JP3845535B2 - Database system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a database system, for example, a write-back method, which operates by matching data related to the contents of a plurality of databases.
[0002]
[Prior art]
Conventionally, when a plurality of databases exist and the same data is held between them, it is common to synchronize between tables using a replication function. FIG. 12 shows a processing procedure for synchronizing tables when two databases A and B exist. That is, the databases A and B are connected by the communication line 100. The two databases A and B are defined as replications that can be updated from both directions. Updates in the database A are propagated and reflected in the database B, updates in the database B are propagated and reflected in the database A, Both tables are synchronized.
[0003]
At this time, an operation for synchronization will be described. First, in the process indicated by (1) in FIG. 12, the table 003 of the database A is updated. At this time, a transaction as shown in FIG. 5 is used.
[0004]
Next, in the process indicated by (2) in FIG. 12, the transaction used for updating the table 003 of the database A is stored in the transaction queue set in the database A.
[0005]
On the other hand, in the process shown in (1), the table 002 of the database B is updated. FIG. 6 shows an example of the transaction at this time. Further, in the processing shown in (2), the transaction used for updating the table 002 of the database B is stored in a transaction queue set in the database B.
[0006]
As a result, there is a difference between the data updated in the respective databases in the tables 002 and 003 on the respective databases A and B. Therefore, it is necessary to propagate as follows. First, in the processes shown in (3) and (3), transactions stored in the transaction queue are executed against the opposing database at a certain timing to eliminate the data difference between the tables. .
[0007]
Transactions stored in the transaction queues of the databases A and B are generally executed at a timing corresponding to a request for referring to as accurate data as possible, or at a fixed time interval.
[0008]
[Problems to be solved by the invention]
However, the content of the transaction stored in the transaction queue is unknown as to which data in which table is updated unless it is executed. That is, when referring to a database, it is not known whether the data to be referred to is the latest or whether there is a difference from the latest data. Therefore, even if it waits for the propagation process to be completed while trying to refer to the latest data, the data may be propagated regardless of the data to be referenced. It is not efficient to do.
[0009]
For example, as shown in FIG. 13, the table 003 of the database A is updated in (1), and the transaction used for updating the table 003 of the database A in the process shown in (2) is Save in the transaction queue set in database A. At this time, when a reference request is made to the table 002 of the database B in the process shown in (3), the propagation of the transaction stored in the transaction queue set in the database A is shown in (4). Even if the reference shown in (5) is performed, the propagation process performed in (4) is an update to the table 003 of the database B, and is not an update to the table 002 of the database B to be referenced. Waiting for processing has been made.
[0010]
In addition, when transaction propagation is executed at regular intervals, there is a possibility that the currently referenced data may not be the latest data, although the waiting time is not required. There was a problem that it was unclear. In addition, if the same data in the same table in the databases A and B is rewritten to a different value at the same time, a propagation process conflict occurs, and one of the data needs to be discarded. A situation occurs in which the data is not updated.
[0011]
The present invention has been made as a solution to such problems, and its purpose is to always access the latest data, and when there is no difference in the data to be referenced, the data is transmitted without executing propagation. To provide an accessible database system. Another object of the present invention is to provide a database system that can appropriately ensure the identity of data among a plurality of databases without causing a competition of propagation processing.
[0012]
[Means for Solving the Problems]
The database system according to the present invention is a database system in which two or more stations are connected via a communication circuit. Each station is provided with a database including a plurality of tables each including a plurality of records, and for each record. When the data is updated in the database of the other station, the data in the database of the own station is updated based on the storage area in which the identification information of the other station that has changed the data of the record is stored and the transaction. Change means to be executed, and identification information control means for setting the identification information of the own station in the storage area provided in the corresponding record of the corresponding table with respect to the database of the other station when the change is made by the change means , A transaction queue in which the transaction used in the changing means is accumulated, and the local station Reference request or if there is a change request to database, or changes in other stations based refers to the storage area in the record corresponding to the range according to the request whether the identification information of the other station is set is made Detecting means for detecting whether or not, and when the detecting means detects that a change has been made in the other station, the relevant transaction is propagated from the transaction queue of the relevant other station and the database of the relevant other station is updated to the local station database. and applied at both the data match means for erasing the identification information of other stations that are set in the corresponding storage area, when a reference request to the database of the station, and no change by the detector As soon as the determination is made, the database of the local station is referred to and the reference result is returned to the reference request source. And a data return means for referring to the local database and returning the reference result to the reference request source after the reflection process by the data matching means when it is detected that the change has been made. .
