JP5867902B2 - Asynchronous database replication method - Google Patents

Description

  The present invention relates to a database asynchronous replication method for performing replication to a backup system database based on a front system database.

  Conventionally, as a method of replicating to the backup system database based on the front system database, the update log when the front system database is updated is transferred to the backup system asynchronously, and the same update processing is executed on the backup system. There is a method to follow. When performing asynchronous replication, there is a possibility that updates may be lost when a front system failure occurs. However, there is less impact on front system processing compared to synchronous type, and the throughput of front system processing is reduced due to replication processing. There is an advantage that can be suppressed.

  Here, consider a case where an update log is transferred directly from the front system to each backup system and asynchronous replication is performed for each of them with the configuration of one front system and multiple backup systems. Since the replication processing in each backup system is independent, the progress of each replication also proceeds independently. In such a configuration, when a front system failure occurs, one of the backup systems becomes a new front system, and the same replication process is started from the new front system to the remaining backup systems.

  FIG. 1 is a diagram showing a configuration of such a front system and a plurality of backup systems.

  At this time, as shown in FIG. 1, if the update log that has not yet reached or reflected in the new front system 1 has already been replicated in the remaining backup system 2, the backup database is in the preceding state. Therefore, inconsistency between the databases of the new front system 1 and the backup system 2 occurs. As a result, it is impossible to continue the replication process with the new front-backup system configuration. For this reason, when such a configuration is adopted, the replication progress status of the backup system 1 serving as the new front system must be preceded by other backup systems 2.

  As a conventional method for replicating a front system database to a plurality of backup systems and constructing a new front system-backup system in the event of a front system failure, as disclosed in Patent Document 1, another method is used. There is a cascade method that replicates to other backup systems. However, in the case of the cascade method, the arrival delay of the update log to the backup system at the end increases. In addition, it has been pointed out that when a communication failure or the like occurs in the relay backup system, it also affects the replication to the backup system at the end.

Japanese Patent No. 3958757

  In this way, in a system configuration that requires high availability, in addition to the configuration of the front base-backup base, a base that replicates the database for another purpose during normal operation of the system and operates as a backup system when necessary It is necessary to consider backup transfer in the added three-site configuration. At this time, it is necessary to consider a method that allows a new front-backup configuration to be continued at the remaining bases in the event of a front base failure.

  The replication method according to a predetermined embodiment of the present invention is a multiple replication method in which a front system database is asynchronously replicated to a plurality of backup systems in a distributed database. Perform disaster recovery so that the alternate backup system switches to the priority backup system to the front system. The replication processing progress of each backup system is controlled so that the backup transfer from the new front system to the new priority backup system is immediately possible. This makes it possible to adopt a new front-backup configuration between the backup systems when a front system failure occurs.

  Preferably, of the plurality of existing backup systems, the progress of replication in the backup system that becomes the new front system when a front failure occurs is notified to the remaining backup systems. In the replication of the remaining backup system, it is determined that replication is possible up to the notified replication progress status, and the replication processing is executed up to the notified replication progress status. By introducing such a mechanism, it is possible to avoid that the replication status of the remaining backup system precedes that of the backup system that becomes the new front system, and when the front failure occurs, the new front system moves to the remaining backup system. Replication can be continued.

  The predetermined embodiment of the present invention is effective in a mission critical system that requires high availability, but may have the following effects.

  First, according to a predetermined embodiment of the present invention, when asynchronous replication is performed from a master database existing in a front system to a database of a plurality of backup systems, the replication progress status of a specific backup system is changed to that of another backup system. It is possible to precede the progress of replication.

  Second, the ability to control the progress of replication enables a specific backup system to operate as a new front system in the event of a front system failure, and to resume replication from the new front system database to the remaining backup systems. In this case, it is possible to avoid the occurrence of data inconsistency due to the replication progress of the databases in both systems.

  In particular, this configuration is effective when the time required until the backup system can operate as a new front system is different. Specifically, the following system configuration can be taken. The high-priority backup system that becomes the new front system in the event of a front system failure performs resource operation and personnel assignment so that it can immediately become the front system. On the other hand, when the front system is normal, the low priority backup system replicates the front system database for information sharing or the like, and performs resource operation and personnel assignment equivalent to the high priority backup system only when the front system fails. In the case of such a system configuration, since the priority order for becoming a new front system is determined in advance between backup systems, this embodiment is effective.

