Method and apparatus for reducing data locking time by removing a lock when journal data is written into a main memory journal queue

• US 5,386,554 A
• Filed: 05/16/1991
• Issued: 01/31/1995
• Est. Priority Date: 05/18/1990
• Status: Expired due to Term

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Nozaki, Masaharu
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Loomis, John C.
• Application Number:
  US07/700,862
• Time in Patent Office: 1,356 Days
• Field of Search: 395/600, 395/650, 395/700, 395/725, 371/12, 371/16.5
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 2201/82 Solving problems relating to consistency ensuring consistency in mirrored systems G06F11/2064
• CPC Class Codes: G06F 2201/825 the problem or solution involving locking
• CPC Class Codes: G06F 9/466 Transaction processing
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

A data management system, which is equipped with a data file and a journal file and adapted to reserve the exclusive occupation for data and then update the data, is characterized by releasing the exclusive occupation of data prior to writing of a journal to the journal file and of corresponding updated data to the data file. When data on a data base is updated, the exclusive occupation states of data to be updated are reserved first and then updating of the data is executed on a data buffer placed on a main memory and journals on the updating of the data are collected. The collected journals are entered into a first-in first-out journal queue in units of transactions. The exclusive occupation states of data on a corresponding transaction is released after corresponding journals have been stored in the journal queue. The journals stored in the journal queue are written to the journal file in units of transactions in the order in which they are entered into the queue. Subsequently updated data is written to the data file. The data in the data file is restored using the journal file when a system failure occurs in the data base.

Claims

1. An on line transaction processing method comprising the steps of:

requesting to update data on a data base having a plurality of data files which must not be inconsistent;

executing, in response to said requesting step, operations of reserving an exclusive occupation for a piece of data, updating the data and collecting a journal on updating of the data on a plurality of pieces of data, constituting one transaction, on said plurality of data files in sequence;

storing the journal collected in said executing step in a unit of transactions into a volatile memory;

releasing the exclusive occupation of the data related to the transaction in which the journal is stored into the volatile memory during said storing step;

writing the journals stored in said storage step to a non-volatile journal file in the unit of transactions after releasing of the exclusive occupation of the data; and

writing updated data to said data files for transactions whose journals have been written to said non-volatile journal file, and outputting a message on updated data.

2. A data management system comprising:

reserving means for reserving an exclusive occupation state for data;

updating means for updating data whose exclusive occupation is reserved and collecting a journal;

volatile memory means for storing said journal collected by said updating means;

releasing means for releasing the exclusive occupation of the data after the storage of said journal to said volatile memory means; and

means for writing said journal stored in said volatile memory means to a non-volatile journal file after the releasing of the exclusive occupation of the data by said releasing means.

3. A data management system according to claim 2, in which said reserving means and said updating means reserve exclusive occupation state of data and collect journals in sequence for a plurality of pieces of data constituting one transaction.

4. A data management system according to claim 2, in which said volatile memory means is a first-in first-out memory.

5. A data management system according to claim 2, in which said volatile memory means stores, in a group, journals on updating of a plurality of pieces of data constituting one transaction.

6. A data management system according to claim 2, in which said releasing means releases, in a group, the exclusive occupation states of a plurality of pieces of data corresponding to one transaction.

7. A data management system according to claim 2, in which said releasing means includes means for writing updated data to a data file after the writing of the journals to said non-volatile journal file.

8. A data management system according to claim 7, further comprising restoring means for restoring data in said data file using said non-volatile journal file when a system failure occurs in said data management system.

9. A data management system according to claim 8, in which said data writing means includes means for storing a clean log indicating the termination of the writing of data to said data file in said volatile memory means for a transaction in which the writing of updated data to said data file is terminated, said journal writing means writes the clean log stored in said volatile memory means to said non-volatile journal file, and said restoring means restores data on said data file only for updated data the writing of which to said data file is not completed in response to said clean log stored in said non-volatile journal file.

10. A data management system according to claim 2, further comprising:

a journal queue for receiving the journals,

wherein the releasing means releases the data file before any of the journals are stored in the non-volatile memory from the journal queue and before the data file is updated,

and wherein the journals are stored in the non-volatile memory in batches.

11. A data management system equipped with a data file and a non-volatile journal file and adapted to reserve exclusive occupation of data and then update the data, comprising:

access means for reserving exclusive occupation of data on which a request is made, executing updating of the requested data in a data buffer placed on a main storage device and collecting one or more journals;

releasing means for storing the journals collected by said access means in a volatile first-in first-out journal queue in a unit of each transaction and releasing the exclusive occupation of data after the completion of the storing of the journals in the volatile journal queue;

journal writing means for writing the journals stored in the volatile journal queue for each said transaction into the non-volatile journal file in the order in which the journals are stored in the volatile journal queue; and

data writing means for writing into the data file updated data of the data buffer into the data file for each said transaction whose journals have been written in to the non-volatile journal file.

12. A data management system according to claim 11, further comprising restoring means for restoring data in said data file to have consistency using said journal file when a fault occurs in said data management system.

13. A data management system according to claim 12, in which said data writing means includes means for entering a clean point log indicating the completion of the writing of data to said data file into said journal queue for a transaction whose updated data has been written to said data file;

and said journal writing means includes means for writing said clean point log entered into said journal queue to said journal file.

14. A data management system according to claim 13, in which said restoring means restors data on said data file on the basis of said clean point log stored in said journal file only for updated data the writing of which to said data file is not completed.

15. A data management system according to claim 11, in which said main memory is a volatile memory device, and said data file and said journal file are formed on external storage units each of which is formed of a nonvolatile memory device.

16. A data management system according to claim 15, in which said data file comprises a plurality of files which are related to each other and required to have consistency.

17. A data management system according to claim 11, further comprising inputting means for inputting a command to update data and outputting means for outputting a message on updated data, and in which said access means operates in response to said command from said inputting means, and said outputting means outputs said message for a transaction which occurs after the writing of data to said data file is completed by said data writing means.

18. A data management system according to claim 11, in which said access means, when data to be updated is not present on said data buffer, transfers the data from said data file to said data buffer and updates the data on said data buffer after reserving exclusive occupation of the data, and said data writing means writes updated data on said data buffer to said data file.

19. A data management system according to claim 11, further comprising means for fetching journals from said journal queue in the order in which they are entered into said queue and writing them to said journal file in the order in which they are fetched from said journal queue.

20. A data management system according to claim 11, further comprising:

a journal queue for receiving the journals,

wherein the releasing means releases the data file before any of the journals are stored in the non-volatile memory from the journal queue and before the data file is updated,

and wherein the journals are stored in the non-volatile memory in batches.