Method and mechanism for efficient implementation of ordered records

• US 7,039,773 B2
• Filed: 04/29/2003
• Issued: 05/02/2006
• Est. Priority Date: 04/29/2003
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Loaiza, Juan R., Srihari, Vinay, Bamford, Roger J., Pareek, Alok, Joshi, Vikram, Nithrakashyap, Arvind, Hu, Wei Ming, Bosman, Tudor
• Primary Examiner(s): MOAZZAMI, NASSER G
• Application Number:
  US10/426,471
• Publication Number:
  US 20040221116A1
• Time in Patent Office: 1,099 Days
• Field of Search: 711/158, 711161-162, 707200-204, 714 15- 16, 714 18- 20
• US Class Current: 711/158
• CPC Class Codes: G06F 16/284 Relational databases
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

An improved method, mechanism, and system for implementing, generating, and maintaining records, such as redo records and redo logs in a database system, are disclosed. Multiple sets of records may be created and combined into a partially ordered (or non-ordered) group of records, which are later collectively ordered or sorted as needed to create an fully ordered set of records. With respect to a database system, redo generation bottleneck is minimized by providing multiple in-memory redo buffers that are available to hold redo records generated by multiple threads of execution. When the in-memory redo buffers are written to a persistent storage medium, no specific ordering needs to be specified with respect to the redo records from the different in-memory redo buffers. While the collective group of records may not be ordered, the written-out redo records may be partially ordered based upon the ordered redo records from within individual in-memory redo buffers. At recovery, ordering and/or merging of redo records may occur to satisfy database consistency requirements.

Claims

1. A method for implementing records in a computing system, comprising:

establishing two or more in-memory buffers;

distributing records across the two or more in-memory buffers;

identifying a cross-sectional interval for the two or more in-memory buffers, the cross-sectional interval based upon a temporal period, the cross-section intervals corresponding to a subset of zero or more records for each of the two or more in-memory buffers, the subset of zero or more records for each of the two or more in-memory buffers being temporally ordered; and

writing the subset of zero or more records for each of the two or more in-memory buffers to the persistent storage medium, wherein records written to the persistent storage medium are not sequentially ordered as an entire group.

2. The method of claim 1 in which the records written to the persistent storage medium are later processed.

3. The method of claim 2 wherein the records written to the persistent storage medium are converted into sequentially ordered records during the later processing.

4. The method of claim 2 wherein database recovery is performed during the later processing.

5. The method of claim 1 in which the two or more log buffers comprise two or more in-memory redo buffers and the records comprise redo records.

6. The method of claim 1 in which the records distributed across the two or more in-memory buffers are ordered within each in-memory buffer.

7. The method of claim 1 in which the records written to the persistent storage medium are partially ordered.

8. The method of claim 7 in which sequentially-ordered records are generated by merging the partially ordered records written to the persistent storage medium.

9. The method of claim 1 in which the records written to the persistent storage medium do not have pre-allocated locations for individual records.

10. The method of claim 1 in which a system checkpoint is synchronized to the temporal period of the cross-sectional interval.

11. The method of claim 10 in which later processing against the records written to the persistent storage medium is performed against a cross-sectional interval following the system checkpoint.

12. The method of claim 11 in which database recovery is performed during the later processing.

13. The method of claim 1 in which the records are distributed across the two or more in-memory buffers to balance load across the two or more in-memory buffers.

14. The method of claim 1 in which the records comprise log information describing changes for remote storage devices.

15. A method for implementing records in a computing system, comprising:

establishing multiple buffers in the computing system;

distributing records across the multiple buffers;

identifying an interval within the multiple buffers to write the records to a persistent storage medium; and

writing the records to the persistent storage medium, wherein the records written to the persistent storage medium are not sequentially ordered as an entire group.

16. The method of claim 15 in which the computing system comprises a database system, wherein the records comprise redo records, the multiple buffers comprise multiple redo buffers, and the interval comprises a log-write interval.