  Further, from the viewpoint of disaster recovery, the location of the backup system having a high priority as the new front system is selected at a location far from the front system. At this time, if the backup system with a lower priority is closer to the front system than the backup system with a higher priority, the priority is lower unless replication progress control is performed as in the predetermined embodiment of the present invention. There is a high possibility that the replication progress of the backup system will be preceded. In the case of such a system arrangement, the predetermined embodiment of the present invention is effective.

  Third, since the replication progress status can be controlled, the following selection can be performed from the viewpoint of operation when the backup system preceded by the replication progress status is operated as a new front system. That is, the application is started after all the unreflected update logs stored in the update log storage table are reflected in the database. Alternatively, an unreflected update log stored in the update log storage table is discarded and the application is immediately started.

  Fourth, because the update log itself is transferred directly from the front system to each backup system, if a failure occurs in the backup system that becomes the new front system during normal operation of the front system, the new operation can be It is possible to switch the backup system as the front system. That is, an operation of changing the reflection permission serial number of the backup destination backup system to the last received update log serial number is performed. Then, an operation of changing the priority backup system of the update log reflection permission serial number management table of the front system to the switching destination backup system is performed.

  As described above, even when a failure occurs in the backup system serving as the new front system, it is possible to switch the other backup system to the backup system serving as the new front system simply by performing an operation within each system. In addition, when a failure occurs in a backup system that relays data as in the cascade method, there is no effect on the replication processing to other backup systems, so the fault tolerance is improved.

It is a figure which shows schematic structure of the conventional front-backup system. 1 is a block diagram showing a schematic configuration of a front-backup system according to an embodiment of the present invention. It is a figure which shows an example of the update log reflection permission serial number management table 006 in a present Example. It is a figure which shows an example of the update log reflection serial number management table 110 in a present Example. 6 is a flowchart illustrating a process in which a certain update log is reflected in the database 101 of the backup system 1 and the database 201 of the backup system 2 in the present embodiment. It is a block diagram which shows schematic structure of the front-backup system which concerns on other embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, the same code | symbol is attached | subjected to the same element and the overlapping description is abbreviate | omitted.

  FIG. 2 is a block diagram showing a schematic configuration of a front-backup system according to an embodiment of the present invention. As shown in the figure, in this embodiment, the front-backup system includes a front system 0, a backup system 1, and a backup system 2, and each system includes a database 001, a database 101, and a database 201, respectively.

  The database 001 included in the front system 0 is a database that is a replication source of the database of each backup system. The backup system 1 is a backup system that operates as a new front system when a failure occurs in the front system 0. The backup system 2 is a backup system that operates as a backup system even when a failure occurs in the front system 0.

  The front system 0 includes an application 002 that updates the database 001. The application 002 includes an update log storage unit 003. The update log storage unit 003 stores an update log of database update in the update log storage table 004. The update log storage table 004 stores the update log together with a serial number indicating the update log generation order. The update log transmission unit 005 indicates the update log stored in the update log storage table 004, the update log sequence number and the reflection permission sequence number, the update log reception unit 107 on the backup system 1, and the update log reception unit 207 on the backup system 2. Send to. Here, the reflection permission serial number is the last serial number of the update log that may be reflected by each backup system. The update log transmission unit 005 has a function of resending the update log in response to a request from the update log reception units 107 and 207.

  FIG. 3 is a diagram illustrating an example of the update log reflection permission serial number management table 006 in the present embodiment. In the figure, a reflection permission serial number 0061 is a reflection permission serial number that is referred to when the update log transmission unit 005 performs transmission. The notified reflection permission serial number 0062 represents the reflection permission serial number already notified to each backup system. The priority backup system 0063 represents a backup system serving as a reference when determining the reflection permission serial number.

  Returning to FIG. 2, the update log receiving unit 107 of the backup system 1 stores the update log received from the update log transmitting unit 005 in the update log storage table 108. The received reflection permission serial number is recorded in the update log reflection serial number management table 110. The update log reflection unit 109 refers to the update log reflection sequence number management table 110, acquires the update log to be updated from the update log storage table 108, and reflects the update log in the database 101.

