JPH0962557A - Distributed database system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the data transfer volume and to imposse no load on a communication line neither a database processor of the other party by repairing only required data to be accessed at the time of issue of an inquiry request. SOLUTION: It is assumed that a data base DB1 of a database processor 10-1 has a part to be repaired in a partial item K. The database processor 10-1 is started with this part left unrepaired. Thereafter, inquiry requests related to items other than the part to be repaired are normally accepted, but the part to be repaired is repaired when an inquiry request including this part is issued. That is, the processor 10-1 receives data presentation for a proper access timing of the pertinent part from other database processors 10-2 to 10-4 and not only responds to inquiries but also repairs data of this part.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed database system for controlling a plurality of database processing devices each storing a database in which some items are duplicated and stored in a distributed manner.

[0002]

2. Description of the Related Art When the content or structure of a database becomes complicated, the database may be distributed and managed by a plurality of database processing devices that take in only the necessary items of the database. In such a system, each database processing device manages an arbitrary item in the data stored with another database processing device such that they overlap at least partially. Therefore, for example, when there is a request to update or delete an item of a certain data, all databases including the item in the data are targets of the deletion or update. That is, when such an inquiry request is made to one of the database processing devices, the database processing device updates or deletes the corresponding item of its own data and accesses the other database processing device to make the corresponding request. Modify or delete parts at the same time. For such management, each database processing device
A storage device is provided for displaying the database configuration and the like of all database processing devices.

Further, when a certain database processing device temporarily suspends the processing due to some trouble or the like, access for update or deletion from another database processing device cannot be accepted. Therefore, when recovering from a failure, it is necessary to collectively repair the items of data that are redundantly held in each database processing device. For such a purpose, a device that performs availability determination processing in which each database processing device distinguishably displays items of data that can be used and items of data that cannot be used in the databases of all other database processing devices Is provided (Japanese Patent Laid-Open No. 5-108453).

[0004]

The conventional distributed data system as described above has the following problems to be solved. In a conventional system, a database processing device that has recovered from a failure requests the transfer of the relevant part to ensure consistency when there is an item of data that duplicates the item of each data of another database processing device. Copy and repair your own database.

However, when copying data from such another database, a large amount of data is transferred at once. Therefore, the communication line connecting the database processing devices, which serves as a data transfer path, is exclusively used for this transfer process for a long time. Further, the database processing device on the data providing side is also occupied for a long period of time. Therefore,
There is a problem in that the overhead of the communication line and its database processing device becomes large, and normal inquiry processing and the like are hindered.

[0006]

The present invention adopts the following constitution in order to solve the above points. (Structure 1) The distributed data system of the present invention comprises a plurality of database processing devices in which arbitrary items in predetermined data are stored in a distributed manner by overlapping at least some of them.
All or some of the database processing devices include an inquiring device that processes a database usage request, configuration information indicating the data configuration of each database processing device, and availability information indicating whether or not each item of the data is available. There is a request to use the directory storage device that stores and the database storage device that acquires the latest information from another database and updates the availability information when the database processing device is restarted after being suspended. At this time, the availability information of the corresponding item in the data stored in its own database processing device is referred to, and if it is not available, the data of the access unit amount including the corresponding item is accepted from another database processing device, and the corresponding Failure to repair items and modify applicable availability information for directory storage It provided with a recovery device.

(Explanation) All or part of the database processing device means that this structure may be adopted only in a part of the database processing devices and the conventional method may be adopted in the other database processing devices. Is. The database use request includes various requests by general-purpose database access commands such as update, addition, deletion, and search reference. The inquiry device is a device that interprets the request and controls access to the database. A large amount of data is stored in the database processing device, and each data is composed of several items. Some database processors store data containing only the appropriate number of items, and other database processors store data containing some other item. Therefore, each of the database processing devices includes at least partially common items in its data. All may be common, or may include items that are not present in any other database processing device. The configuration information is data for displaying the contents.

The directory storage synchronization device plays a role of preparing availability information so that the database processing device can start operation while the unusable item remains when it is restarted. The availability information that cannot be used includes information in which a failure has occurred for some reason and information that should be updated together with other databases is in an unupdated state. The data of the access unit amount means the data transferred in one access when any other database is accessed to refer to the item that is the subject of the inquiry. If the inquiry request cannot be answered in one access, the access is performed twice, and the corresponding item is repaired twice.

(Structure 2) In this system, when there are a plurality of other database processing devices that receive data including the corresponding item, it is preferable to select the database processing device that can receive the corresponding data most economically. (Explanation) The most economical means, for example, when communication costs are low. This is because it is better to receive the same data economically.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to specific examples. FIG. 1 is a block diagram showing an embodiment of a distributed database system of the present invention. The system shown in the figure includes a plurality of database processing devices 10-1 to 10-4.
Are connected via the network 1. The database processing devices have substantially the same configuration. Here, only the database processing device 10-1 shows a specific block diagram.

