JP2017126172A - Database control program, database control apparatus, and database control method - Google Patents

Abstract

An object of the present invention is to increase the speed of a storage process for storing a data group generated based on the execution result of a search process for a DB in the DB. As shown in (1), core # 0 accepts an INSERT SELECT statement 111, executes search processing 112 according to the SELECT phrase as shown in (2), and records group rd1 based on the execution result. 2 is generated. Next, the core # 0 assigns each of the record groups rd1 and rd2 to the cores # 1 and # 2 as shown in (3), and the core # 1 and # 2 as shown in (4). A request for executing a storage process 113 for storing the allocated record in the DB 102 is transmitted. Here, the core # 0 requests the search process 112 to be executed as a process included in the same transaction. The cores # 1 and # 2 execute the storage process 113 as a process included in the same transaction as the search process 112, as indicated by (5). [Selection] Figure 1

Description

  The present invention relates to a database control program, a database control apparatus, and a database control method.

  Conventionally, there is an instruction indicating that a data group generated based on an execution result of executing a search process on a database (DB: DataBase) is stored in the database. As a related prior art, for example, when a transaction includes a record addition instruction to a DB on a file, a record is added to a buffer provided for each transaction, and a record in the buffer is added to the DB by a normal end instruction of the transaction. is there.

Japanese Patent Laid-Open No. 2-138646

  However, in the prior art, it may take a long time to store the data group generated based on the execution result of executing the search process for the DB in the DB. For example, when each processor of a plurality of processors executes storage processing in parallel, if each of the plurality of processors tries to store the data independently in the DB, it is consistent while a certain processor stores the data in the DB. Since exclusive control is performed in order to maintain the performance, other processors will wait and the storage process will be delayed.

  In one aspect, the present invention provides a database control program, a database control apparatus, and a database control method for speeding up a storage process for storing a data group generated based on an execution result of executing a search process for a DB in the DB The purpose is to provide.

  According to one aspect of the present invention, each of a group of data generated based on an execution result obtained when a computer executes a search process for a DB as a process included in a transaction is one of a plurality of processors that can access the DB. Database control for transmitting a request for executing storage processing for storing data allocated to the allocation destination processor to the DB as processing included in the transaction to the allocation destination processor allocated to each of the data groups of the plurality of processors A program, a database control device, and a database control method are proposed.

  According to another aspect of the present invention, any one of data groups generated based on an execution result in which a processor included in a plurality of processors that can access a DB executes a search process for the DB as a process included in a transaction. Database control program, database control apparatus, and database control for acquiring data as a process included in a transaction, acquiring data and storing the acquired data in the database in response to the data assigned to the processor A method is proposed.

  According to one aspect of the present invention, it is possible to increase the speed of a storage process for storing a data group generated based on the execution result of a search process for a DB in the DB.

FIG. 1 is an explanatory diagram illustrating an operation example of the database control method according to the present embodiment. FIG. 2 is an explanatory diagram illustrating a configuration example of the DB management system 200. FIG. 3 is an explanatory diagram illustrating a hardware configuration example of the database control apparatus 101. FIG. 4 is an explanatory diagram illustrating a functional configuration example of the database control apparatus 101. FIG. 5 is an explanatory diagram (part 1) illustrating an example of an operation example of the leader execution body ld and the member execution body mm. FIG. 6 is an explanatory diagram (part 2) illustrating an example of an operation example of the leader execution body ld and the member execution body mm. FIG. 7 is an explanatory diagram showing a detailed example of storage in the same transaction. FIG. 8 is an explanatory diagram showing a detailed example of storage in different transactions. FIG. 9 is a flowchart (part 1) illustrating an example of a processing procedure when an INSERT SELECT statement is received. FIG. 10 is a flowchart (part 2) illustrating an example of a processing procedure when an INSERT SELECT statement is received.

  Embodiments of a disclosed database control program, database control apparatus, and database control method will be described in detail below with reference to the drawings.

  FIG. 1 is an explanatory diagram illustrating an operation example of the database control method according to the present embodiment. A database control apparatus 101 shown in FIG. 1 is a computer that controls the DB 102. The database control apparatus 101 may be, for example, a DB server having the DB 102 or a computer accessible to the computer having the DB 102.

