Method of returning data during insert statement processing

The present invention relates to relational database management systems, and more particularly, to the update and retrieval of data from relational databases by the management systems.

Distributed relational database management systems (RDMS) are well known in the art. FIG. 1 illustrates a conventional RDMS. The system includes a client 101 running an application 102 , where the client 101 is coupled to a server 104 via a network 103 . The server 104 includes a database engine 105 , which in turn includes a query processor 106 and a data manager 107 . The server 104 manages access to the database 108 by the application 102 at the client 101 . Typically, Structured Query Language (SQL) queries are used to issue task requests on the database 108 . The query processor 106 processes and executes these queries or statements, accessing the database 108 via the data manager 107 , utilizing application programming interfaces (API).

When data in the database 108 is to be updated, the application 102 sends one or more SQL statements across the network 103 to the server 104 . For example, INSERT, UPDATE, or DELETE statements may be used. Data in one or more rows of the database 108 are then inserted, updated, or deleted. Sometimes, the database 108 generates values in one or more columns in the row(s) as well. For example, a row pertaining to a sale is inserted, with values in the customer name column, payment information column, and items purchased column. The database then also inserts a timestamp in a fourth column in the same row. The database updates additional columns based on the definition of columns in the table. Values updated by a user can also be modified due to database constraints or due to triggers on the database table.

For a user to know what values were actually written to disk, separate SQL statements must be issued to fetch the data in the updated row(s), such as a SELECT statement. This requires the building of a separate execution plan for each SQL statement, and requires multiple crossings of the data manager 107 . As is known in the art, an execution plan outlines how a server's query processor will actually run a specific query or statement. The query processor 106 parses and optimizes the SQL statement and generates the execution plan. The plan is then executed, using the data manager 107 to manipulate the data. These multiple API crossings between the client 102 , server 104 , query processor 106 , and data manager 107 are inefficient. In some eases, for example, when the database generates a primary key, the only unique value in the data is the generated value. In these cases, it is not possible to retrieve the data that had just been inserted into the table since the primary key is not known by the application.

Accordingly, there exists a need for a method for generating an execution plan for updating and retrieving data from a database in a single process. The method should allow a single statement to be issued for data to be updated and retrieved from the database. The execution plan for the statement should require a single API crossing between the client and the database. The present invention addresses such a need.

A method for generating an execution plan for updating and retrieving data from a database in a single process includes: receiving a statement by a server to update a database with a first set of data and to retrieve a second set of data from the database; building a first execution plan to update the database with the first set of data; building a second execution plan to retrieve the second set of data from the database; and building a single execution plan including a combination of the first and second execution plans. The single execution plan allows for the updating and retrieval of data with a single crossing of an interface between the client and the database. By performing both functions in this manner, the efficiency of performing such tasks on the database is significantly increased.

FIG. 1 illustrates a conventional relational database management system.

FIG. 2 is a flowchart illustrating a preferred embodiment of a method for generating an execution plan for updating and retrieving data from a database in a single process in accordance with the present invention.

FIG. 3 is a flowchart illustrating the method for generating an execution plan for updating and retrieving data from a database in a single process in a Structured Query Language context.

FIG. 4 is a flowchart illustrating in more detail the execution of the single execution plan for an example SQL statement.

FIG. 5 is a flowchart illustrating the execution of the single execution plan for a second example SQL statement.

FIG. 6 is a flowchart illustrating the execution of the single execution plan for a third example SQL statement.

FIG. 7 is a flowchart illustrating the execution of the single execution plan for a fourth example SQL statement.

The present invention provides a method for generating an execution plan for updating and retrieving data from a database in a single process. The following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a patent application and its requirements. Various modifications to the preferred embodiment will be readily apparent to those skilled in the art and the generic principles herein may be applied to other embodiments. Thus, the present invention is not intended to be limited to the embodiment shown but is to be accorded the widest scope consistent with the principles and features described herein.