  FIG. 4 is a diagram illustrating an example of the update log reflection serial number management table 110 according to the present embodiment. The update log reflection sequence number management table 110 manages update log sequence numbers. In the figure, a reflection permission serial number 1101 represents the last serial number of the update log that can be reflected in the database 101 by the update log reflection unit 109. The reflected serial number 1102 represents the last serial number that has been reflected by the update log reflecting unit 109. The update log reflection sequence number management table 210 of the backup system 2 also has the same data structure as the update log reflection sequence number management table 110, and includes a reflection permission sequence number 2101 and a reflected sequence number 2102.

Returning to FIG. 2, the update log receiving unit 207, the update log storage table 208, the update log reflection unit 209, and the update log reflection serial number management table 210 of the backup system 2 have the same functions as those in the backup system 1.

  Next, based on the flow of the update log represented by the solid arrow in FIG. 2, the operation of the present embodiment will be described with reference to FIG. FIG. 5 is a flowchart showing a process in which a certain update log is reflected in the database 101 of the backup system 1 and the database 201 of the backup system 2 in this embodiment. In FIG. 5, the update log and its serial number are mainly described in the process of storing in each table. A description surrounded by a broken line indicates processing in the front system 0. In the description surrounded by a solid line, the left side shows processing in the backup system 1 and the right side shows processing in the backup system 2.

  At the same time that the application 002 of the front system 0 updates the database 001, the update log storage unit 003 stores the update log in the update log storage table 004 (S501). When storing the update log, a serial number indicating the update log generation order is assigned.

  The update log transmission unit 005 periodically monitors the update log addition status in the update log storage table 004, and if a newly stored update log exists, acquires the update log and its serial number. Further, the update log transmission unit 005 determines a reflection permission serial number to be transmitted to each backup system. The reflection permission serial number is determined as follows.

  First, when transmitting to the system indicated by the priority backup system 0063 in the update log reflection permission sequence number management table 006 (the backup system 1 in the embodiment of FIG. 3), the sequence number of the last update log acquired from the update log storage table 004 is displayed. Use a reflection permission serial number.

  Next, when transmitting to a system other than the system indicated in the priority backup system 0063 of the update log reflection permission serial number management table 006 (backup system 2 in this embodiment), the serial number of the reflection permission serial number 0061 of the update log reflection permission serial number management table 006 Is the permission serial number.

  The update log transmission unit 005 transmits the acquired update log, its serial number, and the determined reflection permission serial number to the update log reception unit 107 of the backup system 1 and the update log reception unit 207 of the backup system 2, respectively (S502, S507). ). When the transmission is completed, the notified reflection permission serial number 0062 of the update log reflection permission serial number management table 006 is updated.

  The update log reception unit 107 of the backup system 1 stores the update log received from the update log transmission unit 005 in the update log storage table 108. Further, the reflection permission serial number 1101 of the update log reflection serial number management table 110 is updated with the value of the received reflection permission serial number (S503). In the backup system 1, the reflection permission serial number 1101 is equal to the serial number of the last update log received from the update log transmission unit 005. After the update log reflection serial number management table 110 has been updated, the update log reception unit 107 acquires the value of the reflected serial number 1102 in the update log reflection serial number management table 110 and returns the serial number to the update log transmission unit 005 (S505). ).

  The update log reflection means 109 of the backup system 1 monitors the update log reflection serial number management table 110. If the value of the reflection permission serial number 1101 is larger than the value of the reflected serial number 1102, the update log reflection serial number management table 110 reflects it from the update log storage table 108. The update log up to the serial number of the permitted serial number 1101 is reflected in the database 101. In the backup system 1, since the value of the reflection permission serial number 1101 is the last serial number of the received update log, all the received update logs are reflected. After the update log is reflected, the serial number of the reflected serial number 1102 in the update log reflection serial number management table 110 is updated (S504).