  The DB 102 is a set of data created and managed in a certain format. For example, the DB 102 manages customer data, sales data, and the like of users accessing the database control apparatus 101 in a certain format. Then, the database control apparatus 101 executes a search process for the DB 102 or a storage process for the DB 102 according to a command from a terminal operated by the user.

  The DB 102 may be divided into a plurality of storage areas. The divided storage area is called “partition”. By dividing into partitions, the performance of the DB 102 and the ease of management can be improved. For example, the administrator of the DB 102 may design the DB 102 to store a table in which a set of data is represented by vertical columns and horizontal records in one partition. Alternatively, the administrator of the DB 102 may design the DB 102 so that each record included in one table is classified by the value of the record key and each classified group is stored in a separate partition. In the example of FIG. 1, the DB 102 is divided into partitions pt1 and pt2.

  Here, the command from the terminal operated by the user is data according to the specification of an inquiry language for performing data manipulation and definition. As an inquiry language, SQL is mentioned, for example. In the following description, a command from a terminal operated by a user will be described as SQL. For example, SQL for performing a search process has a syntax that starts with a SELECT phrase, and is called a SELECT statement. Also, the SQL that performs the storage process has a syntax that starts with the INSERT phrase, and is called an INSERT statement.

  In the SQL specification, there is a syntax indicating that a data group generated based on an execution result of executing a search process on the DB 102 is stored in the DB 102. By executing SQL according to this syntax, data obtained by processing data stored in the DB 102 can be stored in the DB 102. This SQL has a syntax that starts with an INSERT phrase and includes a SELECT phrase on the way, and is called an INSERT SELECT statement.

  When the INSERT SELECT statement is executed, the database control apparatus 101 searches the table specified by the FROM phrase after the SELECT phrase according to the condition specified by the WHERE phrase. Then, the database control apparatus 101 generates a new record group by designating a column name, an arithmetic operator, a group function, or the like of the search source table designated by the SELECT clause based on the record group satisfying the condition. . Then, the database control apparatus 101 stores the generated record group in a table specified by the INSERT phrase.

  However, when processing according to the INSERT SELECT statement is performed, it may take time to store the generated record group in the DB 102. Specifically, since the record group generated by the SELECT phrase is stored in the partition one record at a time, if the number of generated record groups increases, the time required for the storage process increases. Thus, for example, it is conceivable to increase the speed of the storage process by causing each of a plurality of processors to execute the storage process in parallel. However, if each of the plurality of processors is to be stored in a partition independently, exclusive control is performed to maintain consistency while a certain processor stores data in a partition. Therefore, the other processors wait until they can acquire the lock, and the storage process is delayed.

  Therefore, in the present embodiment, a method will be described in which each core executes the INSERT phrase as the same transaction as the SELECT phrase processing for the INSERT SELECT statement.

  An operation example of the database control apparatus 101 will be described with reference to FIG. The database control apparatus 101 has a plurality of processor cores in one processor package as a plurality of processors. Alternatively, the database control apparatus 101 may have a plurality of CPUs (Central Processing Units) instead of a plurality of processor cores. In the following description, the processor core is referred to as a “core”.

  As shown in (1) of FIG. 1, the core # 0 accepts an INSERT SELECT statement 111 as one of a plurality of cores included in the database control apparatus 101. For example, core # 0 receives an INSERT SELECT statement 111 from a client terminal, a Web server, or the like. Alternatively, the core # 0 may acquire the INSERT SELECT statement 111 included in the stored procedure to be executed by executing the stored procedure. Here, the stored procedure is a program stored in the DB 102 as a series of processing procedures for the DB 102.

  Then, as shown in (2) of FIG. 1, the core # 0 executes the search process 112 according to the SELECT phrase of the INSERT SELECT statement 111, and sets the record groups rd1 and rd2 based on the execution result of executing the search process 112. Generate. Core # 0 does not perform the search process 112 by itself, acquires the execution results of other cores executing the search processes from other cores, and generates record groups rd1 and rd2 based on the acquired execution results May be. Alternatively, the core # 0 may acquire the record groups rd1 and rd2 generated by other cores without generating the record group from the search process 112 and the execution result.