The database processing apparatus 10-1 shown in the figure owns a database DB1 in which data D to be inquired is stored. The data D is composed of an appropriate number of items K. In order to execute such inquiry processing of the database DB1, the inquiry processing device 3, the failure recovery device 4,
Directory storage device 6, directory storage synchronization device 7
Is provided.

The inquiry processing device 3 is a part for processing an external request for use of the database DB1, for example, an inquiry request for update, addition, deletion, search, reference or the like. That is, it is a device having a conventionally well-known structure that performs a process of taking out corresponding data in the database DB1, editing necessary items, and replying to an inquiry destination. The failure recovery device 4 is a part that performs a certain restoration process called a state guarantee transaction when the database processing device 10-1 is restarted when the operation is stopped for some reason. Incidentally, in the case of the database processing apparatus of the present invention, this restoration processing is performed for each data of the access unit amount which will be described later.

The directory storage synchronization device 7 is operated by another database processing device 10-1 when the database processing device 10-1 is temporarily stopped for some reason and then restarted.
-2 to 10-4 and the like, which acquires the latest information including the database update status and updates the availability information 12 stored in the directory storage device 6. The directory storage device 6 is a part that stores configuration information 11 and availability information 12 for database management. Configuration information 11
Is information to be displayed including not only the configuration of its own database DB1 but also the configuration of the databases of the other database processing devices 10-1 to 10-4.

FIG. 2 shows a data content explanatory view of the database processing device. As shown in this figure, for example, the database D of the database processing apparatus 10-1 shown in FIG.
It is assumed that B1 stores data composed of items K1, K2, K3, and K4. Although this data is composed of a group of many data, it is shown by one band in the figure. It is assumed that the database processing device 10-2 of FIG. 1 stores the database DB2 shown in this figure. This database DB2 has items K1, K2,
It shall consist of K3. The database processing device 10-3 stores the database DB3. This database DB3 is composed of items K3 and K4.
Further, the database processing device 10-4 uses the database D.
It stores B4. This database DB4 has item K
2, K3 and K4.

That is, as shown in this figure, for example, the database DB1 and the database DB2 are the items K1 and K.
2 and K3 are redundantly stored. However, the item K4 is stored only in the database DB1. Then, for example, when a correction or update process for the item K1 is performed during the inquiry process for the database DB2, the same correction or update process is also performed for the item K1 of the database DB1 at the same time. However, if the database DB1 is inaccessible due to some trouble or the like, only the item K1 of the database DB2 is updated, so that the item K1 of the database DB1 is incorrect information if it is subsequently used for inquiry processing or the like. Will be provided. Therefore, item K of this database DB1
No. 1 is treated as unusable until necessary repair is performed. Conventionally, when the database processing device that manages the database DB1 is restarted, the contents thereof are collectively restored, and such an unusable state does not continue. However, in the present invention, even if there is a portion to be restored in the database, the operation is started, and such a state of being unusable continues.

FIG. 3 shows a specific content explanatory diagram of the configuration information 11 and the availability information 12 stored in the directory storage device 6. This figure shows each database DB1, DB
It shows what kind of items are stored for each data of 2, DB3 and DB4. Items that are not stored are indicated by "-". by this,
The structure of the database in each database processing device becomes clear. On the other hand, looking at, for example, the database DB1, data of items K1 to K4 are stored here. However, from the above explanation, for example, item K
It is assumed that while No. 1 is in the temporary stop of the apparatus, another database processing apparatus performs the update processing and the like and needs the restoration. In this case, as shown in the figure, for example, the database DB1
The item K1 is displayed as being unusable. This is the availability information 12. The above information is stored in the directory storage devices 6 of all the database processing devices.

For example, assume that the database processing device 10-1 shown in FIG. 1 is restarted after being stopped for a certain period of time. In this case, the directory storage synchronization device 7 refers to the configuration information 11 and the availability information 12 of the directory storage device 6 and updates them. That is, immediately after the restart, the directory storage synchronization device 7 notifies the other database processing devices 10-2 to 10-4 that the database processing device 10-1 has been restarted, and at the same time, its own database processing device 10-. While 1 is not operating, it obtains information about what items have been updated and deleted, and need to be repaired. This can be easily obtained by referring to the inquiry record or the like of each database processing device.
Then, the availability information 12 stored in its own directory storage device 6 is updated to the latest one. If there has been an operation such as restarting before, not all the items have been usable so far. Therefore, the availability information is updated by adding the newly disabled ones including the previously disabled ones.

In that case, as shown in FIG.
For example, the item K1 of the database DB1 is displayed as unavailable and the latest version of the data is displayed in the database DB1.
The fact that it is stored in 2 is also displayed at the same time. This is important information for future reference in case you need to repair it. The system of the present invention is characterized in that after the directory storage synchronization device 7 performs such processing, the operation is started in a state where all data is not completely restored.

