JPS63278155A - Management processing system for decentralized data base - Google Patents

Abstract

PURPOSE:To maintain and secure the consistency of data by feeding the updating log of a center of its own collected during a certain period to a remote center and disposing a means reflecting on a data base of its own in the remote center. CONSTITUTION:The data bases 10A, 10B having the data of the same contents are developed on computer systems 1A, 1B. Updating log transfer parts 5A, 5B in data consistency processing parts 4A, 4B transfer updating log information stored in the log data sets 9A, 9B of the center of its own to the remote center periodically at night. Updating reflecting processing parts 6A, 6B receive the updating log information from the remote center, calculates an updating difference and the updating difference of the center of its own and compares. When a remote updating item is not updated in the center of its own, the contents of updating are reflected in the data bases 10A, 10B of the center of its own. When it is updated in the center of its own, whether they are reflected or not decided according to the predetermined level of the center and processed and when the reception processing of the updating log is completed, the center of a transmission side marks a next transfer.

Description

[Detailed Description of the Invention] [Summary] In order to solve the problem that when databases are centrally managed in multiple computer systems, the use of databases other than a specific center is restricted, each system has a database with the same contents in principle. The update logs of the own center collected within a certain period are sent to the other center, and the other center has a means to reflect the update logs in its own data directory, thereby allowing multiple computer systems to operate asynchronously and independently. Maintain and guarantee data integrity of updated databases.

[Industrial application field]

The present invention relates to a distributed database management processing method that maintains and guarantees data integrity even when databases are updated asynchronously and independently at multiple centers.

[Conventional technology]

FIG. 6 shows an example of a conventional distributed system using communication line connections.

When computer systems were connected by communication lines, it became possible to access databases in other systems in real time, and it became possible to centrally manage databases. However, in this case, the database is concentrated in a specific center, and it is not possible to maintain and use the database in another center. Furthermore, when databases are placed in multiple centers, if both centers are updated, consistency cannot be maintained, so only one system can be updated.

For example, in the distributed system shown in FIG. 6(a).

Each user Ul, U2 has their own computer system IA.
, IB, but files x and y are not distributed, so two files x and y cannot be used at the same time.

In the distributed system shown in FIG. 6(B), each user and each file are distributed to computer systems IA and IB, and the file X held by computer system IA is It is designed to be copied to X'. However, since there is no means to reflect the update of file X' in file X, file X' is only for reference. In order for user U2 to update, he must update file X held by computer system IA.

[Problem that the invention seeks to solve]

According to the conventional method described above, the use of the database is limited depending on the system, and the load is concentrated on a specific center.
In addition, the update process requires access to other systems via two lines, resulting in a slow response.

The present invention aims to solve the above problems, and makes it possible to update the database asynchronously and independently, as if each computer system maintains its own database, and to maintain data consistency even if the database is updated asynchronously and independently. Maintain sex/
The purpose is to provide a guarantee.

[Means for solving problems]

FIG. 1 shows an example of the basic configuration of the present invention.

In FIG. 1, LA and IB are computer systems 2A each having a CPU, memory, etc.

2B is a terminal that refers to and updates the database; 3A;

3B is an application processing unit that processes references and updates to the database; 4A and 4B are data consistency processing units that ensure consistency of the databases of each computer system; and 5A and 5B are transfers of update logs to other computer systems. Update log transfer units 6A, 6B are update reflection processing units 7A, which reflect update logs sent from other computer systems to their own systems;
7B is a data management unit that provides a database access function, 8A and 8B are update history storage processing units that collect update log information, 9A and 9B are log data centers where update log information is stored, and IOA and IOB are the same. Represents a database in which content data is expanded.

41 In the present invention, a database 10A has data with the same contents in a plurality of computer systems LA and IB-h.

10B is expanded. These databases 10A, 1
0B is each computer system LA.

In the IB, the information is referenced and updated asynchronously and independently via the application processing units 3A and 3B and the data management units 7A and 7B.