[0013]
Further, in the database system according to the present invention, even when the detecting means receives a change request for the database of the own station, whether or not the change is made in the other station with reference to the storage area in the record corresponding to the range related to the request. Even if it is detected by the detecting means that the change has been made in the other station, the data matching means propagates the relevant transaction from the transaction queue of the relevant other station and stores the data in the database of the relevant other station. The update is reflected in the own station database, and when there is a change request for the own station database, the changing means updates the data in the own station database as soon as it is determined that there is no change by the detecting means. On the other hand, when the detection means detects that a change has been made in another station, the data matching means And wherein the updating the data in the database of the local station after the reflection processing.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of a database system according to the present invention will be described below with reference to the accompanying drawings. In the drawings, the same components are denoted by the same reference numerals and redundant description is omitted. FIG. 1 shows an explanatory diagram of an embodiment of a database system according to the present invention. In this system, a first station shown on the upper side of the figure and a second station shown on the lower side of the figure are coupled via a communication line 100.
[0016]
FIG. 2 shows the system configuration of the two stations. As shown in FIG. 2, each station is constructed as a workstation, personal computer, or other computer. That is, the computer shown in FIG. 2 has a CPU 51 that performs overall control of the apparatus, and a main storage device 52 that stores information such as programs and data used by the CPU 51 is connected to the CPU 51. Furthermore, an input device control unit 54, a communication interface 55, and a storage device control unit 56 are connected to the CPU 51 via a system bus 53. The keyboard controller 54 is connected to a keyboard input device capable of inputting various information and an input device 57 such as a mouse, magnetic card or IC card reader as a pointing device, and communicates via the communication line 100 to the communication interface 55. A communication processing unit 58 is connected to the storage device control unit 56, and a storage device 59 such as a magnetic disk device, floppy disk drive, or MO (magneto-optical disk) drive, which is an auxiliary storage device, is connected to the storage device control unit 56. A display device and its control unit are provided as necessary.
[0017]
The storage device 59 functions as the databases A and B, and is provided with tables indicated by “001”, “002”, and “003”. Here, there are three tables, but the number is naturally not limited to this. Each table is divided into several records, and each record is configured as shown in FIG. A ROWID, which is a row number managed in the database, is set at the top, and when data is changed in the database of another station, a DartyBit for writing the database name (identification information) is prepared. When data changes occur in a plurality of other station databases, their database names are written in DartyBit.
[0018]
Following the above-mentioned DartyBit, a KEY area in which a primary key item is set is provided in a data area released to the user, and after this KEY area, a data area COLUMN001 to 003 (003 in the figure) opened to the user. However, the present invention is not limited to this.). In the above description, data change is a concept including updating existing data and inserting (adding) new data. Therefore, in the case of data update, data is already stored in the data area COLUMN001, but in the case of insertion of new data, no significant data is yet stored in the data area COLUMN001. As shown in FIG. 4, “NULL” is set.
[0019]
As shown in FIG. 1, in order to match the contents of the databases A and B, a queue 10 (10-A, 10-B) that accumulates transactions for changing data in other databases is a main memory. Provided in the device 52. The queue 10-A is a queue that is stored in order to propagate the transaction to the partner database when an update transaction or an insertion transaction occurs in the database A with respect to the tables 002, 003, etc., and the queue 10-B In the database B, when an update transaction or an insertion transaction occurs for the table 002, 003, etc., the queue is stored to propagate the transaction to the partner database. The transactions set here are as shown in FIGS. FIG. 5 shows an example of a transaction used for data update on the table 003, FIG. 6 shows an example of a transaction used for data update on the table 002, and FIG. 7 shows data insertion on the table 003. An example of a transaction used in is shown.
[0020]
The main storage device 52 of each station having the databases A and B is provided with a program corresponding to the flowchart shown in FIG. 8, and this program is executed to match the data in the table 003 in the databases A and B. This will be explained. This example shows a case where a search has occurred for data in the table 003 before a transaction stored in the queue 10-A is propagated.
[0021]
In the process indicated by (1) in FIG. 1, data is updated to the table 003 in the database A (S1 (change means) in FIG. 8). Next, in the process indicated by (2) in FIG. 1, the database name “A” is written in the DartyBit area for the updated record in the table 003 of the database B (S2 in FIG. 8 (identification information control). means)). In the corresponding record at this time, as shown in FIG. 3, “A”, which is the updated database name, is set in the DartyBit area. Next, in the process indicated by (3) in FIG. 1, the transaction used for the update is stored in the transaction queue 10-A set in the database A (S3 in FIG. 8 (data matching means by change control means). Request set to)).