  Next, as indicated by (3) in FIG. 1, the core # 0 assigns each of the record groups rd1 and rd2 to one of the plurality of cores. In the example of FIG. 1, core # 0 assigns record rd1 to core # 1 and assigns record rd2 to core # 2. Core # 0 may assign a record to itself.

  Then, as shown by (4) in FIG. 1, the core # 0 stores the records assigned to the assignment destination core in the DB 102 in the cores # 1 and # 2 that are assignment destination cores to which the record group rd is assigned. A request to execute 113 is transmitted. Here, the core # 0 requests the cores # 1 and # 2 to execute the search process 112 as a process included in the same transaction. A transaction is a method of handling a plurality of processes as integral inseparable processes. In FIG. 1, processing groups handled as the same transaction are surrounded by a broken line.

  The cores # 1 and # 2 that have received the request execute the storage process 113 as a process included in the same transaction as the search process 112 in accordance with the received request, as indicated by (5) in FIG. In the example of FIG. 1, the core # 1 stores the record rd1 in the partition pt1 as the storage process 113. Also, the core # 2 stores the record rd2 in the partition pt1 as the storage process 113. Here, since the storage process 113 executed by the core # 1 and the storage process 113 executed by the core # 2 are the same transaction, the two storage processes 113 do not cause deadlock or waiting for exclusion. . Therefore, the database control apparatus 101 can obtain speeding up by parallel processing. Next, an example in which the database control apparatus 101 is applied as a DB server of a DB management system will be described with reference to FIG.

  FIG. 2 is an explanatory diagram illustrating a configuration example of the DB management system 200. The DB management system 200 includes a database control device 101 and a client terminal 201. The DB management system 200 is a system that manages a DB and responds to access to the DB. The database control apparatus 101 and the client terminal 201 are connected by a network 202.

  The database control apparatus 101 is a computer that manages the DB management system 200. The client terminal 201 is a computer that accesses the database control apparatus 101. Further, the DB management system 200 is not limited to the configuration shown in FIG. For example, the DB management system 200 may include a Web server, and the database control apparatus 101 may accept access to the DB from the Web server.

(Hardware configuration of database control apparatus 101)
FIG. 3 is an explanatory diagram illustrating a hardware configuration example of the database control apparatus 101. In FIG. 3, the database control apparatus 101 includes a CPU 301, a ROM (Read-Only Memory) 302, and a RAM (Random Access Memory) 303. Further, the database control apparatus 101 includes a disk drive 304 and a disk 305, and a communication interface 306. The CPU 301 to the disk drive 304 and the communication interface 306 are connected by a bus 307, respectively. Here, the disk 305 corresponds to the DB 102 shown in FIG. Further, the database control apparatus 101 may be connected to the DB 102 serving as an external apparatus via the communication interface 306.

  The CPU 301 is an arithmetic processing unit that controls the entire database control apparatus 101. The CPU 301 has a plurality of cores.

  The ROM 302 is a nonvolatile memory that stores programs such as a boot program. A RAM 303 is a volatile memory used as a work area for the CPU 301.

  The disk drive 304 is a control device that controls reading and writing of data with respect to the disk 305 according to the control of the CPU 301. As the disk drive 304, for example, a magnetic disk drive, an optical disk drive, a solid state drive, or the like can be adopted. The disk 305 is a nonvolatile memory that stores data written under the control of the disk drive 304. For example, when the disk drive 304 is a magnetic disk drive, a magnetic disk can be adopted as the disk 305. Further, when the disk drive 304 is an optical disk drive, an optical disk can be adopted as the disk 305. When the disk drive 304 is a solid state drive, a semiconductor memory formed by a semiconductor element, that is, a so-called semiconductor disk can be used as the disk 305.

  The communication interface 306 is a control device that controls an internal interface with the network 203 and controls input / output of data from other devices. Specifically, the communication interface 306 is connected to another device via the network 203 through a communication line. As the communication interface 306, for example, a modem or a LAN adapter can be employed.