To more particularly describe the features of the present invention, please refer to FIGS. 2 through 7 in conjunction with the discussion below.

FIG. 2 is a flowchart illustrating a preferred embodiment of a method for generating an execution plan for updating and retrieving data from a database in a single process in accordance with the present invention. First, the server 104 receives a statement to update the database 108 with a first set of data and to retrieve a second set of data from the database 108 , via step 202 . The statement can be from an application 102 at a client 101 or be received locally at the server 104 . The second, retrieved set of data can be the same rows involved in the first, updated set of data, or it can be more or less than the first set of data. Once the statement is received, a first execution plan to update the database 108 with the first set of data is built, via step 204 . Then, a second execution plan to retrieve the second set of data is built, via step 206 . Next, a single execution plan comprising a combination of the first and second execution plans is built, via step 208 . In the preferred embodiment, as the single execution plan is being built, storage for each of the data in the second set of data can be obtained. The statement is then executed according to this single execution plan, via step 210 . The building of the single execution plan is an important aspect of the present invention. It sets forth how the query processor 106 would update the first set of data and retrieve the second set of data with a single crossing of the data manager 107 . By performing both functions in this manner, the efficiency of performing such tasks on the database 108 is significantly increased.

In the preferred embodiment, SQL is used to perform tasks on the database 108 . A single SQL statement, with the syntax, SELECT . . . FROM FINAL TABLE (INSERT . . . ) is used to issue an update and a retrieval of data of the database 108 in a single process. However, other types of statements with different syntax can also be used without departing from the spirit and scope of the present invention.

For example, assume that the following SQL statement was issued to the server 104 :

SELECT C 1 , C 2 , C 3 FROM FINAL TABLE (INSERT into T 1 (C 1 , C 2 ) VALUES (:hv 1 , :hv 2 ))

This SQL statement requests that the values of the variables, hv 1 and hv 2 , be inserted into columns C 1 and C 2 in table T 1 . After updating the table T 1 , the actual values stored in columns C 1 , C 2 , and C 3 are to be retrieved from the table T 1 .

To illustrate the execution of this SQL statement in accordance with the present invention, please refer to FIG. 3 . FIG. 3 is a flowchart illustrating the method for generating an execution plan for updating and retrieving data from a database in a single process in a SQL context. When the server 104 receives the SQL statement above, the query processor 106 first parses the SQL statement, via step 302 . If the SQL statement is determined to not be an INSERT in a SELECT statement, via step 304 , then an execution plan for the SQL statement is built in a conventional manner, via step 312 . However, with the example SQL statement above, the SQL statement is an INSERT in a SELECT statement. Thus, the query processor 106 dismantles the SELECT portion of the statement from the INSERT portion of the statement. An execution plan for the SELECT portion of the statement is then built, via step 306 . An execution plan for the INSERT portion of the statement is also built, via step 308 . Both the execution plans for the SELECT and INSERT portions can be built in a conventional manner. Next, a single execution plan with the combination of the executions plans for the SELECT and INSERT portions of the statement is built, via step 310 .

The query processor 106 then executes the single execution plan, via step 314 . If the SQL statement was not an INSERT in a SELECT statement, via step 316 , then the SQL statement is processed, via step 322 , according to the execution plan conventionally created in step 312 . Since the example SQL statement is an INSERT in a SELECT statement, then the query processor 106 sends a request to the data manager 107 , via step 318 , to insert data and retrieve the data in a single process according to the single execution plan created in step 310 . The data manager 107 then executes the single execution plan, via step 320 . Thus, only one crossing of the API between the query processor 106 and the data manager 107 is required to both update and retrieve data from the database 108 .

For the example SQL statement above, the data manager 107 inserts the values of hv 1 and hv 2 in columns C 1 and C 2 of table T 1 . In doing so, the data manager 107 has a pointer to the row with these new values. Then, using the same pointer, the data manager 107 fetches the actual data in columns, C 1 , C 2 , and C 3 , in the table T 1 . This data is then returned to the application that issued the SQL statement.

