KR20020090711A - Database diagnostic method of DBMS - Google Patents

Abstract

PURPOSE: A method for diagnosing a database of a DBMS(Database Management System) is provided to prevent a fatal error in an operation of a switchboard by diagnosing the disorder of a disk, reporting the particulars of an error, and restoring the database of the disk. CONSTITUTION: The performance of a diagnostic function for the database is decided(S10). If the time to diagnose the database arrives or a command for performing the diagnostic function is inputted, one database file for performing the diagnostic function is opened from the disk of a master processor(S12). The master processor requests a processor having the database the same as the opened database to perform the diagnosis(S14). If the checksum calculated by the processor is different from the checksum calculated by the master processor, a request for the restoration is transmitted to the corresponding processor(S24). The master processor receiving the details of a backup relation from the corresponding processor restores the data by recording the details of the backup relation in the corresponding database(S28).

Description

Database diagnostic method of DBMS

The present invention relates to a method for diagnosing a database of a database management system (DBMS), and more particularly, to a method for diagnosing a database of a DBMS, which enables a diagnosis and recovery of a failure by inspecting a database.

In general, the exchange system is a distributed structure composed of several processes, and DBMS is used to perform comprehensive management such as efficient support of exchange application data, data consistency, data backup, and data diagnosis and recovery. use.

Databases required by applications are created on the host system (offline) and loaded on each processor. When the applications want to search or modify these databases, they are made through the DBMS.

The database is loaded into the main memory of each processor for real-time processing, and one master processor among the processors of the exchange has a disk. The master processor is prepared in case the processor fails and is restarted. It will also place the database on disk.

When the exchange's applications request a change to the database through the DBMS, the DBMS changes the database in the requested processor's main memory and reflects the changes in the database on the master processor's disk.

1 is a diagram illustrating the structure of a DBMS in an exchange, in which a database must exist in each processor's main memory and a disk. A database in the disk dumps a new database to a disk during operation of the exchange. It can be broken if the database on the main memory and the version of the database on the disk are inconsistent, and if the disk is in a bad state or if the exchange's operating system malfunctions.

As described above, when a database is broken, a function of diagnosing an error by examining a broken database is not provided in the related art. Therefore, the operator periodically checks the database for a case where the database is broken. If the backup is stored in a separate storage medium such as (Cartridge Tape), and the database is broken, the backup database is reloaded to the exchange and restored.

As described above, conventionally, since there is no function for diagnosing whether the database is broken during operation of the exchange, it is difficult to determine whether the database is actually broken, and the system malfunctions after reloading the database into the main memory. After discovering that, you know that the database on disk is broken, after a fatal error has already occurred during the exchange.

As described above, since the processors of the exchange must be reloaded in order to recover the broken database, the interruption of the exchange is inevitable, and the reloading of the backed up database reloads the changes of the database that occurred after the last backup. There is a problem that is lost.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems. The present invention compares the contents of a database of a main memory with a database existing on a disk periodically or at the request of an operator to diagnose whether there is a failure in a database existing on a disk, It provides a DBMS database diagnostic method that prevents catastrophic errors in the exchange operation that may occur due to a database error by reporting the error that occurred to the operator and recovering the database existing on the disk. There is this.

1 is a diagram showing the structure of a DBMS in an exchange.

2 is a diagram showing a database structure of a DBMS applied to the present invention.

3 is a flowchart for explaining a database diagnostic method of the DBMS according to the present invention.

The database diagnosis method of the DBMS according to the present invention for achieving the above object comprises the steps of opening a database file for the master processor to perform the diagnostic function when it is necessary to perform a diagnostic function for the database; Requesting a database diagnosis to a processor in which a same database as the open database is stored in the main memory device; Retrieving a backup relation sequentially from a directory / dictionary area of a database stored in the main memory according to the diagnosis request, and receiving the checksum from the processor that calculates a checksum for the contents of the retrieved backup relations; ; Retrieving backup relations sequentially from the directory / dictionary area of the opened database and calculating a checksum for the contents of the retrieved backup relations; It is determined whether the calculated checksum and the checksum received from the processor match each other. If the calculated checksum and the checksum received from the processor do not coincide with each other, the processor restores the backup relation to the backup relation. Requesting process; And restoring data by recording the contents of the backup relation received from the processor that reads and transmits the contents of the backup relation in a predetermined size in response to the request for the restoration of the backup relation in the open database.

Here, the diagnostic function for the database may be performed according to the arrival of a time for performing a diagnosis on a database preset from an operator through an operator terminal and an input of a command for performing a diagnostic function on the database through an operator terminal. It features.

The method may further include repeating the request for diagnosing the database and restoring the data by the number of backup relations of the database.

Hereinafter, a method for diagnosing a database of a DBMS according to a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Figure 2 is a road showing the database structure of the DBMS to be applied to the present invention, the database used in the DBMS is largely divided into two areas, namely directory / dictionary area (a) and relation area (b) Are distinguished.

The relation area (b) described above is a set of relations used in a DBMS, and a relation is a table in which data defined by a user is collected.

As described above, relations are divided into two kinds of relations, such as backup relations and memory-only relations. When a backup relation is changed in main memory, The relations that must be backed up to the disk's database are also referred to as simple relations.