  The update log reception unit 207 of the backup system 2 stores the update log received from the update log transmission unit 005 in the update log storage table 208 in the same manner as the update log reception unit 107 of the backup system 1 (S508), and reflects the update log The value of the reflection permitted serial number 2101 in the serial number management table 210 is updated (S509), and the reflected serial number 2102 is returned. In the backup system 2, the reflection permission serial number 2101 is the reflection permission serial number managed in the update log reflection permission serial number management table 006 of the front system 0.

  The update log reflection unit 209 of the backup system 2 reflects the update log stored in the update log storage table 208 in the database 201 in the same procedure as the update log reflection unit 109 of the backup system 1 (S510).

  When the update log transmission unit 005 of the front system 0 receives the replies from the update log reception unit 107 of the backup system 1 and the update log reception unit 207 of the backup system 2, the reply source system updates the update log reflection permission serial number management table 006. In the case of the prioritized backup system 0063, the reflection permission serial number 0061 of the update log reflection permission serial number management table 006 is updated with the received reflected serial number (S506).

  The update log transmission means 005 of the front system 0 monitors the update log addition status in the update log storage table 004, and even if there is no newly added update log, the update log reflection permission serial number management table 006 is stored. If the reflection permission sequence number 0061 is larger than the notified reflection permission sequence number 0062, only the reflection permission sequence number is transmitted to each backup system (S502, S507). When the update log receiving unit 107 and the update log receiving unit 207 receive only the reflection permission sequence number, the update log reception unit 107 and the update log reception unit 207 perform only the process of updating the reflection permission sequence number in each update log reflection sequence number management table.

  As described above, as shown in FIG. 5, when an update log is reflected in the database 201 of the backup system 2, the update log is always reflected in the database 101 of the backup system 1. As a result, the replication progress status of the database 101 can always be preceded by the database 201.

  A process required when a failure occurs in the front system 0 in the above configuration and the backup system 1 operates as a new front system will be described. When operating as a new front system, functions equivalent to the application 002, the update log storage unit 003, the update log storage table 004, the update log transmission unit 005, and the update log reflection permission serial number management table 006 that existed in the front system 0 Is started respectively. Thereafter, the update log transmission means transmits the reflected serial number 1102 of the update log reflection serial number management table 110 as the reflection permission serial number. Thus, by proceeding with the update log reflection process of the backup system 2, the database 201 can be brought into the same state as the database 101 immediately before the new front system is operated. If the necessary update log is not stored in the update log storage table 208 of the backup system 2, it can be dealt with by retransmitting the necessary update log from the update log storage table 108 of the backup system 1. is there.

  In the above embodiment, when the backup system is operated as a new front system, it is possible to discard the unreflected update log and start the application immediately. On the other hand, in the following embodiment, it is possible to start an application after reflecting all unupdated update logs in the database.

  The received serial number of the update log receiving means 107 of the backup system 1 is used as the reflection permission serial number 0061 of the update log reflection permission serial number management table 006.

  In the case of the above embodiment, there is a possibility that the update log reflection processing to the database 201 of the backup system 2 precedes that of the backup system 1. However, the update log reflected in the database 201 of the backup system 2 is always stored in the update log storage table 108 of the backup system 1. Therefore, when operating as a new front system, the update progress stored in the database 101 can be preceded by the database 201 by reflecting all the update logs stored in the update log storage table 108 to the database 101. In the case of the above embodiment, the update log receiving means 107 of the backup system 2 uses the serial number that can store the update log in the update log storage table 108 as the reflection permission serial number. The delay of the replication process to the database 201 can be reduced.

  FIG. 6 is a block diagram showing a schematic configuration of a front-backup system according to another embodiment of the present invention. In the embodiment shown in FIG. 2, the communication between the backup system 1 and the backup system 2 when the front system 0 is operating normally is inactive. However, in the embodiment shown in FIG. Assume that communication between backup systems is active.

  In this embodiment, the backup system 1 includes an update log reflected serial number transmission unit 111, and the backup system 2 includes an update log reflected serial number reception unit 212. The update log reflected serial number transmission unit 111 of the backup system 1 acquires the value of the reflected serial number 1102 from the update log reflection serial number management table 110 and transmits it to the update log reflected serial number reception unit 212 of the backup system 2. The update log reflected serial number receiving unit 212 of the backup system 2 updates the reflection permitted serial number 2101 of the update log reflecting serial number management table 210 with the serial number received from the update log reflected serial number transmitting unit 111.

