data to the access plan generating means and the access plan holding means when an operation is performed to change the contents of the table data managed by the database server. The corresponding access program An access plan discarding unit for instructing to discard the access plan, and each access plan discarding unit sets and holds a threshold value for the number of times the contents of the table data change, and sets the number of times the contents of the table data changes. If it is detected and the number of times is compared with the threshold value and the threshold value is exceeded, the access plan holding unit is instructed to discard the access plan exceeding the threshold value.
[0008]
According to the present invention, by setting a threshold value in the access plan discarding means, even when the content of data managed by the database server changes, it is necessary to immediately change the access plan in response to a search inquiry request. When there is no significant influence, it is possible to prevent the access plan from being discarded unnecessarily, and the optimization efficiency can be improved.
[0009]
Also, the query optimization processing device of the present invention is a query optimization processing device that optimizes the search for query requests input to a heterogeneous database integration system using a wrapper architecture, and that receives a query request input from a user. A request input means; one or more database servers each managing one or more table data; and a time cost for executing a query inquiry request provided for each database server. A wrapper module that estimates a prediction of the number of results, an access plan generation unit that converts an inquiry request from the inquiry request input unit into an operator tree structure, and generates an access plan according to an estimate obtained from the wrapper module; and access An inquiry request is issued to each database server according to the access plan generated by the run generation means, or an access plan execution means for executing a search process in place of the database server, and given by the inquiry request input means The inquiry and the history of the access plan generated by the access plan generation means are held, and if there is an inquiry that is semantically equivalent to the newly given inquiry in the history, the corresponding access plan is passed to the access plan generation means. When an operation that changes the contents of the table data managed by the database server is performed corresponding to each access plan holding means and each database server, the corresponding access plan is discarded for the access plan holding means. Access plastic to instruct The access plan execution means measures the time taken for each execution of the inquiry request, and the access plan holding means takes the execution of each access plan together with the history of the inquiry. The access plan discarding means sets a threshold value of a ratio to the minimum value of the execution time, and when the ratio between the execution time and the minimum value of the access plan exceeds the threshold value, the access plan discarding means An instruction to discard a plan and an access plan using data of the same database server is provided.
[0010]
According to the present invention, when there is a significant change in the response time of the database server due to a factor other than the information monitored by the access plan discarding means, it is possible to know that an inappropriate access plan has been used. The number of times of using a simple access plan can be reduced, and the optimization efficiency can be improved.
[0011]
Also, the query optimization processing method of the present invention is a query optimization processing method for optimizing a search for a query request input to a heterogeneous database integration system using a wrapper architecture to optimize a query to a database server. An inquiry request input step in which an input means receives an input of an inquiry request from a user, and a search for each of the database servers provided with one or more wrapper modules each managing one or more table data according to the inquiry request A wrapper module processing step for estimating a time cost and a prediction of the number of results when executing an inquiry request, and an access plan generation unit converts the inquiry request into an operator tree structure, An access plan generation step for generating an access plan according to the estimate obtained by the processing step, and an access plan execution means issues an inquiry request to the database server according to the access plan generated at the access plan generation step. Or an access plan execution step for executing a search process in place of the database server, an access plan holding means for inputting an inquiry request input in the inquiry request input step, and an access plan generated in the access plan generation step. An access plan holding step for holding an access plan corresponding to a newly inputted inquiry request in the history and storing the corresponding access plan to the access plan generating step; The database An access plan discarding step for instructing to discard the corresponding access plan in the access plan holding step when there is a change. The process sets and holds the threshold of the number of times the contents of the table data change, detects the number of times the contents of the table data change, and compares the number of times with the threshold to hold the access plan. An instruction is given to discard an access plan that exceeds the threshold held in the process.
