Distributed type of database system, consistency managing...

Details Distributed type of database system, consistency managing... Distributed type of database system, consistency managing...

1998-12-07

2001-09-04

Black, Thomas (Department: 2171)

Data processing: database and file management or data structures

Database design

Data structure types

C707S793000, C707S793000, C707S793000, C707S793000

Reexamination Certificate

active

06286004

ABSTRACT:

FIELD OF THE INVENTION
The present invention relates to a distributed type of database system enabling consistency management to prevent generation of contradiction in copied data even in an extremely large scale database system extending over extremely large number of sites, a consistency managing method for the same, and a computer-readable recording medium with a program for making a computer execute each step of the method recorded therein.
BACKGROUND OF THE INVENTION
In a distributed type of database system, copied data for an object is distributed among a plurality of sites, and a plurality of transactions are concurrently executed. Therefore, it is very important to preserve consistency of copied data distributed over the system.
As one example of consistency managing method in a distributed type of database system based on the conventional technology, there is a so-called optimistic control system based on time stamps provided under centralized management (described as “optimistic control system” hereinafter).
FIG. 21
is a block diagram showing general configuration of a distributed type of database system realizing the optimistic control system. This distributed type of database system comprises a main site
2100
having master data, and a plurality of database (DB) sites
2101
to
2104
each having copied data generated from the master data. The main site
2100
centralizedly manages consistency of the master data and copied data. When transaction for updating is generated in any of the DB sites
2101
to
2104
, the DB sites
2101
to
2104
can not update copied data stored in the site before updating is permitted by the main site
2100
.
Processing in the optimistic control system consists of three phases of read-out phase, check phase and write phase. Generally, in the read-out phase, data is read out and written in only a private area for transaction (in other words, work area). In the check phase, checking is executed for any conflict with other transaction for insuring the possibility of serialization. Further, in the write phase, when updating is permitted as a result of checking in the check phase, updated data is permanently written in a database at each of the DB sites
2101
to
2104
, and when updating is not permitted, the transaction is aborted.
The read-put phase, check phase and write phase are specifically described with reference to FIG.
21
. For instance, when transaction for updating is generated in the DB site
2101
, the DB site
2101
reads out the copied data to be updated into the private area for transaction and updates the read-out copied data in the private area for transaction (read-out phase). Then the DB site
2101
transmits an updating request information including a value after updating to the main site
2100
. In response to the request for updating from the DB sites
2101
, the main site
2100
plays a role as an arbitrator for any conflicting request for updating and determines whether the request for updating is to be permitted or not (check phase). When the main site
2100
permits the request for updating, the DB site
2101
updates copied state stored in the site and executes processing for permanently writing the updated copied data (write phase). On the other hand, when the request for updating is not permitted because of any conflicting request for updating, the DB site
2101
aborts the generated transaction.
The main site
2100
determines whether the request for updating is to be permitted or not according to a time stamp given to the data to be updated. In addition, the main site
2100
updates, when permitting the request for updating, the master data, and transmits a value after updating also to the DB sites
2102
to
2104
other than the DB site
2101
having transmitted the request for updating. As a result, also in the DB sites
2102
to
2104
, the corresponding copied data is updated using the value after updating transmitted from the main site
2100
. Thus a consistency of data is maintained in the entire system. In other words, the possibility of serializing data is insured in the check phase described above, and enhanced consistency management is enabled for preventing generation of contradictions among data.
In the optimistic control system described above, as the centralized management system is employed, the authorized latest value is preserved in the main site
2100
. Therefore, when data consistency is checked in the check phase in the optimistic control system, each of the DB sites
2101
to
2104
is required only to issue an inquiry to the main site
2100
and it is not required to lock other copied data. In addition, even when any of the DB sites is down, there is a advantage that the DB site can be ignored to continue the processing.
In the conventional technology as described above, data consistency over a system as a whole can be preserved. However, when an extremely large distributed type of database system extending over an extremely large number of database sites is constructed, it is physically impossible to deliver, each time data is updated, information indicating updating of the data to each of the extremely large number of distributed DB sites from the main site.
Furthermore, in the conventional technology as described above, when a data is updated, there is generated a time lag in delivery of information indicating updating of the data between a DB site that receives the updated data first and a DB site that receives the updated data last. Therefore, consistency management capable of suppressing generation of data contradiction in the entire system at an arbitrary point of time is physically impossible.
In addition, as the main site arbitrates conflicts among requests for updating from a plurality of DB sites, in an extremely large distributed type of database system, work load on the main site remarkably increases. In other words, when a data is updated, the main site is required to transmit information indicating updating of the data to each of the DB sites, and in a case of an extremely large scale distributed type of database system extending over an extremely large number of database sites, work load on the main site remarkably and disadvantageously increases.
Furthermore, because of delay in processing due to remarkable increase of work load on the main site, sometimes transaction in each DB site may be aborted. In such a case, each DB site executes the same transaction again and this results in an increase of a number of unnecessary processing. Especially, in the time stamp-based optimistic control system, priority of transaction for updating in each DB site is determined according to a time stamp, and it is generally believed that the possibility of abortion of transaction is higher as compared to the lock system in which an earlier request for updating is preferentially treated.
In addition, although arbitration for solving conflict among request for updating is executed according to a time stamp, however, there is no global time in each site in a distributed type of database system. Therefore, when time clock used for management in each DB site is not coincident, time of updating of data does not always indicate which data is new or old, which makes it impossible to preserve consistency among data.
SUMMARY OF THE INVENTION
It is a first object of the present invention to enable enhanced consistency management capable of preventing generation of contradiction among data even in an extremely large scale distributed type of database system that extends over an extremely large number of database sites.
It is a second object of the present invention to enable reduction of work load on the main site, even in a case where centralized management in which only the main site can permit updating of data is employed.
It is a third object of the present invention to reduce a number of aborted transactions and improve the processing efficiency in the main site as well as in each DB site by employing centralized management by the main site for making it possible for each DB site to check

Affiliated with

Also associated with

Rating

Distributed type of database system, consistency managing... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Distributed type of database system, consistency managing..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed type of database system, consistency managing... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2486221