Consistency unit replication in application-defined systems

• US 20040193952A1
• Filed: 03/27/2003
• Published: 09/30/2004
• Est. Priority Date: 03/27/2003
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Parham, Jeffrey B., Singh, Ram P., Narayanan, Charumathy
• Application Number:
  US10/401,214
• Publication Number:
  US 20040193952A1
• Time in Patent Office: Days
• US Class Current: 714/13
• CPC Class Codes: G06F 16/273 Asynchronous replication or reconciliation

Abstract

Architecture for replicating a consistency unit in an application-defined system. Changes in a source replica effect metadata changes in both a change unit and the associated consistency unit. In response to a synchronization request by a destination, the source enumerates an updated consistency unit, enumerates all updated change units of the consistency unit, bundles the change units, and sends the bundles the change units for transmit to the destination. The destination applies the bundled change units in a single transaction after detecting and resolving conflicts. The process continues for each change.

Claims

1. A system that facilitates data replication, comprising:

a change tracking component that tracks metadata related to a replicated version of a consistency unit that spans across one or more homogenous collections of data; and

a reconciliation component that compares the metadata respectively, resolves conflicts using the metadata, and converges the replicated versions of the consistency unit.

2. The system of claim 1, the replicated versions of the consistency unit converged in a single transaction.

3. The system of claim 1, the metadata linked with a logical record relationship.

4. The system of claim 1, the conflicts detected in association with at least one of a logical record level, row level and column level.

5. The system of claim 4, the conflicts resolved at the logical record level by a comparison of logical record lineage metadata at a logical record level of the consistency unit and logical record lineage metadata of a logical record level of the replicated version, after which a winning logical record lineage metadata is selected according to a predetermined conflict policy.

6. The system of claim 4, the conflicts resolved at the row level by comparing respective logical record lineages of the consistency unit and the replicated version at the logical record level to detect a conflict, after which respective row lineages of the conflicting rows are compared to select a winner, the winner selected according to a predetermined conflict policy.

7. The system of claim 4, the conflicts resolved at the column level by comparing respective logical record lineages of the consistency unit and the replicated version at the logical record level to detect a conflict, after which respective row lineages of the conflicting rows are compared to detect a conflict, and in response to detecting the conflict at the row level, versions of the column data are compared to confirm the conflict exists, the conflict resolved by selecting a winning column according to a predetermined conflict policy.

8. The system of claim 4, a record at the row level including row lineage data and a record at the column level including column version data, such that conflict detection at the row level uses the row lineage data and conflict detection at the column level uses the column version data.

9. The system of claim 1, the reconciliation component resolving conflicts at a logical record level utilizing logical record metadata associated therewith, which logical record metadata includes logical record lineage data that tracks a change history for the consistency unit.

10. The system of claim 1, the reconciliation component resolving conflicts at a row level utilizing row metadata, which row metadata includes row lineage data that tracks change history of the row metadata, the history including information of at least one of the replica that updated the row and the version of the replica that updated the row.

11. The system of claim 1, the reconciliation component resolving conflicts at a column level utilizing column version metadata, which column version metadata tracks information of at least one of identity data of the replica that updated the column version and the column version.

12. The system of claim 1, wherein semantic relationships of the metadata of the consistency unit are preserved when converging of the consistency unit with the replicated version fails.

13. The system of claim 1, the reconciliation component converges data at a lower level when the conflict is detected at a higher level.

14. The system of claim 1, the conflicts resolved by the reconciliation component by selecting a winner logical record and a loser logical record in accordance with a predetermined conflict policy and overwriting the logical record lineage of the loser logical record with the logical record lineage of the winner logical record.

15. The system of claim 1, the changes in the metadata of the replicated versions tracked with the change tracking component such that the consistency unit can be converged.

16. The system of claim 1, the reconciliation component utilizing change enumeration to facilitate changes to a plurality of rows of the consistency unit, such changes including at least one of updates and, inserts and deletes.

17. The system of claim 16, the change enumeration utilizing view definitions to ensure enumeration according to a logical record link definition.

18. A computer including the system of claim 1.

19. A network of a plurality of clients and servers including the system of claim 1.

20. A method for facilitating data replication, comprising:

tracking metadata related to a replicated version of a consistency unit that spans across one or more homogenous collections of data;

comparing the metadata respectively;

resolving conflicts using the metadata; and

converging the replicated versions of the consistency unit.

21. The method of claim 20, the replicated versions of the consistency unit converged in a single transaction.

22. The method of claim 20, further comprising the step of linking the metadata with a logical record relationship.

23. The method of claim 20, further comprising the step of detecting the conflicts in association with at least one of a logical record level, row level and column level.

24. The method of claim 23 resolving conflicts at the logical record level by, comparing logical record lineage metadata at a logical record level of the consistency unit and logical record lineage metadata of a logical record level of the replicated version;

and

selecting a winning logical record lineage metadata according to a predetermined conflict policy.