  Further, when the administrator of the DB management system 200 directly operates the database control apparatus 101, the database control apparatus 101 may have hardware such as a display, a keyboard, and a mouse. Although not shown, the client terminal 201 also includes hardware included in the database control apparatus 101 and hardware such as a display, a keyboard, and a mouse.

(Functional configuration example of the database control apparatus 101)
FIG. 4 is an explanatory diagram illustrating a functional configuration example of the database control apparatus 101. The database control apparatus 101 includes a search process execution unit 401, an allocation unit 402, a transmission unit 403, an acquisition unit 404, and a storage process execution unit 405. The search processing execution unit 401 to the storage processing execution unit 405 serving as the control unit realize the functions of the respective units when each core of the CPU 301 executes a program stored in the storage device. Specifically, the storage device is, for example, the ROM 302, the RAM 303, the disk 305, etc. shown in FIG. The processing results of each unit are stored in a register of each core of the CPU 301, a cache memory of each core of the CPU 301, a RAM 303, and the like.

  The search processing execution unit 401 to the transmission unit 403 are included in the reader execution body ld. On the other hand, the acquisition unit 404 and the storage processing execution unit 405 are included in the member execution body mm. The leader executor ld performs a search process by designating the SELECT phrase, and distributes records to be stored in the partitions. The member execution body mm stores the distributed record in the partition. The search processing execution unit 401 to the storage processing execution unit 405 are included in each core of the database control apparatus 101, and any core may execute the leader execution body ld, and any core executes the member execution body mm. May be. In addition, the core that executes the leader execution body ld may execute the member execution body mm. In the example of FIG. 4, the core # 0 executes the leader execution body ld, and the cores # 1 and # 2 execute the member execution body mm.

  In response to receiving the INSERT SELECT statement 111, the search process execution unit 401 executes the search process 112 corresponding to the SELECT phrase as a process included in the transaction. Alternatively, the search process execution unit 401 executes the search process 112 as a process included in the transaction when the INSERT SELECT statement included in the stored procedure is acquired by executing the stored procedure stored in the DB 102. Also good.

  The allocation unit 402 allocates each of the record groups that are the execution result of the SELECT phrase in the INSERT SELECT statement 111 to one of the plurality of cores. Specifically, the allocation unit 402 allocates each record group to one of the plurality of cores by securing a buffer corresponding to the allocation destination core and storing the records in the buffer in a distributed manner. In the example of FIG. 4, the allocating unit 402 stores a record to be allocated to the core # 1 in the buffer b1, and stores a record to be allocated to the core # 2 in the buffer b2.

  Further, the assigning unit 402 may distribute and assign data groups to a small number of cores out of the number of partitions and the number of cores.

  The transmission unit 403 executes the storage process 113 corresponding to the INSERT phrase in the INSERT SELECT statement 111 as a process included in the transaction of the search process 112 to the allocation destination core to which each of the record groups of the plurality of cores is allocated. Send a request to

  The acquisition unit 404 acquires a record assigned to each core in response to each of the record groups that are the execution result of the SELECT phrase in the INSERT SELECT statement 111 being assigned to one of the plurality of cores. Specifically, the acquisition unit 404 acquires a record assigned to each core from a buffer corresponding to each core. In the example of FIG. 4, the acquisition unit 404 of the core # 1 acquires a record from the buffer b1, and the acquisition unit 404 of the core # 2 acquires a record from the buffer b2.

  The storage process execution unit 405 executes the storage process 113 as a process included in the transaction of the search process 112. In the example of FIG. 4, the storage process execution unit 405 of the core # 1 stores the acquired record in the partition pt1 or the partition pt2 as the storage process 113. An example of in which partition the acquired record is stored will be described with reference to FIG.

  Next, operation examples of the leader execution body ld and the member execution body mm will be described with reference to FIGS.

  FIG. 5 is an explanatory diagram (part 1) illustrating an example of an operation example of the leader execution body ld and the member execution body mm. FIG. 6 is an explanatory diagram (part 2) illustrating an example of an operation example of the leader execution body ld and the member execution body mm. 5 and 6, it is assumed that the number of partitions is 3, the number of cores of the CPU 301 is three or more, and three member execution bodies mm are generated.