FIG. 4 is a flowchart illustrating in more detail the execution of the single execution plan for the example SQL statement. First, the result table for the example SQL statement is determined, via step 401 . To determine the result table, the data manager 107 first inserts the values of the host variables, hv 1 and hv 2 , from buffers into the columns, C 1 and C 2 , specified in the INSERT portion of the statement, via step 402 . The buffers would have been created during the creation of the execution plan. Then, the data manager 107 returns the values in the columns, C 1 , C 2 , and C 3 , specified in the SELECT portion of the statement to the buffers, via step 403 . Once the result table is determined in this manner, the query processor 106 returns the data in the buffers to the requesting application, via step 404 .

In addition to updating and retrieving data, other database capabilities can also be made available through the method in accordance with the present invention. For example, the filtering and sorting of data can be used. In the SQL context, predicates, cursors, subselects, and ordered by functionalities can be made available. In fact, the full capabilities of SQL can be offered.

Consider a second example SQL statement:

DECLARE CS 1 CURSOR FOR

SELECT C 1 , C 2 , C 3 FROM FINAL TABLE

(INSERT into TI(C 1 , C 2 )

(SELECT C 4 , C 5 FROM Stuff))

WHERE Numstuff>:hv 3 ;

A single execution plan is built for the second example SQL statement according to steps 302 through 304 , as described above in FIG. 3 . FIG. 5 is a flowchart illustrating the execution of the single execution plan for the second example SQL statement. First, the result table is determined, via step 501 , where more than one row may qualify. To determine the result table, first, a workfile is created, via step 502 . Then, a row containing the columns, C 4 and C 5 , specified in the subquery, is selected from the subquery table, Stuff, via step 503 . The values of these columns, C 4 and C 5 , are then inserted into the target table, T 1 , in the columns, C 1 and C 2 , specified in the INSERT portion of the statement, via step 504 . The values of the columns, C 1 , C 2 , and C 3 , specified in the SELECT portion of the statement are then inserted into the workfile, via step 505 . Here, while the values of columns C 1 and C 2 are updated, the value of column C 3 is also returned, along with the updated values of columns C 1 and C 2 . If more rows quality, via step 506 , then steps 503 through 505 are repeated for each row. Once all qualifying rows are processed, the filter is performed, via step 507 . For the second example SQL statement above, all values in the workfile where Numstuff is greater than the value of the host variable, hv 3 , are retained. Once the results table is determined in this manner, the query processor 106 returns the values from the workfile to the requesting application, via step 508 .

Consider a third example SQL statement:

DECLARE CS 1 CURSOR FOR

SELECT C 1 , C 2 , C 3 FROM FINAL TABLE

(INSERT into TI(C 1 , C 2 )

(SELECT C 4 , C 5 FROM Stuff))

WHERE Numstuff>:hv 3

ORDER BY C 2 ;

A single execution plan is built for the third example SQL statement according to steps 302 through 314 , as described above in FIG. 3 . FIG. 6 is a flowchart illustrating the execution of the single execution plan for the third example SQL statement. First, the result table is determined, via step 601 , where more than one row may qualify. To determine the result table, first, a workfile is created, via step 602 . Then, a row containing the columns, C 4 and C 5 , specified in the subquery is selected from the subquery table, Stuff, via step 603 . The values of these columns, C 4 and C 5 , are then inserted into the target table, T 1 , in columns, C 1 and C 2 , specified in the INSERT portion of the statement, via step 604 . The values of the columns, C 1 , C 2 , and C 3 , specified in the SELECT portion of the statement are then inserted into the workfile, via step 605 . If more rows quality, via step 606 , then steps 603 through 605 are repeated for each row. Once all qualifying rows are processed, the filter is performed, via step 607 . For the third example SQL statement above, all values in the workfile where Numstuff is greater than the value of the host variable, hv 3 , are retained. Next, a sort workfile is created, via step 608 . The values from the filtered workfile is inserted into the sort workfile, via step 609 . The sort is then performed, via step 610 . According to the ORDER BY portion of the third example SQL statement, the values in the sort workfile are sorted according to the values in the column, C 2 . Once the results table is determined in this manner, the query processor 106 returns the values from the sort workfile to the requesting application, via step 611 .