25. The method of claim 23 resolving conflicts at the row level by further,

comparing respective logical record lineages of the consistency unit and the replicated version at the logical record level to detect a conflict;

and

comparing respective row lineages of the conflicting rows; and

selecting a winner according to a predetermined conflict policy.

26. The method of claim 23 resolving conflicts at the column level by further,

comparing respective logical record lineages of the consistency unit and the replicated version at the logical record level to detect a conflict;

comparing respective row lineages of the conflicting rows to detect a conflict;

comparing versions of the column data to confirm the conflict exists; and

selecting a winning column according to a predetermined conflict policy.

27. The method of claim 23, a record at the row level including row lineage data and a record at the column level including column version data, such that conflict detection at the row level uses the row lineage data and conflict detection at the column level uses the column version data.

28. The method of claim 20 resolving conflicts at a logical record level utilizing logical record metadata associated therewith, which logical record metadata includes logical record lineage data that tracks a change history for the consistency unit.

29. The method of claim 20 resolving conflicts at a row level utilizing row metadata, which row metadata includes row lineage data that tracks change history of the row metadata, the history including information of at least one of the replica that updated the row and the version of the replica that updated the row.

30. The method of claim 20 resolving conflicts at a column level utilizing column version metadata, which column version metadata tracks information of at least one of identity data of the replica that updated the column version and the column version.

31. The method of claim 20, wherein semantic relationships of the metadata of the consistency unit are preserved when converging of the consistency unit with the replicated version fails.

32. The method of claim 20 converging data at a lower lever when the conflict is detected at a higher level.

33. The method of claim 20 resolving conflicts by further,

selecting a winner logical record and a loser logical record in accordance with a predetermined conflict policy;

and

overwriting the logical record lineage of the loser logical record with the logical record lineage of the winner logical record.

34. The method of claim 20, further comprising the step of tracking changes in the metadata of the replicated versions with a change tracking component such that the consistency unit can be converged.

35. The method of claim 20, further comprising the step of enumerating changes to facilitate changing a plurality of rows of the consistency unit, such changes including at least one of updates and, inserts and deletes.

36. The method of claim 35, further comprising the step of providing view definitions to ensure enumeration according to a logical record link definition.

37. A method of facilitating data replication, comprising:

tracking semantically-related data changes between a source data collection and a first destination data collection of a homogenous collection of data;

linking metadata associated with the semantically-related data changes according to a logical record relationship to form a consistency unit; and

converging the consistency unit of data changes with a second destination data collection of the homogenous data collection in a single transaction.

38. The method of claim 37, further comprising:

detecting conflicts by comparing changes of the consistency unit with the second destination data collection; and

resolving the conflicts by selecting a winner according to a predetermined conflict policy.

39. The method of claim 38, the conflict detected in association with at least one of a logical record level, row level and column level.

40. The method of claim 39, the conflict detected and resolved at each of the logical record level, row level, and column level by first comparing respective logical record lineages of the source data collection and second destination data collection.

41. The method of claim 39, a record at the row level including row lineage data and a record at the column level including column version data, such that conflict detection at the row level uses the row lineage data and conflict detection at the column level uses the column version data.

42. The method of claim 40, wherein semantics of the semantically related changes are preserved when converging of the consistency unit is retried.

43. The method of claim 38, further comprising converging the data at a lower level when the conflict is detected at a higher level.

44. The method of claim 37, further comprising the step of detecting and resolving a conflict between the data consistency unit and the second destination data collection, the conflict resolved by selecting a winner logical record and a loser logical record in accordance with a predetermined conflict policy and overwriting the logical record lineage of the loser with the logical record lineage of the winner.

45. The method of claim 37, further comprising the step of tracking changes in the first destination data collection with a change tracking component, which change tracking component tracks replication metadata such that the consistency unit of data changes can be converged with the second destination data collection.

46. The method of claim 37, the data collection is a relational database.

47. A system for facilitating data replication, comprising:

means for identifying semantically-related data changes between a source data collection and a first destination data collection of a homogenous collection of data;

means for linking metadata associated with the semantically-related data changes according to a logical record relationship to form a consistency unit; and

means for converging the consistency unit of data changes with a second destination data collection of the homogenous data collection in a single transaction.

48. A system for facilitating data replication, comprising:

means for identifying metadata related to a replicated version of a consistency unit that spans across one or more homogenous collections of data;

means for comparing the metadata respectively;

means for resolving conflicts using the metadata; and

means for converging the replicated versions of the consistency unit.

49. A method of facilitating data replication, comprising:

identifying data changes between a source data collection and a first destination data collection;

creating a table of consistency unit change tracking information that is propagated to a second destination as a consistency unit; and

converging the consistency unit of data changes with a second destination data collection in a single transaction.

50. The method of claim 49, the table including at least one of a unique tag and a logical record link each defining an association between a change unit and a consistency unit.

51. The method of claim 50, further comprising the step of updating the table in response to writing the unique tag at the first destination data collection.

52. The method of claim 49, source data collection representative of a directory services architecture.