  5 and 6, it is assumed that the database control apparatus 101 receives SQL1 shown in FIG. 5 from the client terminal 201. The contents of SQL1 are "INSERT INTO T4-SELECT-FROM (SELECT-FROM T1, T2) A LEFT OUTER JOIN-T3-". Here, SQL1 performs the SELECT clause 501 on the data obtained by externally joining the table T3 with the result obtained by performing the SELECT clause 502 of FIG. 5 on the tables T1 and T2 as an axis. SQL indicating that the result is stored in the table T4.

  The reader execution body ld scans from the partitions pt1_1, 1_2, and 1_3 storing the data of the table T1 as the SELECT processing of (ld-1) shown in FIG. 5, and generates records that match the conditions. At this time, the leader executor ld scans in parallel from the partitions pt1_1, 1_2, and 1_3. The leader executor ld sorts the generated data and stores it in the sorting result st1. The leader executor ld also performs the processing of (ld-2) SELECT shown in FIG. 5 in the same manner as the processing of (ld-1) SELECT shown in FIG. 5, and stores the execution result in the sort result st2. To do.

  Next, the leader execution body ld joins the result of T1 and the result of T2 as the process (ld-3) shown in FIG. The leader executor ld stores the result of the JOIN in the sort result st3. Then, the leader executor ld also performs the processing of (ld-4) SELECT shown in FIG. 5 in the same manner as the processing of (ld-1) SELECT shown in FIG. 5, and converts the execution result into the sort result st4. To store.

  Then, the leader execution body ld performs INSERT in parallel with T4 while JOINing the execution results of (ld-3) and (ld-4) as the processing of (ld-5) shown in FIG. Specifically, the leader executor ld stores the joined record in the work area wa_ld. The processing after storage is shown in FIG.

  In FIG. 6, the leader execution body ld accumulates the records stored in the work area wa_ld in the buffers b1, b2, and b3 for the three member execution bodies mm1, mm2, and mm3. Wait for a request. The buffers b1, b2, and b3 are prepared for the number of member execution bodies mm by the leader execution body ld.

  Specifically, the leader execution body ld first accumulates records in the buffer b1. The leader execution body ld accumulates records in the buffer b2 when there is no more free space in the buffer b1. Further, the leader executor ld accumulates records in the buffer b3 when there is no free area in the buffer b2, and accumulates records in the buffer b1 when there is no free area in the buffer b3. The leader executor ld repeats these processes until there are no more records stored in the work area wa_ld. In the example of FIG. 6, the leader executor ld stores records rd1 and 2 in the buffer b1, stores records rd3 and 4 in the buffer b2, and stores records rd5 and 6 in the buffer b3.

  Next, the operation of each member execution body mm will be described. Since the operations of the member execution bodies mm1 to mm3 are the same, the operation of the member execution body mm1 will be described. The member execution body mm1 requests the buffer in which the record is stored to the leader execution body ld as the processing of (mm-1) shown in FIG. 6, and acquires the record from the buffer b1 in the example of FIG. When the acquisition of the record ends, the member execution body mm1 notifies the leader execution body ld of the release of the buffer b1. The member execution body mm1 stores the acquired record in the work area wa_mm1. In the example of FIG. 6, the member execution body mm1 stores the records rd1 and rd2 obtained from the buffer b1 in the work area wa_mm1.

  Then, the member execution body mm1 stores the record stored in the work area wa_mm1 in any of the partitions pt4_1, 4_2, and 4_3 storing the data of the table T4 as the process (mm-2) illustrated in FIG. For example, the member execution body mm1 stores the record rd1 in the partition pt4_1 and stores the record rd2 in the partition pt4_2.

  Here, with respect to which partition to store, each partition distributes records to be stored based on the key of the record to be stored. For example, partition pt4_1 stores records whose key values are “1” to “3”, partition pt4_2 stores records whose key values are “4” to “6”, and partition pt4_3 stores key values. Stores records from “7” to “9”. The member execution body mm1 refers to the value of the key to be stored and determines a storage destination partition from among the partitions pt4_1, 4_2, and 4_3.

  Next, a storage example in the partition according to the present embodiment will be described with reference to FIG.

