System and method for refreshing a table using epochs

• US 7,818,297 B2
• Filed: 03/31/2003
• Issued: 10/19/2010
• Est. Priority Date: 03/31/2003
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Peleg, Nitzan, Sherman, Yuval, Zeller, Hansjorg
• Primary Examiner(s): STACE, BRENT S
• Application Number:
  US10/403,500
• Publication Number:
  US 20040193622A1
• Time in Patent Office: 2,759 Days
• Field of Search: 707/2, 707/8, 707/201, 707/202, 707/625, 707/624, 707/610, 707/609, 707/999.002, 707/999.008, 707/999.201, 707/999.202
• US Class Current: 707/625
• 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: G06F 16/275 Synchronous replication

Abstract

The disclosed embodiments relate to a system for refreshing a table to produce a refreshed table. The system may include a refresh log that may contain a plurality of entries that each comprise an epoch identifier. The system may also include a refresh manager that may be adapted to perform a refresh operation on the table using entries that have corresponding epoch identifiers.

Claims

1. A system for refreshing a table to produce a refreshed table, the system comprising:

a refresh log that contains a plurality of entries, each of the plurality of entries comprising an epoch identifier, wherein the epoch identifier identifies a subset of the plurality of entries that have been added to the refresh log since a previous refresh operation on the table, wherein the epoch identifier is adapted to synchronize refresh operations applied to network-distributed elements of the table with refresh operations applied to the refresh log; and

a refresh manager that performs a refresh operation on the table using entries that have corresponding epoch identifiers, wherein the refresh manager is configured to increment the epoch identifier before performing the refresh operation.

2. The system for refreshing the table set forth in claim 1, wherein the corresponding epoch identifiers represent epoch numbers that have been created since the previous refresh operation on the table.

3. The system for refreshing the table set forth in claim 1, wherein the refresh manager changes the epoch identifier in association with a refresh operation.

4. The system for refreshing the table set forth in claim 1, wherein the entries relate to updates performed on a base table on which the refreshed table is based.

5. The system for refreshing the table set forth in claim 4, wherein the base table and the refresh log are co-located on common partitions.

6. The system for refreshing the table set forth in claim 4, wherein the refresh manager locks the base table and changes the epoch identifier in association with a refresh operation.

7. The system for refreshing the table set forth in claim 1, wherein the epoch identifier is stored as metadata for a database.

8. The system for refreshing the table set forth in claim 1, wherein the epoch identifier is visible as runtime information for a base table upon which the refreshed table is based.

9. The system for refreshing the table set forth in claim 1, wherein the refreshed table is a materialized view.

10. The system for refreshing the table set forth in claim 1, wherein the table comprises at least a portion of a distributed, shared-nothing database.

11. A tangible machine readable medium storing information, the tangible medium comprising:

a base table;

a refresh log that contains a plurality of entries, each of the plurality of entries corresponding to an update that was made to the base table; and

code adapted to perform the updates to the base table by a process that also places entries into the refresh log, wherein each of the plurality of entries comprises an epoch identifier that identifies a subset of the plurality of time entries that have been added to the refresh log since a previous refresh operation on the table, wherein each of the plurality of entries in the subset has the same epoch identifier, and wherein the epoch identifier is adapted to synchronize refresh operations applied to network-distributed elements of the table with refresh operations applied to the refresh log.

12. The tangible medium set forth in claim 11, wherein the refresh log is created because the base table has a materialized view defined thereon.

13. The tangible medium set forth in claim 11, wherein the base table comprises at least a portion of a distributed, shared-nothing database.

14. A computer network, comprising:

a plurality of computer systems coupled for communication with each other, each of the plurality of computer systems comprising a partition that stores a portion of a base table and a refresh log, the refresh log containing a plurality of entries, each of the plurality of entries comprising an epoch identifier, wherein the epoch identifier identifies a subset of the plurality of entries that have been added to the refresh log since a previous refresh operation on the base table, and wherein the epoch identifier is adapted to synchronize refresh operations applied to network-distributed elements of the table with refresh operations applied to the refresh log; and

a refresh manager that performs a refresh operation on the base table to provide a refreshed base table using entries that have corresponding epoch identifiers, wherein the refresh manager is configured to increment the epoch identifier before performing the refresh operation.

15. The computer network set forth in claim 14, wherein the corresponding epoch identifiers represent epoch numbers that have been created since the previous refresh operation on the table.

16. The computer network set forth in claim 15, wherein the refresh manager changes the epoch identifier in association with a refresh operation.

17. The computer network set forth in claim 15, wherein the entries relate to updates performed on the portion of the base table stored on the same partition as the refresh log that contains the refresh information.

18. The computer network set forth in claim 15, wherein the refresh manager locks the base table and changes the epoch identifier in association with a refresh operation.

19. The computer network set forth in claim 15, wherein the epoch identifier is stored as metadata for a database.

20. The computer network set forth in claim 15, wherein the epoch identifier is visible as runtime information for the base table.

21. The computer network set forth in claim 15, wherein the refreshed table is a materialized view.

22. The computer network set forth in claim 15, wherein the base table comprises at least a portion of a distributed, shared-nothing database.

23. A method of updating a table to produce a refreshed table, the method comprising:

locking a base table that contains information on which the table is based;

incrementing an epoch identifier associated with the base table from a first value to a second value while the base table is locked;

unlocking the base table;

storing the epoch identifier as part of an entry in a refresh log, wherein the epoch identifier identifies the entry as one of a plurality of entries that have been added to the refresh log since a previous refresh operation on the table, and wherein the epoch identifier is adapted to synchronize refresh operations applied to network-distributed elements of the table with refresh operations applied to the refresh log; and

refreshing the table to produce the refreshed table using the entries in the refresh log that have epoch identifiers of the first value.

24. The method of updating the table set forth in claim 23, comprising creating the refreshed table using data corresponding to epoch numbers that have been created since the previous refresh operation on the table.

25. The method set forth in claim 23, comprising storing the base table and the refresh log on a common partition.

26. The method set forth in claim 23, comprising creating a materialized view based on the table.

27. The method set forth in claim 23, comprising changing the epoch identifier in association with a refresh operation.

28. The method set forth in claim 23, comprising locking the base table and changing the epoch identifier in association with a refresh operation.

29. The method set forth in claim 23, comprising storing the epoch identifier as metadata for a database.

30. The method set forth in claim 23, comprising making the epoch identifier visible as runtime information for the base table.