One-phase commit in a shared-nothing database system

• US 6,845,384 B2
• Filed: 11/21/2003
• Issued: 01/18/2005
• Est. Priority Date: 08/01/2003
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Pruscino, Angelo, Bamford, Roger J., Chandrasekaran, Sashikanth
• Primary Examiner(s): Mizrahi, Diane D.
• Application Number:
  US10/718,875
• Publication Number:
  US 20040148289A1
• Time in Patent Office: 424 Days
• Field of Search: 707/200, 707/10, 707/9, 707/205, 711/150, 712/203
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2379 Updates performed during online database operations; commit processing
• 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
• CPC Class Codes: Y10S 707/99956 File allocation

Abstract

Techniques are provided for handling distributed transactions in shared-nothing database systems where one or more of the nodes have access to a shared persistent storage. Rather than coordinate the distributed transaction using a two-phase commit protocol, the coordinator of the distributed transaction uses a one-phase commit protocol with those participants that have access to the transaction status information maintained by the coordinator. The transaction status information may reside, for example, in the redo log of the coordinator. In the case that the coordinator fails, those participants can determine the state of the distributed transaction based on information stored on the shared disk. In addition, the coordinator is able to determine whether it is possible to commit the distributed transaction based on information that is stored on the shared disk by the participants, without those participants entering a formal “prepared” state.

Claims

1. A method for performing a distributed transaction in a shared-nothing database system, the method comprising:

on a first shared-nothing node of said shared-nothing database system, causing a coordinator that is coordinating the distributed transaction to store information that indicates status of said distributed transaction on a persistent storage device;

wherein the persistent storage device is accessible to a participant that is to perform one or more operations as part of said distributed transaction;

wherein the participant resides on a second shared-nothing node of said shared-nothing database system; and

on the second shared-nothing node of said shared-nothing database system, causing the participant to determine the status of said distributed transaction by reading the information from the persistent storage device.

2. The method of claim 1 wherein:

the participant is a first participant of a plurality of participants in said distributed transaction;

the plurality of participants includes a second participant that does not have access to said persistent storage device; and

the method further comprises the step of the coordinator interacting with the second participant according to a two-phase commit protocol.

3. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 2.

4. The method of claim 1 further comprising the steps of:

the coordinator committing the distributed transaction;

after the coordinator commits the distributed transaction, the coordinator sending a commit message to the participant; and

preventing the information that indicates the status of the distributed transaction from being overwritten or deleted until a set of conditions is satisfied, wherein one condition in said set of conditions is that the coordinator receives a commit acknowledge message from said participant.

5. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 4.

6. The method of claim 1 further comprising the steps of:

the participant sending a first piece of information to the coordinator, wherein the first piece of information is associated with work performed by said participant as part of said distributed transaction; and

the coordinator performing a comparison between the first piece of information and information associated with a redo log of said second shared-nothing node; and

the coordinator determining whether to commit the transaction based, at least in part, on said comparison.

7. The method of claim 6 wherein piece of information includes a log-sequence-number of the latest change made by the participant as part of the distributed transaction.

8. The method of claim 7 wherein the step of sending includes the steps of:

the participant identifying a message that is being sent to said first shared-nothing node for a purpose unrelated to the distributed transaction; and

piggybacking the log-sequence number on said message.

9. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 8.

10. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 7.

11. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 6.

12. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 1.

13. A method for performing a distributed transaction in a shared-nothing database system, the method comprising:

assigning a participant to perform one or more operations as part of said distributed transaction;

wherein the participant resides on a first shared-nothing node of said shared-nothing system;

causing said participant to store, on a persistent storage device, status information that indicates changes made by the participant during performance of said one or more operations;

wherein the persistent storage device is accessible to a coordinator that is responsible for coordinating said distributed transaction;

wherein the coordinator resides on a second shared-nothing node of said shared-nothing database system;

on said second shared-nothing node of said shared-nothing database system, causing said coordinator to determine, based on the status information on said persistent storage device, whether the participant has written to persistent storage changes produced by performance of the one or more operations; and

the coordinator process determining whether the distributed transaction can be committed based, at least in part, on whether the participant has written to persistent storage changes produced by performance of the one or more operations.

14. The method of claim 13 wherein:

the step of causing said participant to store, on a persistent storage device, status information that indicates changes made by the participant during performance of said one or more operations includes

causing said participant to store redo information in a redo log on said persistent storage device; and

the step of causing said coordinator to determine, based on the status information on said persistent storage device, whether the participant has written to persistent storage changes produced by performance of the one or more operations includes

inspecting the redo log of the participant to determine whether the redo information for said changes have been written to said persistent storage.

15. The method of claim 13 wherein:

the participant is a first participant of a plurality of participants in said distributed transaction;

the plurality of participants includes a second participant that stores status information on a second persistent storage device that is not accessible by said coordinator; and

the method further comprises the step of the coordinator interacting with the second participant according to a two-phase commit protocol.

16. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 14.

17. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 15.

18. The method of claim 13 wherein:

the information on said persistent storage device indicates that the participant has not written to persistent storage changes produced by performance of the one or more operations; and

the method further comprises the coordinator sending a force redo message to the participant to cause the participant to write to persistent storage the changes produced by performance of the one or more operations.

19. The method of claim 18 wherein the step of sending a force redo message includes the steps of:

identifying a message that is being sent to said first shared-nothing node for a purpose unrelated to the distributed transaction; and

piggybacking the force redo message on said message.

20. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 19.

21. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 18.

22. A computer-readable medium carrying one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform the method recited in claim 13.