  FIG. 7 is an explanatory diagram showing a detailed example of storage in the same transaction. A plurality of member execution bodies mm are operated as the same transaction as the leader execution body ld, so that even if a plurality of member execution bodies mm store records in parallel, deadlock or exclusion between member execution bodies mm Waiting is suppressed.

  For example, as shown in FIG. 7, the member execution body mm1 stores records rd1, rd3, and rd5 in pages pg1, pg3, and pg2, respectively, as the processes (1), (3), and (5) shown in FIG. Store. The member execution body mm2 stores records rd2 and rd4 in pages pg2 and pg1, respectively, as the processes (2) and (4) shown in FIG. Here, since the member execution bodies mm1 and mm2 are included in the same transaction, they can be stored in parallel on the same page without performing exclusive control.

  Further, since the member execution bodies mm1 and mm2 are included in the same transaction, they are also stored as one transaction in the log file lg storing the operation history for the DB. In the example of FIG. 7, log information lg1 to lg5 is stored in the log file lg. For example, the log information lg1 indicates that the record rd1 is stored in the page pg1. The log information lg1 to lg5 is stored as one transaction. The database control apparatus 101 only needs to roll back one transaction with reference to the log file lg as in the case of not storing in parallel when rolling back.

  Next, a case where data is stored in different transactions will be described with reference to FIG.

  FIG. 8 is an explanatory diagram showing a detailed example of storage in different transactions. In FIG. 8, it is assumed that the member execution bodies mm1 and mm2 included in the database control apparatus are operated as a transaction different from the transaction of the leader execution body ld. In this case, when page exclusion is performed, the member execution bodies mm are mutually exclusive, and a deadlock state may occur. Further, a plurality of transactions for each member execution body mm are combined into one.

  For example, as shown in FIG. 8, the member execution body mm1 stores records rd1, rd3, and rd5 in pages pg1, pg3, and pg2, respectively, as the processes (1), (3), and (5) shown in FIG. Store. Further, the member execution body mm2 tries to store the records rd4 and rd2 in the pages pg2 and pg1, respectively, as the processes (2) and (4) shown in FIG. However, since the member execution bodies mm1 and 2 are included in different transactions, the member execution body mm2 stores the records rd2 and rd4 in the pages pg1 to pg1 in which the member execution body mm1 stores the records rd1, rd3, and rd5. I can't.

  Since the member execution bodies mm1 and mm2 are included in different transactions, log information lg1 to lg5 indicating the results of the member execution bodies mm1 and mm2 are also stored in the log file lg as different transactions. For example, in the example of FIG. 8, the log information lg1, lg3, lg5 and the log information lg2, lg4 are stored as different transactions. Therefore, the relationship between the member execution bodies mm is unknown. The commit log is also output separately for each transaction. Therefore, in this case, the database control apparatus performs processing for combining the transactions into one.

  Next, a flowchart illustrating processing performed when the INSERT SELECT statement described in FIGS. 5 to 7 is received will be described with reference to FIGS. 9 and 10.

  FIG. 9 is a flowchart (part 1) illustrating an example of a processing procedure when an INSERT SELECT statement is received. FIG. 10 is a flowchart (part 2) illustrating an example of the processing procedure at the time of receiving an INSERT SELECT statement. The INSERT SELECT statement process is a process executed by the database control apparatus 101 when an SQL that becomes an INSERT SELECT statement is received. Here, in the description of FIG. 10, since the processing contents of the member execution body mm are the same for each member execution body mm, the description will be given using the member execution body mm1, and the operation of the other member execution bodies mm will be described. Omitted.

  The leader executor ld starts a transaction (step S901). After starting the transaction, the leader executor ld starts processing of the SELECT phrase shown by (ld-1) to (ld-4) in FIG.

  Next, the leader execution body ld creates a member execution body mm (step S902). Also, the leader execution body ld acquires the created buffer b for member execution bodies (step S903). Next, the leader executor ld determines whether or not the member executors mm have been created for the number of partitions or the number of cores of the CPU 301 that is small (step S904). If the member execution bodies mm have not been created for the number of partitions or the number of cores of the CPU 301 that is small (step S904: No), the leader execution body ld proceeds to the processing of step S902.