The update history storage processing units 8A and 8B are database IOA
, when updating the IOB, the logical key of the record,
Update log information such as changed item names and changed item values are stored in log data sets 9A.

Output to 9B.

The update log transfer units 5A and 5B are operated during the night, for example.

A process of periodically transferring update log information stored in the log data sets 9A and 9B of the own center to the other center is executed. This transfer can also be done asynchronously.

When the update reflection processing units 6A and 6B receive the update log information sent from the partner center, the update reflection processing units 6A and 6B calculate the update difference (
The final value (or start value and final value) of the update content for each logical key and the update difference at the own center are calculated and the two are compared. If the updated items at the other center have not been updated at the own center, the updated contents are reflected in the databases 10A and 10B of the own center. If the own center has also been updated, it is determined whether or not to reflect the update based on the predetermined level of the center, and processing is performed accordingly. When the reception process for the update log is completed, this fact is reported to the originating center, and the third originating center marks the 9th transfer bone.

[Effect]

According to the present invention, the databases IOA and IOB basically hold data with the same content, and can be referenced and updated asynchronously and independently in the respective computer systems LA and IB. By updating the database IOA
, IOB, although there is a temporary mismatch between them, the data matching processing unit 4A.

With 4B, updates in other computer systems are regularly reflected in the database of the own system.
It becomes possible to maintain and guarantee data integrity.

Although the example in FIG. 1 shows distributed database management in two computer systems, data consistency can be similarly guaranteed in distributed database management in three or more computer systems.

〔Example〕

2(A) and 2(B) are transfer flows according to an embodiment of the present invention, FIG. 3 is an explanatory diagram of data matching processing, and FIG.
An example of the processing configuration of the illustrated data matching processing section.

FIG. 5 shows an explanatory diagram of the operation side according to an embodiment of the present invention.

The transfer flow of the system according to the embodiment of the present invention shown in FIG. 1 will be described below with reference to FIG. 1st
Although the computer system IA shown in the figure will be described as the sending center and the computer system IB as the receiving center, the reverse transfer is also the same.

It can be processed independently. The numbers ■~@ in the following description correspond to ■~[phases] shown in 5.FIG. 2 (AL (B)).

■ As shown in Figure 2 (A), when there is an update to the database IOA in the job of the 9 originating center,
By the update history storage processing unit 8A.

The update history data is stored in the log data set 9A.

(2) Log data is periodically collected and transferred to another computer system, such as when a service ends, and at that time a transfer log record indicating the start of one transmission is generated and written at the final position of the log data set 9A.

- Read the log record of update data from the log data set 9A, and if there is update data with the same logical key, integrate them to create 2 outgoing data and store it in the 2 outgoing data center.

■ Send 9 outgoing data sets to the receiving center using existing file transfer processing functions.

Also, when the computer system IA, which is the originating center, previously received the update data of the computer system IB,
If there is something that could not be reflected in the own database IOA, a data set consisting of updated data that has not been reflected on the sending side is sent to the receiving side center. After that, computer system I
Start B's reception job.

■ As shown in Figure 2 (B), in the receiving center's job, the update data that was not reflected in its own data base IOB when it was previously received from the originating center, and the update of the originating data center received this time. An update difference is calculated from the data, and 9 originating side difference data sets are created using the difference record. In addition, the receiving side log data center 9B stores the updated data that has not been reflected in the previous time at the sending center and the updated data up to that point at the receiving center.
Calculate the update difference with.

Generate delta records to create a receiver delta dataset. Then, compare the contents of the first differential data set on the transmitting side and the differential data set on the receiving side, and
Create a reflection data storage data set consisting of the updated data to be reflected. Updated items that were updated at the originating center but not updated at the receiving center will be reflected in 2 data, but for items updated at the 9-ryo center, only one of the updated data will be reflected depending on the predetermined priority level of the center. Generate reflection data so that it is valid.

■ Read the reflected data record from the reflected data storage data center and reflect the updated contents in the database IOB. Also, at this time, if there is data that cannot be reflected in the database IOB because another user is occupying the resource.