Further, at the time of restoration, in the system of the present invention, the data of the access unit amount is updated. FIG. 4 illustrates an explanatory diagram showing the contents of data update of such an access unit amount. In the present invention, when an inquiry request is made for an item having a failure, that is, an item requiring repair after the start of operation, the part is repaired only for the first time. At this time, the restoration is not performed collectively for all the data. If all the data are collectively repaired, the communication line is occupied for a long time at that time, and a large load is applied to the database processing device of the other party who receives the data transfer, as in the conventional case.

On the other hand, in the system of the present invention, when any one of the databases is accessed in order to refer to the item that is the object of the inquiry, only the data transferred by the single access is restored. This is called an access unit amount, but if this amount of restoration is performed, the load on the communication line and the other party's database processing device will be small. As shown in FIG. 4, for example, with respect to all the data in the database DB1, it is assumed that the data of the access unit amount is the amount shown by hatching in the figure. In this case, this amount of data restoration is performed each time there is an inquiry request, and if there are several inquiry requests, restoration of all data is naturally completed.

The access unit amount of data is a sufficiently small amount of data as compared with all the data, and more than one data, the recovery efficiency is high. Specifically, 1
When an inquiry request is made to the item K1 of the data of the item, in one access, about 1000 items of data are transferred from the tens of thousands of data from the database processing device having the latest data. This transfer amount is an amount set depending on the capacity of the buffer memory and other circumstances.

The operation of the apparatus of the present invention will be described below in order with reference to a concrete flow chart. FIG. 5 is a flowchart of restart processing after a failure occurs. First, step S
If a failure occurs in 1, the temporary stop process is performed in step S2. Then, when it is restarted in step S3, as described above, the directory storage synchronization device 7 operates to store the necessary availability information 12 in the directory storage device 6. That is, here, the availability information 12
To the latest one. Of course, at that time, the information as shown in FIG. 3 described above is also stored at the same time. Thus
Start operation of the device.

Next, the operation when there is an inquiry for a search request will be described with reference to FIG. First, when an inquiry request is made, in step S1, the inquiry processing device 3 requests the directory storage synchronization device 7 to determine availability of the corresponding item. In step S2, the directory storage synchronization device 7 refers to the directory storage device 6 to confirm an item that has not been restored. If the inquiry request includes only usable items, the process jumps from step S3 to step S8, and the search process and the reply are performed as usual. On the other hand, when it is determined that the item needs to be repaired, the directory storage device 6 determines the availability of the same item of another database processing device in step S4 and notifies the failure recovery device 4 of the availability.

That is, the stored contents of the directory storage device 6 shown in FIG. 3 include information indicating whether or not other databases can be used. Therefore, the usable database is determined and the failure recovery device 4 is notified. In step S5, when one database processing device is designated, the failure recovery device 4 makes a data transfer request to the database processing device. On the other hand, when two or more databases are designated, the latest data is requested to the database processing device with the lowest cost.

A low-cost database processing device is, for example, a device with low communication cost. Communication charges will increase if data is transferred from a long-distance database processing device, so a short-distance data processing device is selected. Next, step S
At 6, the failure recovery device 4 accepts a predetermined data unit containing the relevant item from its database processing device. Then it updates its own database. Then, the process proceeds to step S7, the display indicating that the directory storage device 6 is unusable is erased, the directory storage device 6 is made available, and the process proceeds to step S8.
Perform normal search processing and answer. If it is not possible to respond to the requested inquiry after receiving the data transfer in one access unit, it is possible to receive the data transfer in multiple times. The same applies when an inquiry such as reference or deletion is made.

As described above, it is possible to prevent a situation in which a large amount of data transfer for data restoration is concentrated when a failure occurs and the database processing device once stops its operation and is restarted.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a distributed database system of the present invention.

FIG. 2 is an explanatory diagram of data contents of a database processing device.

FIG. 3 is an explanatory diagram of stored contents of a directory storage device.

FIG. 4 is an explanatory diagram of data updating for each access unit.

FIG. 5 is an operation flowchart of restart processing after a failure occurs.

FIG. 6 is a flowchart of a search operation.

[Explanation of symbols]

 3 Query Processing Device 4 Failure Recovery Device 6 Directory Storage Device 7 Directory Storage Synchronous Device 10-1 to 10-4 Database Processing Device DB1 Database

Claims (2)

[Claims] 1. A plurality of database processing devices in which arbitrary items in predetermined data are at least partially overlapped with each other and distributed and stored, and a database use request is issued to all or some of the database processing devices. An inquiry device to be processed, a configuration information indicating a data configuration of each database processing device, a directory storage device storing availability information indicating whether or not each item of the data is available, and a database processing device after the suspension. When restarting, the latest information is obtained from another database and the availability information is updated, and when there is a request to use the database, the corresponding information in the data stored in its own database processing device Refer to the availability information of the item to be used, and if it is not available, from another database processor Distributed database system that accept data access unit amount including those items and repair the desired item, and further comprising a fault recovery apparatus for changing the appropriate availability information in the directory memory. 2. The database processing device capable of receiving the most economically applicable data is selected when there are a plurality of other database processing devices that accept the data including the corresponding item. Distributed database system.