That is, backup relations are relations that should not be lost in any situation of the system, and simple memory relations are relations that should be loaded with data produced at the initial restart of the processor.

The directory / dictionary area (a) contains information about each relation in order for the DBMS to access the relations such as location information of each relation or information on whether each relation is a backup relation or a simple storage relation.

3 is a flowchart for explaining a database diagnostic method of the DBMS according to the present invention.

First of all, it is time to perform the diagnosis on the database preset from the operator through the operator terminal or determine whether to perform the diagnosis function on the database according to the command input for performing the diagnosis function on the database through the operator terminal. (S10).

As a result of the determination in step S10, when the time for performing the diagnosis on the database arrives or when the diagnosis function for the database is to be performed according to a command input for performing the diagnosis function on the database through the operator terminal, the master processor is the master. One database file for performing a diagnostic function is opened on the disk of the processor (S12).

After that, the master processor sends a diagnostic request to a processor in which the same database as the open database is stored in the main memory in order to check whether the contents of the database opened in the disk and the database stored in the main memory of the processor match each other. (S14). For example, if the database opened from disk is a database stored in the main memory of A processor, the A processor makes a request for diagnosis of the database.

As described above, the processor receiving the diagnostic request from the master processor sequentially searches the backup relation in the directory / dictionary area of the database, and calculates a checksum of the contents of the found backup relation (S16).

Then, the checksum calculated in step S16 is transmitted to the master processor (S18).

On the other hand, the master processor also searches the backup relations sequentially in the directory / dictionary area of the open database, similar to the processor requested to diagnose the diagnosis, and calculates a checksum for the contents of the retrieved backup relations (S20).

Thereafter, the master processor determines whether the check sum calculated in step S20 and the check sum received from the processor coincide with each other (S22).

As a result of the determination of step S22, if the check sum calculated in step S20 and the check sum received from the processor coincide with each other, the process proceeds to step S14 to check the next backup relation and performs database diagnosis with the processor. If the check sum calculated in step S20 and the check sum received from the processor do not match each other, the master processor requests the processor to restore the backup replication (S24).

As described above, the processor, which is requested to recover the backup relation from the master processor, may store the main memory in a block unit of a predetermined size, for example, a block unit of 512 bytes in order to minimize the number of communication between the processors and input / output to the disk. The contents are read from the corresponding backup relation of the database stored in the device and transmitted to the master processor (S26).

Meanwhile, the master processor receiving the contents of the backup relation from the processor recovers the data by writing the contents of the received backup relation to the corresponding database on the disk (S28).

If the size of the backup relation is larger than that of one block (512 bytes), steps S26 and S28 are repeated as much as the size of the relation.

Thereafter, it is determined whether the recovery of the backup relation has ended, and when the recovery of the backup relation has ended, the diagnosis and recovery of all the backup relations existing in the database opened in step S12 are performed to recover the database. It is determined whether has ended (S30, S32).

As a result of the determination of step S32, when the recovery of the database opened in step S12 is not terminated, the process proceeds to step S14 to request a database diagnosis from the processor to check the remaining backup relations, and the step S12. When the recovery of the database opened in step S ends, the master processor closes the database file opened in step S12 from the disk (S34).

Thereafter, the above steps S12 to S34 are repeated to diagnose and recover all databases existing on the disk (S36).

The database diagnosis method of the DBMS of the present invention is not limited to the above-described embodiments, and may be variously modified and implemented within the range allowed by the technical idea of the present invention.

According to the database diagnosis method of the DBMS of the present invention as described above, by comparing the contents of the database of the main memory and the database existing on the disk periodically or at the request of the operator to diagnose whether there is a failure in the database on the disk In the event of an error, by reporting the error to the operator and recovering the database existing on the disk, it is possible to prevent a catastrophic error in the operation of the exchange that may occur due to a database error.

Claims (3)

In the case where it is necessary to perform a diagnostic function for the database, opening a single database file for performing a diagnostic function by the master processor; Requesting a database diagnosis to a processor in which a same database as the open database is stored in the main memory device; Retrieving a backup relation sequentially from a directory / dictionary area of a database stored in the main memory according to the diagnosis request, and receiving the checksum from the processor that calculates a checksum for the contents of the retrieved backup relations; ; Retrieving backup relations sequentially from the directory / dictionary area of the opened database and calculating a checksum for the contents of the retrieved backup relations; It is determined whether the calculated checksum and the checksum received from the processor match each other. If the calculated checksum and the checksum received from the processor do not coincide with each other, the processor restores the backup relation to the backup relation. Requesting process; And recovering data by recording the contents of the backup relation received from the processor that reads and transmits the contents of the backup relation to a predetermined size in response to the request for the recovery of the backup relation in the open database. Database diagnostic method. The method of claim 1, wherein the diagnostic function for the database, A method of diagnosing a database of a DB, characterized in that it is performed according to the arrival of a time for performing a diagnosis on a database preset from an operator through an operator terminal and a command input for performing a diagnostic function on a database through an operator terminal. The method of claim 1, further comprising repeating the request for diagnosing the database and restoring the data as many times as the number of backup relations in the database.