Consider a fourth example SQL statement:

DECLARE CS 1 INSENSITIVE SCROLL CURSOR FOR

SELECT C 1 , C 2 , C 3 FROM FINAL TABLE

(INSERT into TI(C 1 , C 2 )

(SELECT C 4 , C 5 FROM Stuff))

WHERE Numstuff>:hv 3

ORDER BY C 2 ;

A single execution plan is built for the fourth example SQL statement according to steps 302 through 314 , as described above in FIG. 3 . FIG. 7 is a flowchart illustrating the execution of the single execution plan for the fourth example SQL statement. First, the result table is determined, via step 701 , where more than one row may qualify. To determine the result table, first, a workfile is created, via step 702 . Then, a row containing the columns, C 4 and C 5 , specified in the subquery is selected from the subquery table, Stuff, via step 703 . The values of these columns, C 4 and C 5 , are then inserted into the target table, T 1 , in the columns, C 1 and C 2 , specified in the INSERT portion of the statement, via step 704 . The values of the columns, C 1 , C 2 , and C 3 , specified in the SELECT portion of the statement are then inserted into the workfile, via step 705 . If more rows quality, via step 706 , then steps 703 through 705 are repeated for each row. Once all qualifying rows are processed, the filter is performed, via step 707 . For the fourth example SQL statement above, all values in the workfile where Numstuff is greater than the value of the host variable, hv 3 , are retained. Next, a sort workfile is created, via step 708 . The values from the filtered workfile are inserted into the sort workfile, via step 709 . The sort is then performed, via step 710 . According the ORDER BY portion of the statement, the values in the sort workfile are sorted according to the values in the column, C 2 . Next, an updateable workfile is created, via step 711 . The values from the sort workfile are then inserted into the updateable workfile, via step 712 . Once the results table is determined in this manner, the query processor 106 returns the values from the updateable workfile to the requesting application, via step 713 .

When the database system is a distributed database system, during processing of any of the example SQL statements above, the server 104 additionally returns an indicator that data has been updated in the database 108 . This indicator is important for commit transaction processing, as the requester needs to know if any updates occurred, and if so, to commit the changes. If no updates were made, then no commit processing is needed.

The method in accordance with the present invention takes advantage of new database inter-component communications. The query processor 106 executes the SQL statement while the data manager 107 component manipulates the data. Thus, the query processor 106 tells the data manager 107 what to do. With the present invention, this inter-component communication is minimized by having the query processor 106 inform the data manager 107 to update data and retrieve data at the same time, i.e., with one crossing of the API between the query processor 106 and the data manager 107 . Execution costs are saved by not calling the API a second time to retrieve the values.

A method for generating an execution plan for updating and retrieving data from a database in a single process has been disclosed. The method builds a single, combined execution plan for updating and retrieving data with a single crossing of the API between the client and the database. By performing both functions in this manner, the efficiency of performing such tasks on the database is significantly increased. A single statement can be used to perform both functions. If errors occur in retrieving the data, the insertion of data is not necessarily affected. In addition, the full capabilities of database, such as the filtering and sorting of data, can be made available.

Although the present invention has been described in accordance with the embodiments shown, one of ordinary skill in the art will readily recognize that there could be variations to the embodiments and those variations would be within the spirit and scope of the present invention. Accordingly, many modifications may be made by one of ordinary skill in the art without departing from the spirit and scope of the appended claims.