[0022]
Next, in the process indicated by (◯ 4) in FIG. 1, it is assumed that a reference transaction occurs for the table 003 in the database B (S4 in FIG. 8). In this case, the DartyBit area of the reference range record is searched to detect whether or not the updated database name exists (S5 (detection means) in FIG. 8). If the updated database name exists, this data needs to be updated by propagation. Therefore, as shown in FIG. 1 (○ 5), the transaction related to the contents of the transaction queue 10-A of the database A is processed. Propagate (S6 (data matching means) in FIG. 8). Thereby, the data update in the database A is reflected in the database B. Next, the corresponding database name in the DartyBit area is erased by reflecting the data according to the above propagation (S7 (identification information control means) in FIG. 8), and in the process shown in (◯ 6) in FIG. The latest data reflected in step S8 is returned as a reference result (S8 (data return means) in FIG. 8).
[0023]
FIG. 9 shows processing when data is updated before data stored in the queue 10-A is propagated. Corresponding to this case, the main storage device 52 of each station having the databases A and B is provided with a program corresponding to the flowchart shown in FIG. Since the data in the table 003 is matched, this will be described.
[0024]
In the process indicated by (1) in FIG. 9, data is updated to the table 003 in the database A (S11 (changing means) in FIG. 10). Next, in the process indicated by (2) in FIG. 9, the database name “A” is written in the DartyBit area for the updated record in the table 003 of the database B (S12 (identification information control in FIG. 10). means)). In the corresponding record at this time, as shown in FIG. 3, “A”, which is the updated database name, is set in the DartyBit area. Next, in the process indicated by (3) in FIG. 9, the transaction used for the update is stored in the transaction queue 10-A set in the database A (S13 in FIG. 10 (data matching means by the change control means). Request set to)).
[0025]
Next, in the process indicated by (◯ 4) in FIG. 9, it is assumed that an update transaction has occurred for the table 003 of the database B (S14 in FIG. 10). In this case, the DartyBit area of the record in the update range is searched to detect whether or not the updated database name exists (S15 (detection means) in FIG. 10). If there is an updated database name here, this data needs to be updated by propagation. Therefore, as shown in (○ 5) of FIG. 9, an update transaction to the table 003 of the database B is waited. (Change control means).
[0026]
During this time, as shown in (6) of FIG. 9, the transaction related to the contents of the transaction queue 10-A of the database A is propagated (S16 (data matching means) of FIG. 10). Thereby, the data update in the database A is reflected in the database B. Next, the corresponding database name in the DartyBit area is deleted by reflecting the data according to the propagation (S17 (identification information control means) in FIG. 10).
[0027]
As a result of the propagation, the corresponding database name in the DartyBit area is deleted, and in the process shown in (7) of FIG. 9, the update to the table 003 of the database B that was kept waiting in (5) of FIG. A transaction is executed (S18 in FIG. 10 (change means, change control means)). Next, in the process indicated by (8) in FIG. 9, the database name “B” is written in the DartyBit area for the updated record in the table 003 of the database B (S19 (identification information control means) in FIG. 10). ). Next, in the process indicated by (9) in FIG. 9, the transaction used for the update is stored in the transaction queue 10-B set in the database B (S20 (change control means) in FIG. 10).
[0028]
Note that the database data change includes insertion (in the case where the transaction is INSERT) in addition to the update as described above. FIG. 7 shows a transaction in which the transaction is an INSERT statement. As described above, when the transaction is an INSERT statement, there may be a case where there is no record in the partner database where DartyBit should be set. In such a case, a record in which only the primary key value is set in the KEY area is inserted into the counterpart database, and the database name related to the transaction is set in the DartyBit area. In the thus inserted record, the database name “A” is set in the DartyBit area as shown in FIG.
[0029]
As shown in the above embodiment, if there is no changed database name in the DartyBit of the reference range (record) of the database, it is guaranteed that the data to be referred to is the latest, and step S5 in FIG. Since the process branches to NO in step S15 in FIG. 10, the latest data is accessed without performing unnecessary propagation processing before the reference. On the other hand, when there is a changed database name in the DartyBit of the reference range (record) of the database, it is referenced after the propagation process is executed in front of the reference and the table is synchronized. , Always ensure access to the latest data. Furthermore, in the case of a database change, if there is a database name that has been changed in the DartyBit of the change range (record), it is referenced after the propagation process is executed in advance of the change and the table is synchronized. Therefore, in the case of a conflict, processing in an appropriate order is ensured, and the latest data is always changed.
[0030]
The present invention can also be applied when replication is defined for a database of three or more stations. FIG. 11 shows a replication configuration of databases A, B, and C of three stations according to the present invention. In this example, the data update is performed on the table 003 of the database A in the process indicated by (1) in FIG. Next, in the process indicated by (2) in the figure, the database name “A” is written to DartyBit for the record updated above in the table 003 of the database B. At the same time, in the process indicated by (3) in FIG. 11, the database name “A” is written into DartyBit for the updated record in the table 003 of the database C as well.