[0012]
Also, the query optimization processing method of the present invention is a query optimization processing method for optimizing a search for a query request input to a heterogeneous database integration system using a wrapper architecture to optimize a query to a database server. One or more inquiry management input steps in which the input means receives an inquiry request from the user and one or more wrapper modules each manage one or more table data according to the inquiry request input in the inquiry request input step. A wrapper module processing step for estimating a time cost and a prediction of the number of results when executing a search query request to each of the database servers to be accessed, and an access plan generation means is input by the query request input step. An access plan generation step for converting the received inquiry request into an operator tree structure and generating an access plan according to the estimate obtained by the wrapper module processing step, and an access plan execution means are generated in the access plan generation step. An access plan execution step of issuing an inquiry request to the database server according to the access plan or executing a search process instead of the database server, and an access plan holding means are input by the inquiry request input step The inquiry request and the access plan generated in the access plan generation step are held as a history, and if there is an inquiry in the history that is semantically equivalent to the newly input inquiry request, the corresponding access plan is stored in the access plan. Generation process The access plan holding process to be passed and the access plan discarding means monitor changes in the contents of the table data managed by the database server, and when there is a change, instruct to discard the corresponding access plan in the access plan holding process An access plan discarding step, wherein the access plan execution step measures the time taken to execute the inquiry request every time the inquiry request is executed, and the access plan holding step holds each inquiry request history together with the history of the inquiry request. A minimum value of the time taken to execute the access plan is held, and in the access plan discarding step, a threshold value of a ratio to the minimum value of the execution time is set, and a ratio between the execution time and the minimum value of the access plan is the threshold value Use the access plan and data from the same database server An instruction to discard the access plan is provided.
[0013]
A program to be executed by a computer according to the present invention causes the computer to execute the above method.
[0014]
A recording medium according to the present invention records the above program.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
Exemplary embodiments of an inquiry optimization processing apparatus, an inquiry optimization processing method, a program for causing a computer to execute the method, and a recording medium storing the program will be described in detail below with reference to the accompanying drawings. To do.
[0020]
(Embodiment 1)
FIG. 1 is a block diagram showing a system configuration of a query optimization processing apparatus according to the present invention. A client (terminal) 101 and a plurality (two in the illustrated example) of database servers 102 (102a and 102b) are connected to the query optimization processing apparatus 100.
[0021]
The query optimization processing apparatus 100 includes a query request input unit 111, an access plan generation unit 112, an access plan holding unit 113, and an access plan execution unit 114. Further, a wrapper module 121a and access plan discarding means 122a in the database server 102a, and a wrapper module 121b and access plan discarding means 122b in the database server 102b are provided. In FIG. 1, the connection configuration of the wrapper module 121b and the access plan discarding means 122b provided on the database server 102b side which is the lower level apparatus and the higher level apparatus (main means of the query optimization processing apparatus 100) is the database server 102a side. The description is omitted.
[0022]
FIG. 2 is a block diagram illustrating a specific example of the hardware configuration of the query optimization processing apparatus 100. The user inputs an inquiry request from the client 101, for example, in the form of an SQL sentence. The input inquiry request is sent to the application server 202 constituting the inquiry optimization processing apparatus 100 through a communication line (LAN or the like) 201. The application server 202 includes a CPU 203, a memory 204 having a program area and a data area, and a storage device (hard disk or the like) 205.
[0023]
(Access plan generation means 112)
The access plan generation unit 112 first checks whether the access plan holding unit 113 has an access plan that can process the inquiry request. If it is not in the access plan holding means 113, an access plan is created as follows. First, the inquiry request is converted into a tree structure representing a combination of relational algebras. The operators (processes) corresponding to the relational algebra are as follows.
[0024]
Access: The table data of the database server 102 is fetched sequentially.
Select: Filter data according to conditions.
Project: Extracts some of the attributes that make up the tuple of the table.
Join: Joins two tables according to conditions.
Union: Creates the sum of two tables.
Intersect: Creates the product of two tables.
Difference: Creates the difference between two tables.
[0025]
Among the above, Access is not a relational algebra operator, but is a minimum necessary function when the database server 102 does not have a search function.
[0026]
(Access plan holding means 113)
The access plan holding unit 113 holds the inquiry request and the access plan on the memory 204 using, for example, a known hash map structure. In this case, for example, a character string of an SQL sentence representing an inquiry request can be used as the key of the hash map. The access plan holding unit 113 further holds, on the memory 204, an array that associates the database servers 102a and 102b with a list of access plans that use the data. In order to uniquely identify the database servers 102a and 102b, for example, sequence numbers uniquely assigned to the database servers 102a and 102b can be used. The access plan list may use an array of pointers indicating positions on the hash map.
[0027]
(Access plan execution means 114)
The access plan execution unit 114 executes search processing according to the access plan obtained by the access plan generation unit 112.
