Fault-tolerant transaction-oriented data processing

• US 5,465,328 A
• Filed: 01/14/1994
• Issued: 11/07/1995
• Est. Priority Date: 03/30/1993
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Mohan, Chandrasekaran, Dievendorff, Richard
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Snyder, Glenn
• Application Number:
  US08/181,521
• Time in Patent Office: 662 Days
• Field of Search: 395/575, 395/600, 395/650, 371/12, 364/282.4, 364/282.1, 364/281.3
• US Class Current: 714/15
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 2201/87 Monitoring of transactions
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

In transaction processing systems, it is known for resource-updating operations within a transaction to be backed out at the request of an application program following detection of error conditions during processing of the transaction. If the error condition is very likely to recur, it may be undesirable for the operations request to be presented to the application exactly as before. A transaction-oriented data processing system and a method of transaction-oriented data processing are provided in which operation requests or data packets may be marked to be excluded from the effects of application-requested backouts.

Claims

1. A method of fault-tolerant transaction-oriented data processing in which resource updates performed by a first application program are made permanent and visible to other application programs only on successful completion of the transaction, and are backed out following discovery of an error condition prior to successful completion, comprising the steps of:

initiating execution of a first unit of work, said first unit of work including one or more operations initiated by said first application program;

specifying whether operations within said first unit of work are to be excluded from the effects of application-requested backouts following detection of error conditions within said first unit of work;

responsive to detection of an error condition by said first application program, backing out resource updates performed in said first unit of work whilst ensuring that excluded operations are not made available to other application programs and are not backed out by said backing out step;

establishing a unit of work which includes said excluded operations.

2. A method according to claim 1, wherein:

said ensuring step includes the step of acquiring request duration locks on said excluded operations to prevent said excluded operations from being made available to other application programs.

3. A method according to claim 2, further including the steps of:

performing said established unit of work;

releasing said request-duration locks after said performing step.

4. A method according to claim 1 wherein said excluded operations include a transaction-initiating operation.

5. A method according to claim 1, further including the steps of:

marking excludable operations;

storing a list of pointers to said excluded operations; and

examining said stored list during said ensuring step to locate said excluded operations.

6. A method according to claim 1 in which inter-program communication between application programs is carried out using message queuing and in which transaction requests comprise messages sent to queues, wherein committing said established unit of work causes a faulty message which raised an error to be deleted from the message queue.

7. A method according to claim 1 wherein said ensuring step includes the step of acquiring request duration locks on said excluded operations to prevent said excluded operations from being made available to other application programs.

8. A method according to claim 7, further including the steps of:

performing said established unit of work;

releasing said request-duration locks after said performing step.

9. A fault-tolerant transaction-oriented data processing system including an inter-program communications manager providing communications services between application programs, in which resource updates performed by a first application program are made permanent and visible to other application programs only on successful completion of the transaction, and are backed out following discovery of an error condition prior to successful completion, the system comprising:

means for specifying, within an instruction issued by said first application program, whether an operation within a first unit of work is to be excluded from the effects of application-requested backouts following detection of error conditions within said first unit of work;

means within said inter-program communications controller, responsive to detection of an error condition by said first application program, for backing out resource updates performed in said first unit of work;

means for ensuring that any excluded operation is not made available to other application programs and is not backed out by said backing out means; and

means within said inter-program communications controller for establishing a unit of work which includes said excluded operations.

10. The system of claim 9, further including:

means for marking excludable operations;

means for storing a list of pointers to said excluded operations;

means within said ensuring means for examining said list to locate said excluded operations.