The reflected data is saved as an unreflected data set. This unreflected data is used to create the originating side difference data center in the process (2) above when update data is received the first time.

- When the update reflection process to the database IOB is completed, a reply record is generated to notify the originating center of the completion of the process, and a reply data center is created.

■ Generate a transfer log record indicating completion of reception.

Store it in own log data set 9B.

■Reply dataset created by processing■.

Send it to the originating center and start the reply reception job. As a result, the process [phase] shown in FIG. 2(A) operates.

[Phase] The originating center receives the transfer result from the receiving center, and based on the contents of the reply data center,
The transfer log record written in the above process (2) indicating the start of the call is updated to indicate the end of the transfer. Next, when updating data is sent to the receiving center, the updated data after this transfer log record is used to create 9 outgoing data cents.

As in the example below, by saving the unreflected update data as an unreflected data set,
Even data that could not be reflected because it was occupied by another user during transfer and reception can be dealt with later. Also, by securing a data set for saving 2 reflected data.

= 11 = It is also possible to recover if the job ends abnormally during transfer reception.

Next, a specific example of data matching processing will be explained according to FIG.

As shown in FIG. 3(A), the database 1° consists of DB records R1, R2, . . . having two items 11 to i1. Now, by one update, 3 items, for example, items if, i5, of record R1. i8. Suppose that ill has been updated. As shown in FIG. 3(b), the update data consists of a record indicating the update item number, a record with the value of each item before the update, and a record with the value of each item after the update. One of the items is a logical key that identifies the record. The updated data after the previous transfer log record indicating the end of transfer is to be transferred to the receiving center.

For example, if the value of item i5 in the same record is updated from X to 2, and then from 2 to y, the two updated data are integrated at the time of transfer.

Generate outgoing data indicating that X has been updated to y.

The generation of reflected data at the receiving center is
This is done as shown in Figure (c). In addition.

To simplify the explanation, only updated data for one record is illustrated.

At the receiving center, it is the unreflected portion from the previous reception.

One originating side difference data set is created from the originating data of the originating data center received this time. for example.

Regarding item i5, a difference record is generated assuming that the old value rWJ has been updated to the new value ryJ. Similarly, a receiving side difference data set is created from the unreflected data from the previous reception at the second sending center and the receiving side log data set.

By comparing the difference data set on the sending side and the difference data set on the receiving side, a reflected data storage data set consisting of the reflected data to be reflected in the own data domain is created.

The data matching processing units 4A and 4B shown in FIG.

For example, it is composed of a group of programs that perform the processing shown in FIG.

In the transfer data transmission process 40, the transfer log record addition process 41. A log record reading process 421, a record output process 43 to an outgoing data set, a print process 44 to a Japanese line printer, etc. are performed.

In the processing 50 for reflecting the reply notification from the receiving center, processing 51 for rewriting the transfer log record, processing 523 for reading the work data center, printing processing 53 on the Japanese line printer, and the like are performed.

A reception driver 60 that receives update data performs differential data center generation processing 61 such as assembling differential records, outputting differential records, and merging items that have the same logical key for differential records with one originating side;
Do this for the receiving side. Thereafter, a process 62 for generating a nine reflection data set and a hint list is performed by rewriting the entry number, which is a physical key of the database, setting a hit list, and generating reflection target data.

In the database reflection process 63, a 2-notification reply record is generated, the resource management file is accessed, the record is checked, and the data is written to the database. At this time, if there is unreflected data due to competitive exclusive control or the like, it is output to the unreflected data storage center. Finally, exclusive control is released by accessing the resource management file.

If the job ends abnormally during transfer reception, the reception recovery driver 70 is activated. The reception recovery driver 70 performs 1 processing 71 for expanding the key of the reflected data from the reflected data storage data center, and 2 processing 72 for generating a hint list of the reflected data.

Processing 73 for reflecting on the database and processing 74 for accessing the resource management file are performed. Note that the detection of the break point is
This is done using the resource management file.