[0028]
(Access plan discarding means 122a, 122b)
Access plan discarding means 122 (122a, 122b) is prepared for each of the database servers 102a, 102b. The access plan discarding means 122a and 122b are not limited to operations through the wrapper module 121 (121a and 121b), and operations that change the contents of the managed table data to the database server 102 (102a and 102b). When done, the access plan holding means 113 is instructed to discard the access plan using the data of the database server 102 (102a, 102b).
[0029]
Means for monitoring that the contents have changed varies depending on the implementation form of the database servers 102a and 102b. For example, when the database servers 102a and 102b are DBMSs (database management systems), it can be assumed that the content has changed due to the update operation being performed. Further, for example, when the database servers 102a and 102b are simple files, a form in which the time stamps of the files are monitored at regular intervals can be considered.
[0030]
A specific instruction for discarding the access plan is performed by transmitting the sequence numbers of the database servers 102a and 102b to the access plan holding means 113. The access plan holding unit 113 acquires a list of access plans to be discarded from the array that associates the database servers 102a and 102b with the list of access plans by the given sequence numbers of the database servers 102a and 102b, and deletes the list from the hash map. To do.
[0031]
Next, FIG. 3 is a diagram illustrating an example of converting an SQL statement representing a certain inquiry request into an operator tree structure. The access plan generation unit 112 creates one or more tree structures representing queries, extracts partial trees that use the table data managed by each database server 102 (102a, 102b) for each tree structure, and extracts each database. An estimate is instructed to the wrapper modules 111a and 111b corresponding to the servers 102a and 102b. Since the cost when the sub-trees are processed by the database servers 102a and 102b is obtained from the responses from the wrapper modules 111a and 111b, the one with the smallest cost is selected as the access plan.
[0032]
When it is determined that the access plan execution unit 114 assumes a lower cost than the cost returned by the wrapper modules 121a and 121b by Select or Join, the cost when the access plan execution unit 114 takes over is used.
[0033]
The inquiry request shown in FIG. 3 is a specific example for an in-company employee database. The inquiry request 300 is represented by the following expression (SQL sentence).
select employee. Name, department. Name from employee, department, employee affiliation
where employee. Address like '% Bunkyo-ku%'
and employee affiliation. Employee address = employee. ID
and employee affiliation. Department number = Department. ID
[0034]
This inquiry request is “search for the name, department information and department name (name) of an employee of a person living in Bunkyo-ku by address”. The tree structure shown in FIG. 3 is a triple tree structure. One of the lower nodes uses, as employee information, a table 301 (“employee ID + name + address”), which is a table in which an employee ID, a name, and an address are associated with each other. The operator 310 is a Select (Select “address = Bunkyo-ku”) that narrows down data including a ward.
[0035]
The lower node branches from an operator 311 that is a Join (Join “Employee affiliation.Employee number = Employee.ID”) that joins the table 301 and the table 302 for data having the same employee ID. Further, these lower nodes branch from an operator 312 that is a Join (Join “Employee affiliation.Department number = Department.ID”) that joins the tables 302 and 303 with respect to data having the same employee affiliation.
[0036]
In this example, the database server 102a manages the table 301 that associates IDs, names, and addresses, which are employee information, and the database server 102b manages the table 302 that associates employee numbers and department numbers belonging to the employees. Assume that another database server (not shown in FIG. 1) manages the table 303.
[0037]
FIG. 4 is a flowchart showing an access plan generation process. Hereinafter, the contents of the illustrated process will be described. First, the inquiry request input unit 111 receives an inquiry request input from a user (step S401). Next, the access plan generation unit 112 inquires whether there is an access plan in the access plan holding unit 113 that can process this inquiry request (step S402). If there is an access plan corresponding to the access plan holding unit 113 (step S403: Yes), the process ends without generating an access plan.
[0038]
If there is no access plan that can process the inquiry request in the access plan holding means 113 (step S403: No), a tree structure expressing combinations of relational algebras of operators representing the inquiry request is listed (step S404). Next, one tree structure is extracted (step S405).
[0039]
Then, an unprocessed one is extracted from the subtree from which data can be obtained from a single database server (for example, the database server 102a) (step S406). Next, an estimate is instructed to the wrapper module 121a corresponding to the database server (102a), and the cost when the sub-tree is processed by the database server 102a is obtained (step S407). Thereafter, if there is an unprocessed subtree (step S408: Yes), the process returns to step S406 to perform processing on the unprocessed subtree. The processes in steps S406 and S407 are performed on all subtrees listed in step S405. If there is no unprocessed partial tree (step S408: No), the process proceeds to step S409.
