Method and apparatus for sequencing transactions globally in a distributed database cluster

• US 9,785,691 B2
• Filed: 05/09/2008
• Issued: 10/10/2017
• Est. Priority Date: 09/09/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: Open Invention Network LLC
• Inventors: Wong, Frankie, Yu, Xiong, Wang, Elaine
• Primary Examiner(s): Shmatov, Alexey
• Application Number:
  US12/149,927
• Publication Number:
  US 20090106323A1
• Time in Patent Office: 3,441 Days
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor

Abstract

A system and method for receiving and tracking a plurality of transactions and distributing the transactions to at least two replication queues over a network. The system and method comprise a global queue for storing a number of the received transactions in a first predetermined order. The system and method also comprise a sequencer coupled to the global queue for creating a copy of each of the transactions for each of said at least two replication queues and for distributing in a second predetermined order each said copy to each of said at least two replication queues respectively, said copy containing one or more of the received transactions.

Claims

1. A system configured to receive and track a plurality of committed transactions and to distribute the committed transactions to at least two replication queues over a network, the system comprising:

a first computer system comprising at least one controller, the at least one controller further comprising;

a global queue configured to store a number of the received committed transactions in a first predetermined order, the received committed transactions being received from an application server separate from the first computer system;

a sequencer coupled to the global queue configured to;

prepare a batch of transactions from the number of the received committed transactions;

remove from said batch of transactions a rolled back transaction; and

distribute, in a second predetermined order that is different from the first predetermined order, said batch of transactions to each of said at least two replication queues associated with at least two respective database controllers in an active-active configuration after said rolled back transaction has been removed from said batch, thereby mitigating update conflicts between databases,

the active-active configuration comprising the at least two respective database controllers that are respectively configured to process transactions while the other is processing transactions; and

an indoubt transaction queue in communication with the sequencer configured to receive, from the global queue, an indoubt transaction that has unknown status and that, when another transaction is received in the global queue, blocks the another transaction from being processed.

2. The system according to claim 1, wherein the sequencer is further configured to distribute said batch of transactions at a predetermined time interval.

3. The system according to claim 1, wherein the sequencer is further configured to distribute said batch of transactions when the number of the transactions within the global queue exceeds a predetermined value.

4. The system according to claim 1, wherein the sequencer is further configured to distribute said batch of transactions upon the earlier of:

a predetermined time interval; and

the number of the transactions within the global queue exceeds a predetermined value.

5. The system according to claim 4, wherein each of the transactions comprises an update transaction and a unique transaction id assigned by the sequencer.

6. The system according to claim 5, further comprising a global disk queue in communication with the global queue configured to receive and store the transactions, instead of in the global queue, in response to the global queue being above a global threshold.

7. The system according to claim 6, wherein each of said at least two replication queues have a corresponding replication disk queue configured to receive and store the transactions from the global queue, instead of in the replication queue, in response to the corresponding replication queue being above a replication threshold.

8. The system according to claim 7, wherein the global queue is further configured to receive the transactions from the global disk queue when the global disk queue is other than empty, and the global queue is further configured to receive the transactions from the application server when the global disk queue is empty.

9. The system according to claim 5, wherein the update transaction comprises at least one of a read, insert, update or delete request for at least one database in communication with at least one of said at least two replication queues.

10. The system according to claim 5, further comprising a resent transaction queue configured to store the transactions when the transactions repeated a request for the transaction id.

11. The system according to claim 1, wherein the global queue is a searchable first-in first-out pipe.

12. The system according to claim 11, wherein the sequencer is further configured to assure the order of transactions in the global queue remain consistent with their execution order at a database server coupled to at least one of the replication queues.

13. The system according to claim 11, wherein the global queue is further configured to store an indexed and randomly accessible data set.

14. The system according to claim 1, wherein the global queue and sequencer are hosted on a network entity selected from the group comprising:

a central control server and a peer-to-peer node.