17. The method of claim 15 in which the records written to the persistent storage medium comprises partial ordering, the partial ordering comprising an ordering for each subset of records corresponding to the interval of each of the multiple buffers.

18. The method of claim 15 in which the interval is temporally based.

19. The method of claim 18 in which a record corresponds to the interval if the record is associated with a timestamp that falls within the boundaries of the interval.

20. The method of claim 15 in which the records written to the persistent storage medium do not have pre-allocated locations for individual records.

21. The method of claim 15 in which a system checkpoint is synchronized against the boundaries of the interval.

22. The method of claim 15 further comprising:

establishing a boundary of the interval; and

determining if a pending operation will result in an individual record beyond the boundary of the interval.

23. The method of claim 22 in which the boundary of the interval is published as a visible beacon.

24. The method of claim 22 in which the interval is delayed for completion of the pending operation.

25. The method of claim 22 in which the pending operation is synchronized to a later interval.

26. The method of claim 15 in which the records written to the persistent storage medium are later processed.

27. The method of claim 26 wherein the records written to the persistent storage medium are converted into sequentially ordered records during the later processing.

28. The method of claim 26 wherein database recovery is performed during the later processing.

29. A method for applying ordered records, in which one or more intervals of records have been persistently stored to a persistent storage medium, each of the one or more intervals of records comprising a self-contained set of records for a given interval, comprising:

(a) identifying a checkpoint;

(b) identifying an interval of records corresponding to the checkpoint;

(c) creating an ordered group of records by ordering records within the interval of records;

(d) applying the ordered group of records;

(e) determining whether an additional interval of records are to be applied; and

(f) repeating (c) through (e) until no further intervals of records are to be applied.

30. The method of claim 29 in which the ordered records are applied to a database system, wherein the records comprise redo records.

31. The method of claim 29 in which (c) comprises merging or sorting the records within the first interval of records.

32. The method of claim 29 in which the checkpoint is synchronized to a boundary of the one or more intervals of records.

33. The method of claim 29 in which the checkpoint is not synchronized to a boundary of the one or more intervals of records, and wherein any records corresponding to changes already within the database at the checkpoint are discarded.

34. The method of claim 29 in which the interval of records and the additional interval of records are applied in order.

35. The method of claim 29 in which a determination is made whether the ordered group of records is missing any individual records.

36. The method of claim 35 in which the ordered group is discarded without being applied.

37. The method of claim 35 in which a determination is made whether the missing individual records will cause an error, and the ordered group is applied if no error will result.

38. A structure in a computing system for implementing records on a persistent storage medium, comprising an on-disk log comprising two or more intervals of records, each of the two or more intervals of records comprising a self-contained set of records for a given interval, wherein each of the two or more intervals of records comprises a collection of unordered records.

39. The structure of claim 38 in which a log-write header separates each of the two or more intervals of records.

40. The structure of claim 38 in which each of the two or more intervals of records comprises a partially ordered collection of ordered subsets of records.

41. The structure of claim 40 in which a strand header separates each ordered subset of records.

42. The structure of claim 40 in which each ordered subset of records corresponds to a specific log in memory.

43. The structure of claim 38 in which the computing system comprises a database system, wherein the records comprise redo records and the on-disk log comprises an on-disk redo log.

44. A computer program product comprising a tangible computer usable medium having executable code to execute a process for implementing records in a computing system, the process comprising the steps of:

establishing two or more in-memory buffers;

distributing records across the two or more in-memory buffers;

identifying a cross-sectional interval for the two or more in-memory buffers, the cross-sectional interval based upon a temporal period, the cross-section intervals corresponding to a subset of zero or more records for each of the two or more in-memory buffers, the subset of zero or more records for each of the two or more in-memory buffers being temporally ordered; and

writing the subset of zero or more records for each of the two or more in-memory buffers to the persistent storage medium, wherein records written to the persistent storage medium are not sequentially ordered as an entire group.