Transfer log record addition processing 8 as the common processing unit 80
1. Transfer log record rewriting process 82, transfer log record reading process 83. Print processing 84. Record input/output processing 85 and the like are provided.

The comparison target log range of =15- in the update reflection processing units 6A and 6B shown in FIG. 1 is set to transfer #1 to transfer # shown in FIG.
This will be explained according to 10 operation examples.

Transfer #1: From computer system (B) to computer system (
When update B data is sent to A), the computer system (A)
), updates B and A are compared.

Transfer #2: In the computer system (B), transfer #2 compares the tertiary (al, (bl). Note that in the following explanation, "survivors" refer to valid data that has not yet been processed on the other side. This means updated data.

(al　Update A that survived #1 and update C (Updated enough to reflect bl #1 Transfer #3: Compare the next ta) and tb>.

(al (#1 Ten updates D for unreflected parts), ten updates that survived in #2 F (bl#2 Ten updates E for unreflected parts) Transfer #4: Next fat, (Compare bl.

(Al#3 Unreflected tenth update H (bl (#2 Unreflected tenth update E) that survived #3 Update G Transfer #5: Compare next +al, (bl).

fal l#2 10 updates of unreflected portions of E) that survived in #3 10 updates of G) that survived in #4 10 update I fbl#4 10 updates of unreflected portions J Transfer #6: Next +al , (b).

fal #5 10 updates of unreflected portions K (b) Of (#4 10 updates of unreflected portions J), those that survived in #5 are updated and transferred #7: Compare the next fal, (bl).

fa) #6 unreflected 10 updates 0 fb) (((#4 unreflected 10 update J) survives #5) 10 updates) and #6 survivors 10 updates P 10 updates R Transfer #8: Next Compare (al, (bl).

fal (#4 unreflected tenth update J), #5 survival ten update ten update P (b) #6 unreflected tenth update 0 + update Q transfer #9: next fat, (compare bl .

(al (#7 unreflected tenth update Q), 4↑8 surviving ten update 5 (b) #7 out of (#8 unreflected tenth update R), #7 surviving ten update T Transfer #10: Next Compare (al, (bl).

(at (#8 Unreflected 100 update R), #9 surviving 10 updates ■ (bl #9 Unreflected 10 updates Data consistency of each computer system can be guaranteed even if the logs are transferred asynchronously.However, in order to simplify the processing.

The update log transfer process may be performed by each computer system in synchronization, for example, at night after the end of a general online service, and the processing of unreflected data may be omitted.

C. Effects of the Invention] As explained above, according to the present invention, data integrity can be maintained and guaranteed even if databases with similar contents held by multiple computer systems are updated asynchronously and independently. This enables efficient distributed database management.

[Brief explanation of drawings]

FIG. 1 is an example of the basic configuration of the present invention, FIGS. 2(A) and 1(B) are transfer flows according to an embodiment of the present invention, FIG. 3 is an explanatory diagram of data matching processing, and FIG. 4 is the same as that shown in FIG. 1. FIG. 5 is an explanatory diagram of the operation side according to an embodiment of the present invention, and FIG. 6 shows an example of a conventional system. In the figure, LA and IB are computer systems, 2A and 2B are terminals, 3A and 3B are application processing units, 4A and 4B are data matching processing units, 5A and 5B are update log transfer units, and 6A and 6
B is update reflection processing unit, 7A, 7B is data management unit, 8A
, 8B is an updated layer thickness preservation processing section. 9A and 9B represent log data sets, and IOA and IOB represent databases.

Claims (1)

[Claims] A distributed database management processing method in which a database with the same data is developed in a plurality of computer systems, and each computer system updates the database asynchronously and independently, the method comprising: update history storage processing means (8A, 8B) that collects update log information; update log transfer means (5A, 5B) that collects update log information collected within a certain period and transfers it to another computer system; Update reflection processing means (6A
, 6B), and guarantees data consistency of databases in a plurality of computer systems.