  On the other hand, when the member execution body mm is created for the number of partitions or the number of cores of the CPU 301 that is small (step S904: Yes), the leader execution body ld enters the work area wa_ld as the execution result of (ld-1) in FIG. The stored records are accumulated in the member execution buffer b (step S1001). Next, the leader execution body ld transmits a processing request to each member execution body mm (step S1002).

  The member execution body mm1 waits for a notification from the leader execution body ld (step S1003). Next, the member execution body mm1 confirms the notification content from the leader execution body ld (step S1004). When the process end is received as the notification content from the leader execution body ld (step S1004: process end), the member execution body mm1 ends the INSERT SELECT statement reception process.

  When the processing request is received as the notification content from the leader execution body ld (step S1004: processing request), the member execution body mm1 acquires a record from the member execution body buffer b1 (step S1005). Next, the member execution body mm1 stores the acquired record in the corresponding partition (step S1006). Then, the member execution body mm1 determines whether or not there is a record in the buffer b1 of the member execution body (step S1007). If there is a record in the buffer b1 of the member execution body (step S1007: Yes), the member execution body mm1 proceeds to the process of step S1005.

  On the other hand, when there is no record in the member execution body buffer (step S1007: No), the member execution body mm1 notifies the leader execution body of the completion of processing (step S1008). Then, the member execution body mm1 proceeds to the process of step S1003.

  The leader executor ld that has received the processing completion determines whether or not there is a record in the work area wa_ld (step S1009). If there is a record in the work area wa_ld (step S1009: Yes), the leader executor ld proceeds to the process of step S1001. On the other hand, when there is no record in the work area wa_ld (step S1009: No), the leader executor ld ends the transaction (step S1010). Next, the leader executor ld notifies all member executors of the end of processing (step S1011). Then, the leader executor ld ends the INSERT SELECT statement reception process.

  By executing the INSERT SELECT statement reception process, the database control apparatus 101 can execute each member execution body mm in the same transaction, so that the waiting process due to exclusion is suppressed and the storage process is efficiently performed. It can be performed.

  As described above, for the INSERT SELECT statement, the leader executor ld causes the member executor mm to execute the INSERT phrase as the same transaction as the SELECT phrase processing. Thereby, since the member execution body mm is the same transaction, exclusion does not occur, and speeding up by parallel execution can be obtained.

  Further, the leader executor ld may distribute and assign the data group to a small number of cores out of the number of partitions and the number of cores. Thereby, the database control apparatus 101 can execute the INSERT SELECT statement with the maximum number that can be parallelized.

  Further, the reader executor ld executes a search process corresponding to the SELECT phrase as a process included in the same transaction as the storage process corresponding to the INSERT phrase in response to any core accepting the INSERT SELECT statement. May be. By operating in this way, the database control apparatus 101 can ensure that the search process corresponding to the SELECT phrase and the storage process corresponding to the INSERT phrase are the same transaction.

  In addition, any of the plurality of cores may perform the leader execution body ld. For example, in the database control apparatus 101, the core having the smallest load among the plurality of cores may execute the leader execution body ld. As a result, the database control apparatus 101 can achieve load distribution, for example, when a core with a small load executes the leader execution body ld.

  Further, the member execution body mm may cause each of a plurality of cores to execute the storage process corresponding to the INSERT phrase as the same transaction as the search process corresponding to the SELECT phrase. As a result, the member execution body mm does not cause deadlock or waiting for exclusion, so that the database control apparatus 101 can increase the speed.

  The database control method 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. The database control program is recorded on a computer-readable recording medium such as a hard disk, a flexible disk, a CD-ROM (Compact Disc-Read Only Memory), a DVD (Digital Versatile Disk), and is read from the recording medium by the computer. Executed by. The database control program may be distributed through a network such as the Internet.

  The following additional notes are disclosed with respect to the embodiment described above.

(Supplementary note 1)
Assigning each of the data groups generated based on the execution result of executing the search process for the database as the process included in the transaction to any of a plurality of processors accessible to the database;
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
A database control program for executing a process.