  In the case of the present embodiment, the following changes are required as compared with the embodiment shown in FIG.

  First, the update log reflection permission serial number management table 006 is not necessary. Next, the update log transmission unit 005 of the front system 0 does not transmit the reflection permission serial number. Further, the update log receiving means 107 and 207 do not refer to and update the update log reflection serial number management tables 110 and 210, and do not return the reflected serial number to the update log transmission means 005. Further, the value of the reflection permission serial number 1101 in the update log reflection serial number management table 110 is a special value that allows the update log reflection unit 109 to determine that all update logs may be reflected (for example, the serial number starts from 1). As a value such as −1).

  The present invention is not limited to the above-described embodiment, and can be implemented in various other forms without departing from the gist of the present invention. For this reason, the said embodiment is only a mere illustration in all points, and is not interpreted limitedly. For example, the above-described processing steps can be executed in any order or in parallel as long as there is no contradiction in the processing contents.

A part or all of the above embodiments can be described as in the following supplementary notes, but is not limited thereto.

  (Supplementary note 1) A front-backup system that includes a front system and first and second backup systems, and that replicates the database of the front system to the databases of the first and second backup systems. A front-backup system comprising means for managing the replication progress of the backup system and means for controlling the replication progress of the second backup system with reference to the progress.

  (Supplementary note 2) The managing means manages the first update log reflecting means for reflecting the update log in the database of the first backup system and the serial number of the reflected update log reflected in the database of the first backup system. A front-backup system according to appendix 1, characterized by comprising a management table.

  (Supplementary Note 3) The controlling means includes an update log transmitting means for transmitting the serial number of the reflected update log managed in the management table to the second backup system, and a database in the second backup system based on the serial number. And a second update log reflecting means for reflecting the update log. 2. The front-backup system according to claim 2, further comprising:

  (Appendix 4) When a failure occurs in the front system, the first backup system functions as a new front system, and the second backup system functions as a new priority backup system. A front-backup system as described.

  (Supplementary Note 5) A method of replicating a front system database to a first and second backup system database in a system including a front system and first and second backup systems, the first backup This is a method for managing the replication progress of the second backup system by managing the replication progress of the system and referring to the progress.

0 front system, 001 database, 002 application, 003 update log storage means, 004 update log storage table, 005 update log transmission means, 006 update log reflection permission serial number management table, 0061 reflection permission serial number, 0062 notified reflection permission serial number, 0063 Priority backup system, 1 (first) backup system, 2 (second) backup system, 101,201 database, 107,207 update log receiving means, 108,208 update log storage table, 109,209 update log reflecting means 110, 210 Update log reflection serial number management table, 1101, 2101 Reflection permission serial number, 1102, 2102 Reflected serial number, 111 Update log reflected serial number transmission means, 212 Update log reflected serial number reception means.

Claims (4)

A front-backup system comprising a front system and first and second backup systems, wherein the front system database replicates to the first and second backup system databases,
Means for managing the replication progress of the first backup system;
Means for referring to the progress status and controlling the progress status of replication of the second backup system;
Equipped with a,
The means for managing is
First update log reflecting means for reflecting an update log in the database of the first backup system;
A management table for managing a serial number of the reflected update log reflected in the database of the first backup system;
Ru with a front - backup system. The means for controlling is
Update log transmitting means for transmitting a serial number of the reflected update log managed in the management table to the second backup system;
Second update log reflecting means for reflecting an update log in the database of the second backup system based on the serial number;
The front-backup system according to claim 1 , comprising: When a failure in the front system is generated, the first backup system functions as a new front system, the second backup system, according to claim 1, characterized in that the function as a new priority backup system Or the front-backup system according to 2. In a system comprising a front system and first and second backup systems, a method of replicating the front system database to the first and second backup system databases,
A first backup system managing a replication progress status of the first backup system;
The front system, with reference to the progress, and a step of controlling the progress of replication of the second backup system,
The managing step includes
First update log reflecting means for reflecting the update log in the database of the first backup system;
A management table for managing a serial number of the reflected update log reflected in the database of the first backup system;
A method comprising:

Images