15. A system configured to:

receive a plurality of transactions from at least one application server comprising a processor, distribute the transactions to at least two replication queues associated with at least two respective database controllers in an active-active configuration, and apply the transactions to a respective plurality of databases controlled by the at least two respective database controllers,

wherein the at least two respective database controllers in the active-active configuration are configured to process transactions while the other is processing transactions, the system being separate from the at least one application server and comprising;

a director coupled to each of said at least one application server configured to capture a plurality of database calls therefrom as the plurality of transactions; and

a controller configured to receive each of the plurality of transactions and further configured to;

store the transactions within a global queue in a predetermined order,

generate a batch of committed transactions from the plurality of transactions stored in the global queue for each of said at least two replication queues,

remove a rolled back transaction from said batch of committed transactions,

transmit, after removing the rolled back transaction, in the predetermined order said batch of committed transactions to each of said at least two replication queues, thereby mitigating update conflicts between databases, and

move an indoubt transaction from the global queue to an indoubt queue to allow a further transaction in the global queue to be processed.

16. The system according to claim 15, further comprising:

at least two replication servers comprising said at least two replication queues,

wherein each of said at least two replication servers is coupled to each of the databases via the at least two respective database controllers, and

wherein the director is further configured to route each of the transactions to one or more of the databases relative to a workload and transaction throughput.

17. The system according to claim 16, further comprising a backup controller configured to receive the transactions from said at least one application server upon failure of the controller, the backup controller comprising a backup global queue, wherein the backup global queue is synchronized with the controller and the backup global queue is a copy of the global queue.

18. The system according to claim 15, wherein the controller is further configured to transmit in the predetermined order said batch of committed transactions such that an order of the transactions in the global queue remains consistent with an execution order at one or more of the at least two respective database controllers coupled to at least one of the at least two replication queues.

19. The system according to claim 15, wherein the controller is further configured to store a resent transaction within a resent transaction queue based on a repeated request for a transaction ID.

20. A method for receiving and tracking a plurality of transactions from an application server and distributing the transactions to at least two replication queues over a network, the method comprising:

storing, by a transaction replicator server separate from the application server, a number of the received transactions in a first predetermined order in a global queue;

creating, by the transaction replicator server, a batch of committed transactions from the transactions for each of said at least two replication queues;

removing, by the transaction replication server, a rolled back transaction from said batch of committed transactions;

distributing, by the transaction replication server in response to the removing, in a second predetermined order said batch of committed transactions to each of said at least two replication queues, the at least two replication queues being associated with at least two respective database controllers in an active-active configuration respectively configured to process transactions while the other is processing transactions, thereby mitigating update conflicts between databases, wherein distributing said batch of committed transactions comprises assuring the order of transactions in the global queue remain consistent with their execution order at a database server coupled to at least one of the replication queues; and

moving an indoubt transaction from the global queue to an indoubt transaction queue during a database recovery process.

21. The method according to claim 20, wherein the distributing of said batch of committed transactions occurs at a predetermined time interval.

22. The method according to claim 20, wherein the distributing of said batch of committed transactions occurs when the number of the transactions within the global queue exceeds a predetermined number.

23. The method according to claim 20, wherein the distributing of said batch of committed transactions occurs upon the earlier of:

a predetermined time interval; and

the number of the transactions within the global queue exceeds a predetermined number.

24. The method according to claim 23, wherein each of the transactions comprises an update transaction and a unique transaction id assigned by a sequencer of the transaction replicator server.

25. The method according to claim 24, wherein the update transaction comprises at least one of a read, insert, update or delete request for at least one database in communication with at least one of said at least two replication queues via a respective one of the at least two respective database controllers.

26. The method according to claim 23, further comprising receiving and storing the transactions within a global disk queue when a global queue storage capacity of the global queue reaches a global threshold.

27. The method according to claim 26, further comprising:

determining whether the global disk queue is other than empty; and

receiving the transaction from the global disk queue rather than receiving the transactions from said application server when the global disk queue is other than empty.

28. The method according to claim 23, further comprising:

determining when at least one of said at least two replication queues are above a replication threshold, each of said at least two replication queues having a corresponding replication disk queue;

storing a number of the transactions within said corresponding replication disk queue based upon the determination; and

sending an alert to notify when said at least two replication queues and a corresponding replication disk queue capacity reach a preselected threshold.

29. The method according to claim 28, further comprising:

redirecting the transactions to at least one of said at least two replication queues being below said preselected threshold, based on receiving the alert.