(Appendix 2) The database is divided into a plurality of storage areas,
The assigning process is:
The data group is distributed and allocated to a small number of processors among the number of the plurality of storage areas and the number of the plurality of processors, respectively.
The database control program according to supplementary note 1, characterized in that:

(Supplementary note 3) The database control program according to supplementary note 1 or 2, wherein one of the plurality of processors is caused to execute the assigning process and the transmitting process.

(Supplementary Note 4)
Executing the search process as a process included in the transaction in response to receiving an instruction indicating that a data group generated based on the execution result of the search process for the database is stored in the database; Let it run
The assigning process is:
Assigning each of the data groups generated based on the execution result of the executed search process to any of the plurality of processors;
The database control program according to supplementary note 3, characterized by:

(Supplementary Note 5) In a processor included in a plurality of processors that can access a database,
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
A database control program for executing a process.

(Supplementary note 6) A database control apparatus having a plurality of processors capable of accessing a database,
Each of the data groups generated based on the execution result of executing the search process for the database as a process included in a transaction is assigned to any of the plurality of processors,
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
A database control device characterized by a control unit.

(Supplementary note 7) A database control apparatus having a plurality of processors capable of accessing a database,
The processors included in the plurality of processors are:
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
A database control apparatus characterized by that.

(Supplementary note 8) A database control apparatus having a plurality of processors capable of accessing a database,
One of the plurality of processors is
Each of the data groups generated based on the execution result of executing the search process for the database as a process included in a transaction is assigned to any of the plurality of processors,
A request to execute storage processing for storing the data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
The processors included in the plurality of processors are:
In response to receiving the request from any of the processors, obtaining data assigned to the processor;
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
A database control apparatus characterized by that.

(Supplementary note 9)
Assign each of the data group generated based on the execution result of executing the search process for the database as a process included in the transaction to one of the multiple processors that can access the database,
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
The database control method characterized by performing a process.

(Supplementary Note 10) A processor included in a plurality of processors capable of accessing a database is
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
The database control method characterized by performing a process.

# 0 to # 2 Core 101 Database controller 102 DB
111 INSERT SELECT statement 401 Search processing execution unit 402 Allocation unit 403 Transmission unit 404 Acquisition unit 405 Storage processing execution unit

Claims (9)

On the computer,
Assigning each of the data groups generated based on the execution result of executing the search process for the database as the process included in the transaction to any of a plurality of processors accessible to the database;
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
A database control program for executing a process. The database is divided into a plurality of storage areas,
The assigning process is:
The data group is distributed and allocated to a small number of processors among the number of the plurality of storage areas and the number of the plurality of processors, respectively.
The database control program according to claim 1.   The database control program according to claim 1, wherein any one of the plurality of processors is caused to execute the assigning process and the transmitting process. In any of the processors,
Executing the search process as a process included in the transaction in response to receiving an instruction indicating that a data group generated based on the execution result of the search process for the database is stored in the database; Let it run
The assigning process is:
Assigning each of the data groups generated based on the execution result of the executed search process to any of the plurality of processors;
The database control program according to claim 3. In a processor included in multiple processors that can access the database,
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
A database control program for executing a process. A database controller having a plurality of processors capable of accessing a database,
Each of the data groups generated based on the execution result of executing the search process for the database as a process included in a transaction is assigned to any of the plurality of processors,
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
A database control device characterized by a control unit. A database controller having a plurality of processors capable of accessing a database,
The processors included in the plurality of processors are:
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
A database control apparatus characterized by that. Computer
Assigning each of the data groups generated based on the execution result of executing the search process for the database as the process included in the transaction to any of a plurality of processors accessible to the database;
A request for executing storage processing for storing data allocated to the allocation destination processor in the database as processing included in the transaction is transmitted to the allocation destination processor to which each of the data groups of the plurality of processors is allocated. ,
The database control method characterized by performing a process. Processors included in multiple processors that can access the database
In response to the fact that any data in the data group generated based on the execution result of executing the search process for the database as a process included in the transaction is assigned to the processor, the data is acquired,
A storage process for storing the acquired data in the database is executed as a process included in the transaction.
The database control method characterized by performing a process.

Images