[0040]
Next, the cost of the entire tree structure is calculated from the cost of the partial tree (step S409). Then, the access plan generating unit 112 determines whether the entire tree structure has the minimum cost (step S410). If it is the minimum (step S410: Yes), the current tree structure is made an access plan candidate (step S411). If it is not the minimum cost (step S410: No), step S411 is skipped.
[0041]
Next, it is searched whether there is still an equivalent tree structure in addition to the current tree structure (step S412). If it is still present (step S412: YES), the processing after step S405 is executed again. On the other hand, if there is no equivalent tree structure in addition to the current tree structure (step S412: No), the access plan holding means 113 (memory) is used to use the current access plan candidate processed in step S411 as an access plan. 204) (step S413), and the process ends.
[0042]
According to the above configuration, in the state where an access plan generated from a certain inquiry request is stored in the access plan holding means 113, the meaning of the inquiry request is as long as the content of data managed by the database servers 102a and 102b does not change. When an equivalent query is requested, the query can be executed without generating a new access plan. Therefore, when an inquiry request is generated, the response time of the database servers 102a and 102b can be shortened by reusing the most suitable access plan while reducing the number of times of exchange with the wrapper module.
[0043]
(Embodiment 2)
Next, a second embodiment of the inquiry optimization processing apparatus of the present invention will be described. In the second embodiment, the configuration of the access plan discarding means 122 described in the first embodiment is different.
[0044]
The access plan discarding means 122 (122a, 122b) of the second embodiment is newly provided as an area of the memory 204, a new area for storing the table data contents, and a threshold value preset as the number of times. Added to. The access plan discarding means 122 compares the number of times in the area of the memory 204 with a threshold value set after adding 1 each time the content is changed. If the number of times exceeds the threshold value, the access plan holding unit 113 is instructed to discard the access plan, and the number of times of the memory 204 is changed (reset) to “0”.
[0045]
According to the above configuration, even if the contents of the data managed by the database server 102 (102a, 102b) change, there is no significant influence on the response to the search inquiry request that requires the change of the access plan immediately. In this case, it is possible to prevent the access plan from being unnecessarily discarded by setting a threshold value in the access plan discarding unit 122 (122a, 122b).
[0046]
(Embodiment 3)
Next, a third embodiment of the inquiry optimization processing apparatus of the present invention will be described. In the third embodiment, the configurations of the access plan holding unit 113, the access plan executing unit 114, and the access plan discarding unit 122 (122a, 122b) described in the first embodiment are different.
[0047]
The access plan execution unit 114 according to the third embodiment acquires the value of the time counter at the start and end of execution, and obtains the execution time from the difference between the values. This execution time is transmitted to the access plan holding unit 113. In the general-purpose computer shown in FIG. 2, the value of the time counter can be easily obtained using a timer (not shown).
[0048]
In addition to the access plan, the access plan holding unit 113 adds an area for recording the minimum execution time to the value held in the hash map on the memory 204. When the execution time is transmitted from the access plan execution unit 114, the minimum execution time is acquired from the hash map, and the execution time and the minimum execution time are transmitted to the access plan discarding unit 122 (122a, 122b). If the access plan discarding unit 122 replies that the access plan should be discarded, the target access plan is deleted from the hash map of the access plan holding unit 113.
[0049]
In addition to the functions described in the first embodiment, the access plan discarding unit 122 has a memory area that holds a threshold on the memory 204. The access plan discarding unit 122 calculates the ratio between the execution time given from the access plan holding unit 113 and the value of the minimum execution time. If the calculated ratio exceeds the threshold, the access plan holding unit 113 stores the access plan in the access plan holding unit 113. A value indicating that should be discarded is passed as a response.
[0050]
According to the above configuration, if there is a significant change in the response time of the database server 102 (102a, 102b) due to factors other than the information monitored by the access plan discarding unit 122, it is determined that an inappropriate access plan has been used. You will know and you can reduce the number of times you use inappropriate access plans.
[0051]
(Embodiment 4)
Next, a fourth embodiment of the inquiry optimization processing apparatus of the present invention will be described. In the fourth embodiment, the configurations of the access plan holding unit 113 and the access plan discarding unit 122 (122a, 122b) described in the first embodiment are different.