[0031]
Further, in the process indicated by (4) in FIG. 11, the transaction used for the update is stored in the transaction queue 10-A set in the database A. In the above, since the identification information set in DartyBit includes the database name (site name) so that it can be determined in which database the update has occurred, when accessing the table 003 in the databases B and C, the DartyBit area In the case of a record in which identification information is set, propagation can be executed only from a transaction queue of a database in which a transaction corresponding to the corresponding database name (site name) exists, which is efficient.
[0032]
【The invention's effect】
As described above, according to the database system of the present invention, when there is a reference request to the database of the local station, the identification indicating whether or not the change is made in the other station provided in the database of the other station Based on the information, it is detected whether or not the change has been made at the other station. If the change has been made at the other station, the data is matched, the database is referenced, and the reference result is requested. Since the process of returning to the original is performed, there is an effect of ensuring the reference of the latest data.
[0033]
Further, according to the database system of the present invention, upon receiving a data change request relating to the contents of the own station database, based on the identification information, it is detected whether or not the other station has changed, and the other station If there is a change, the data is changed after matching the changed data at the other station, so in the case of a conflict, processing in an appropriate order is ensured and the latest data is always updated. There is an effect that changes are made.
[Brief description of the drawings]
FIG. 1 is a diagram showing a replication operation when data is referenced by a database system according to the present invention.
FIG. 2 is a block diagram showing a specific configuration example of one station in the database system according to the present invention.
FIG. 3 is a diagram showing a record configuration of a database in the database system according to the present invention.
FIG. 4 is a view showing record contents of a database corresponding to an insertion transaction in the database system according to the present invention.
FIG. 5 is a view showing an example of an update transaction of a table 003 in the database system according to the present invention.
FIG. 6 is a view showing an example of an update transaction for a table 002 in the database system according to the present invention.
FIG. 7 is a view showing an example of an insertion transaction of a table 003 in the database system according to the present invention.
FIG. 8 is a flowchart for explaining processing in the case of a reference transaction in the database system according to the present invention.
FIG. 9 is a diagram showing a replication operation when data is updated by the database system according to the present invention.
FIG. 10 is a flowchart for explaining processing in the case of an update transaction in the database system according to the present invention.
FIG. 11 is a diagram showing a replication operation in a base system configured by three databases according to the present invention.
FIG. 12 is a diagram showing a replication operation by a database system according to a conventional example.
FIG. 13 is a diagram showing a waste of processing in a replication operation by a database system according to a conventional example.
[Explanation of symbols]
A, B, C Database 10-A, 10-B, 10-C Transaction queue 51 CPU
52 Main Storage Device 54 Input Device Control Unit 55 Communication Interface 56 Storage Device Control Unit 57 Input Device 58 Communication Processing Unit 59 Storage Device 100 Communication Line

Claims (2)

In a database system in which two or more stations are coupled via a communication circuit,
Each station
A database comprising a plurality of tables composed of a plurality of records;
A storage area that is provided for each record and stores identification information of another station that has changed the data of the record when data update is performed in the database of the other station;
Based on the transaction, a changing means for performing data update in the local database,
An identification information control means for setting the identification information of the own station in the storage area provided in the corresponding record of the corresponding table with respect to the database of the other station when a change is made by the changing means;
A transaction queue in which transactions used in the changing means are accumulated;
When there is a reference request or a change request for the database of the own station, the change is made at the other station based on whether the identification information of the other station is set with reference to the storage area in the record corresponding to the range related to the request. Detecting means for detecting whether or not
When detecting that the detector is made of modifications in another station, both when to reflect the update of the database of the corresponding other station propagates the corresponding transaction from the transaction queue corresponding other station to own station database, the corresponding storage area Data matching means for erasing the identification information of other stations set in
When there is a reference request for the database of the own station, the detection unit immediately refers to the database of the local station and returns the reference result to the reference request source when it is determined that there is no change. A data return means for referring to the database of the local station after the reflection processing by the data matching means and detecting the reference result to the reference request source when it is detected that the change has been made in the other station. A database system. If there is a change request for your database,
The detection means detects whether or not a change has been made in another station with reference to a storage area of a record corresponding to the range related to the request,
When the data matching means detects that the detecting means has changed in another station, the data matching means propagates the relevant transaction from the transaction queue of the relevant other station and reflects the update of the database of the relevant other station in the local station database. ,
The change means updates the data in its own database as soon as it is determined by the detection means that there is no change. On the other hand, if the detection means detects that a change has been made in another station, the data 2. The database system according to claim 1, wherein the data in the local database is updated after the reflection processing by the matching means.

Images