JPH04165437A - Data base synchronization system - Google Patents

Abstract

PURPOSE:To control the synchronism on a data base system under multiple control in master-slave relation by an easy method by providing a revision number to each of records as update units of data bases. CONSTITUTION:The main data base 1 of a center station and the slave data bases 2 - 4 of respective local stations are connected by communication lines, and each of the files in the data bases 1 - 4 consists of 1st - (M)th records, each of which has 1st - (N)th fields. In the main data base 1, successive numbers are set first as revision numbers in the (N)th field in the registration order of the records. When a record is updated thereafter, the number obtained by adding 1 to the maximum value among the revision numbers which are already set is set in the (N)th field of a new record after the update, the main data base 1 is updated, and the new record after the update is broadcasted to all the local stations. Consequently, the contents of the master and slave data bases are made coincident by the simple method.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a database synchronization method, and particularly to a database synchronization method for a database system that holds the same data redundantly in a main database and a slave database connected by a communication line. in between.

[Conventional technology]

Conventionally, in database systems that hold the same data redundantly in master and slave databases, in order to match the contents of the slave database with the contents of the master database, the local station holding the slave database has to be restarted after a communication failure or suspension of operation. When doing so, the following two methods are used.

The first method is to partially lock (freeze updates) the main database of the center station, search all its contents and transfer them to the local station, reinitialize the contents of the slave database, and then・This is a method to release the tsukku and start operation.

The second method is to keep all the history in both the center station and the local station (using a journal file), and when the local station restarts, it uses the pre-update image of its own journal file to keep track. This method involves returning the contents of the database to the checkpoint and then updating the contents of the slave database to match the main database using the updated image of the journal file on the center side.

[Problem to be solved by the invention]

Among the conventional database synchronization methods described above, the first method locks the main database, so the main database cannot be updated at all times. When referring to the main database for synchronization management, data buffering and other complicated controls are required, which has the drawback of increasing the scale and cost of the device. or,
When a local station returns to operation from a suspended state, it is necessary to pull out all the contents of the main database and initialize the contents of the secondary database, and there is also the problem that a large amount of traffic is generated for this communication. .

On the other hand, the second method of using journal files requires that multiple stations maintain a large number of journal files, and also manages the synchronization of checkpoints for collecting pre-update images between the center station and local stations. Similar to the first method, this method has the drawback of increasing the scale and complexity of the device.

An object of the present invention is to eliminate the above-mentioned drawbacks and provide a database synchronization method that can manage synchronization in a simple manner in a multi-managed database system with a master-slave relationship.

[Means to solve the problem]

The database synchronization method of the present invention is a database system in which two primary databases and a secondary database, which is a copy of all or part of the primary database, are connected via a communication line, and the contents of the secondary database are always made to match those of the primary database. In the database synchronization method of greater than (or less than) the highest (or lowest) revision number in its slave database when restoring synchronization
The system is configured to retrieve a record with a revision number from the main database and update a record with a corresponding key.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a file configuration diagram of an embodiment of the present invention, and FIG. 2 is a block diagram showing an example of the configuration of a database system.

As shown in Fig. 1, one file of the database consists of M records, numbered 1st to Mth.
~Nth field has N fields. Of the N fields, the first and second fields are used as a key (information that uniquely identifies each record) and secondary keys for searching the database, and the Nth field contains the revision number. field is set.

The main database 1 of the center station and the slave databases 2 to 4 of each local station are connected by communication lines as shown in Fig. 2, and the slave database 2.3 stores all copies of the master database 1. 4 has a partial copy, and each local station must always match the contents of the slave databases 2 to 4 with the main database.

In the main database 1, first set a consecutive number as a revision number (1 to km if there are as many files as shown in Figure 1) in the Nth field in the order in which records are registered so that each record can be uniquely identified. After that, when the record is updated, the maximum revision number (kM) that is already set in the Nth field of the new record will be updated.
1 is added to the number, the main database 1 is updated, and the updated new record is broadcast (broadcast notification) to all local stations.

When each local station receives the update information (new record) sent from the center station, it updates its key (first and second record).
record) and replaces the record with the same key in the slave database of its own station with the new record sent, thereby updating the slave database.

When adding a new record, the center station assigns a new revision number and sends it, just like when updating, and each local station refers to the key and if the corresponding record does not exist, it is determined to be an addition and is registered unconditionally. shall be taken as a thing. However, in the case of the slave database 4 that holds a partial copy, the necessity is determined using a secondary key and the like before registration.

Next, when the local station resumes operation from a suspended state and it is necessary to synchronize the contents of the slave database and the main database, the revision numbers of all records in the slave database of the local station are updated to the latest values. (maximum value) and retrieve update information by searching the main database for records with revision numbers higher than that1.
In the second case, on the slave database side, there is no need to distinguish between the broadcast update information issued by the main database and the search result information from the center station based on a request from the local station as different types of information, and they are processed in exactly the same way. All you have to do is update the secondary database.

The above steps are shown along the time axis, and 7 is 31';
It is. In FIG. 3, the slave database normally receives update information broadcast from the main database up to revision number R=n, and at this point it enters a suspension state for some reason, with revision number R=n+
This shows a case where update information for m records from 1 to R=n+m is not reflected. Afterwards, when the cause of the suspension is removed and restarted, the secondary database retrieves the latest revision number R=n in its own database, requests a search of the main database under the condition of revision number Ran, and sets the revision number R≧n+1. Record group (R'-n+
1. R=n+2. -R=n+m> is retransmitted, and by updating the contents of its own database with this, it matches the main database. If the same record is updated multiple times while the secondary database is out of service, only the latest updated information will necessarily be searched because the old updated record will be re-updated with new updated information.

In the above embodiment, the revision number given to the updated record is assumed to be incremented by 1 for each update, but the same effect can be obtained by making it increase monotonically even if it is not incremented by 1. Obtainable. Furthermore, the same effect can be obtained by sequentially subtracting 1 from the minimum existing revision number instead of adding. Furthermore, although it has been explained that in the initial state, each record in the main database is assigned a different consecutive number, it is not necessary to assign consecutive numbers, and the same number (for example, O) may be assigned to all records.

〔Effect of the invention〕

As described in detail above, the present invention provides a revision number in one field of a record in a database system in which the same data is held in multiple databases connected by communication lines, thereby making it possible to update the main database. There is no need to lock, transmit and initialize all contents, and there is no need for each database to have a large number of journal files, and the effect is that the contents of the master and slave databases can be made consistent with a simple method. be.

[Brief explanation of drawings]

FIG. 1 is a file configuration diagram of an embodiment of the present invention, FIG. 2 is a block diagram of the configuration of a database system, and FIG. 3 is an explanatory diagram of a communication flow between a main database and a slave database. 1... Main database, 2, 3.4...
・Slave database.

Claims (1)

[Claims] 1. In a database system in which a main database and a sub-database consisting of a copy of all or part of the main database are connected via a communication line, the contents of the sub-database are always made to match the main database. In the database synchronization method of When restoring database synchronization, the main database is searched for a record with a revision number greater (or smaller) than the largest (or smallest) revision number in the slave database, and the record with the corresponding key is updated. Characteristic database synchronization method. 2. The database synchronization method according to claim 1, wherein a revision number obtained by adding 1 to the maximum value among all existing revision numbers is assigned to the new record to be updated. 3. The database synchronization method according to claim 1, wherein the new record to be updated is given a revision number obtained by subtracting 1 from the minimum value among all existing revision numbers.