[0052]
The access plan holding means 113 according to the fourth embodiment adds an array for associating the table with the list of access plans, instead of the array for associating the database server 102 (102a, 102b) with the list of access plans. In order to uniquely identify a table, for example, a sequence number is assigned to the table. Further, the access plan discarding means 122a and 122b monitor the change in contents in units of tables, and if the contents have changed, inform the access plan holding means 113 of the sequence number of the table. The access plan holding unit 113 acquires a list of access plans to be discarded from the sequence number in the table and deletes it from the hash map on the memory 204.
[0053]
According to the above configuration, for database servers that can observe changes in the contents of the database server in units of tables, the same database is used even if only the tables whose contents have not changed by discarding the access plan in units of tables are used. It is possible to prevent the reusable access plan from being discarded because another table belonging to the server has been changed.
[0054]
Note that the method related to the query optimization process described in this embodiment can be realized by executing a program prepared in advance on a computer such as a personal computer or a workstation. This program is recorded on a computer-readable recording medium such as a hard disk, floppy (R) disk, CD-ROM, MO, and DVD, and is executed by being read from the recording medium by the computer. The program may be a transmission medium that can be distributed via a network such as the Internet.
[0055]
【The invention's effect】
As described above, according to the present invention, when an access plan generated from a query request is stored in the access plan holding means, the query request and the query request are changed as long as the content of data managed by the database server does not change. When a semantically equivalent query is requested, the query can be executed without generating a new access plan. As a result, the optimal access plan can be reused while reducing the number of times of interaction with the wrapper module, and the processing time can be shortened. Even if the contents of the data managed by the database server change, the access plan is unnecessarily discarded when there is no significant impact on the response to the search query request that would require a change in the access plan. This is advantageous in that the optimization efficiency can be improved.
[0056]
Also, if there is a significant change in the response time of the database server due to factors other than the information monitored by the access plan discarding means, you can know that an inappropriate access plan has been used, and The number of times of use can be reduced, and the optimization efficiency can be improved.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a system configuration of a query optimization processing apparatus according to the present invention.
FIG. 2 is a block diagram illustrating a specific example of a hardware configuration of an inquiry optimization processing device.
FIG. 3 is a diagram illustrating an example of converting an SQL sentence representing a certain inquiry request into an operator tree structure;
FIG. 4 is a flowchart showing an access plan generation process;
[Explanation of symbols]
100 Inquiry optimization processor
101 Client (terminal)
102 (102a, 102b) database server
111 Inquiry request input means
112 Access plan generation means
113 Access plan holding means
114 Access plan execution means
121 (121a, 121b) wrapper module
122 (122a, 122b) Access plan discarding means
201 Communication line (LAN)
202 Application server
203 CPU
204 memory
205 Storage device (hard disk)
300 Inquiry request
Table 301, 302, 303
310, 311, 312 Operator

Claims (6)

In a query optimization processing device that optimizes search for query requests input to a heterogeneous database integration system using a wrapper architecture,
An inquiry request input means for receiving an input of an inquiry request from a user;
One or more database servers each managing one or more table data; and
A wrapper module that is provided for each database server and estimates the time cost and the number of results when executing a search query request,
An access plan generating means for converting an inquiry request from the inquiry request input means into an operator tree structure and generating an access plan according to an estimate obtained from the wrapper module;
An access plan executing means for issuing an inquiry request to each database server according to the access plan generated by the access plan generating means, or executing a search process instead of the database server;
The inquiry given by the inquiry request input means and the history of the access plan generated by the access plan generation means are held, and if there is an inquiry that is semantically equivalent to the newly given inquiry in the history, the corresponding access Access plan holding means for passing the plan to the access plan generating means;
Access provided corresponding to each database server and instructing the access plan holding means to discard the corresponding access plan when an operation that changes the contents of the table data managed by the database server is performed. Plan discarding means,
With
Each access plan discarding means includes:
When the threshold of the number of times the contents of the table data change is set and held, the number of times the contents of the table data change is detected, and when the number of times exceeds the threshold, the access plan holding means Inquiry optimization processing apparatus characterized by instructing to discard an access plan exceeding the threshold . In a query optimization processing device that optimizes search for query requests input to a heterogeneous database integration system using a wrapper architecture,
An inquiry request input means for receiving an input of an inquiry request from a user;
One or more database servers each managing one or more table data; and
A wrapper module that is provided for each database server and estimates the time cost and the number of results when executing a search query request,