45. A system for implementing records in a computing system, comprising:

means for establishing two or more in-memory buffers;

means for distributing records across the two or more in-memory buffers;

means for identifying a cross-sectional interval for the two or more in-memory buffers, the cross-sectional interval based upon a temporal period, the cross-section intervals corresponding to a subset of zero or more records for each of the two or more in-memory buffers, the subset of zero or more records for each of the two or more in-memory buffers being temporally ordered; and

means for writing the subset of zero or more records for each of the two or more in-memory buffers to the persistent storage medium, wherein records written to the persistent storage medium are not sequentially ordered as an entire group.

46. A computer program product comprising a tangible computer usable medium having executable code to execute a process for implementing records in a computing system, the process comprising the steps of:

establishing multiple buffers in the computing system;

distributing records across the multiple buffers;

identifying an interval within the multiple buffers to write the records to a persistent storage medium; and

writing the records to the persistent storage medium, wherein the records written to the persistent storage medium are not ordered as an entire group.

47. A system for implementing records in a computing system, comprising:

means for establishing multiple buffers in the computing system;

means for distributing records across the multiple buffers;

means for identifying an interval within the multiple buffers to write the records to a persistent storage medium; and

means for writing the records to the persistent storage medium, wherein records written to the persistent storage medium are not sequentially ordered as an entire group.

48. A computer program product comprising a tangible computer usable medium having executable code to execute a process for applying ordered records in a computing system, in which one or more intervals of records have been persistently stored to a persistent storage medium, each of the one or more intervals of records comprising a self-contained set of records for a given interval, the process comprising:

(a) identifying a checkpoint;

(b) identifying an interval of records corresponding to the checkpoint;

(c) creating an ordered group of records by ordering records within the interval of records;

(d) applying the ordered group of records;

(e) determining whether an additional interval of records are to be applied; and

(f) repeating (c) through (e) until no further intervals of records are to be applied.

49. A system for applying ordered records in a computing system, in which one or more intervals of records have been persistently stored to a persistent storage medium, each of the one or more intervals of records comprising a self-contained set of records for a given interval, comprising:

(a) means for identifying a checkpoint;

(b) means for identifying an interval of records corresponding to the checkpoint;

(c) means for creating an ordered group of records by ordering records within the interval of records;

(d) means for applying the ordered group of records;

(e) means for determining whether an additional interval of records are to be applied; and

(f) means for utilizing (c) through (e) until no further intervals of records are to be applied.

50. A method for implementing redo records in a database, comprising:

establishing multiple redo buffers in a database system;

distributing redo records across the multiple redo buffers;

identifying a log-write interval within the multiple redo buffers to write the redo records to a persistent storage medium; and

writing the redo records to the persistent storage medium, wherein the redo records written to the persistent storage medium are not ordered as an entire group.

51. The method of claim 50 in which the redo records written to the persistent storage medium comprises partial ordering, the partial ordering comprising an ordering for each subset of redo records corresponding to the log-write interval of each of the multiple redo buffers.

52. The method of claim 50 in which the redo records written to the persistent storage medium do not have pre-allocated locations for individual redo records.

53. The method of claim 50 in which a system checkpoint is synchronized against the boundaries of the log-write interval.

54. The method of claim 50 further comprising:

establishing a boundary of the log-write interval; and

determining if a pending operation will result in an individual redo record beyond the boundary of the log-write interval.

55. The method of claim 54 in which the boundary of the log-write interval is published as a visible beacon.

56. The method of claim 54 in which the log-write interval is delayed for completion of the pending operation.

57. The method of claim 54 in which the pending operation is synchronized to a later log-write interval.

58. The method of claim 50 in which the records written to the persistent storage medium are later processed.

59. The method of claim 58 wherein the records written to the persistent storage medium are converted into sequentially ordered records during the later processing.

60. The method of claim 58 wherein database recovery is performed during the later processing.