An access plan generating means for converting an inquiry request from the inquiry request input means into an operator tree structure and generating an access plan according to an estimate obtained from the wrapper module;
An access plan executing means for issuing an inquiry request to each database server according to the access plan generated by the access plan generating means, or executing a search process instead of the database server;
The inquiry given by the inquiry request input means and the history of the access plan generated by the access plan generation means are held, and if there is an inquiry that is semantically equivalent to the newly given inquiry in the history, the corresponding access Access plan holding means for passing the plan to the access plan generating means;
Access provided corresponding to each database server and instructing the access plan holding means to discard the corresponding access plan when an operation that changes the contents of the table data managed by the database server is performed. Plan discarding means,
With
The access plan execution means measures the time taken to execute each time an inquiry request is executed,
The access plan holding means holds a minimum value of time taken to execute each access plan together with a history of inquiries,
The access plan discarding means sets a threshold value of a ratio with respect to the minimum value of the execution time, and when the ratio between the execution time and the minimum value of the access plan exceeds the threshold value, the access plan and data of the same database server A query optimization processing device that instructs to discard an access plan that uses . In a query optimization processing method for optimizing a query to a database server by optimizing a query request input to a heterogeneous database integration system using a wrapper architecture,
An inquiry request input process in which an inquiry request input means receives an input of an inquiry request from a user;
The wrapper module estimates a time cost and a prediction of the number of results when executing a query query request for each of the database servers that are provided with one or more, each managing one or more table data according to the query request. A wrapper module processing step;
An access plan generating step in which the access plan generating means converts the inquiry request into an operator tree structure and generates an access plan according to the estimate obtained by the wrapper module processing step;
An access plan execution unit that issues an inquiry request to the database server according to the access plan generated in the access plan generation step, or executes a search process instead of the database server;
The access plan holding means holds the inquiry request input in the inquiry request input step and the access plan generated in the access plan generation step as a history, and there is an inquiry that is semantically equivalent to the newly input inquiry request. An access plan holding step of passing the corresponding access plan to the access plan generation step if it is in the history;
An access plan discarding step for monitoring a change in the contents of the table data managed by the database server and instructing to discard the corresponding access plan in the access plan holding step when there is a change; Including
The access plan discarding step includes
The threshold of the number of times the contents of the table data change is set and held, the number of times the contents of the table data change is detected, and when the number of times exceeds the threshold compared with the threshold, the threshold is held by the access plan holding step A query optimization processing method characterized by instructing to discard an access plan that exceeds the threshold value. In a query optimization processing method for optimizing a query to a database server by optimizing a query request input to a heterogeneous database integration system using a wrapper architecture,
An inquiry request input process in which an inquiry request input means receives an input of an inquiry request from a user;
The time cost when the wrapper module executes a query query request for each of one or more database servers each managing one or more table data according to the query request input in the query request input step. And a wrapper module processing process to estimate the number of results,
An access plan generating step in which the access plan generating means converts the inquiry request input in the inquiry request input step into an operator tree structure, and generates an access plan according to the estimate obtained by the wrapper module processing step;
An access plan execution unit that issues an inquiry request to the database server according to the access plan generated in the access plan generation step, or executes a search process instead of the database server;
The access plan holding means holds the inquiry request input in the inquiry request input step and the access plan generated in the access plan generation step as a history, and newly An access plan holding step of passing a corresponding access plan to the access plan generation step when there is a query in the history that is semantically equivalent to the input query request;
An access plan discarding step for monitoring a change in the contents of the table data managed by the database server and instructing to discard the corresponding access plan in the access plan holding step when there is a change; Including
The access plan execution step measures the time taken for each execution of an inquiry request,
The access plan holding step holds a minimum value of time taken to execute each access plan held together with a history of inquiry requests,
In the access plan discarding step, a threshold of a ratio with respect to the minimum value of the execution time is set, and when the ratio between the execution time and the minimum value of the access plan exceeds the threshold, the access plan and the data of the same database server A query optimization processing method characterized by instructing to discard an access plan that uses. A program causing a computer to execute the method according to claim 3 or 4. A recording medium storing the program according to claim 5.

Images