Mirroring data changes in a database system

• US 8,655,847 B2
• Filed: 08/16/2010
• Issued: 02/18/2014
• Est. Priority Date: 08/16/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Kamra, Vaibhav, Madhavarapu, Pradeep, Diaconu, Cristian, Porwal, Priyank, Fang, Jun, Acharya, Srini, Chirica, Alexandru
• Primary Examiner(s): Wu, Yicun
• Application Number:
  US12/856,688
• Publication Number:
  US 20120041928A1
• Time in Patent Office: 1,282 Days
• Field of Search: 707/659
• US Class Current: 707/659
• 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 16/273 Asynchronous replication or reconciliation
• CPC Class Codes: G06F 16/275 Synchronous replication

Abstract

A method for mirroring one or more data changes from a first database server to one or more second database servers. The method includes intercepting one or more operations that make one or more changes to one or more data on the first database server. After intercepting these operations, the method merges the operations into one or more chunks, adds one or more log records having the chunks into a database log. The database log includes one or more database log blocks. The method then writes the database log blocks.

Claims

1. A method for mirroring one or more data changes from a first database server to one or more second database servers, comprising:

by the first database server, intercepting two or more operations that each make a change to data of a particular file stored on the first database server, wherein one of the operations is a filesystem command intercepted from a file system stack and one of the operations is a database command intercepted from a database stack;

by the first database server, detecting an offset location and length of each data change;

by the first database server, analyzing the offset location and length of each data change to the particular file to merge the data changes of the filesystem command operation and the database command operation into a single data change having a merged offset location and merged length, the single data change comprising a chunk;

by the first database server, adding the chunk into a database log, the database log having one or more database log blocks;

storing metadata associated with the particular file in a database;

adding information related to a change to the metadata to a database log block of the database log; and

writing the database log blocks by the first database server.

2. The method of claim 1, wherein the operations are intercepted by a filter driver.

3. The method of claim 2, wherein the filter driver is a kernel mode driver configured to detect the changes.

4. The method of claim 1, wherein the operation from the file system stack comprises one of open, write or close operation on the data.

5. The method of claim 1, wherein the change to the metadata comprises an insert or an update operation on the metadata.

6. The method of claim 1, wherein intercepting the operations comprises:

detecting the operations that make the changes; and

sending the offset location and the length for each change to a chunker.

7. The method of claim 6, wherein the data changes are merged by the chunker.

8. The method of claim 1, wherein merging the data changes comprises combining a subset of the operations, wherein the subset of the operations comprises two or more data changes, wherein each data change in the subset occurs overlapping data.

9. The method of claim 1, wherein merging the data changes comprises combining the operations, wherein the operations comprise data changes occurring in adjacent data.

10. The method of claim 1, wherein the data changes are merged after a predetermined amount of time.

11. The method of claim 1, wherein the data changes are merged for a predetermined amount of data.

12. The method of claim 1, further comprising sending the database log blocks to the second database servers.

13. The method of claim 12, wherein sending the database log blocks comprises:

(a) determining whether one of the database log blocks includes one of the log records having the chunks; and

(b) if the one of the database log blocks includes the one of the log records, then sending data that corresponds to the one of the log records to the second database servers.

14. The method of claim 13, further comprising repeating steps (a)—

(b) for each database log block.

15. A database management system, comprising:

a filesystem filter driver configured to intercept one or more filesystem operations that make one or more changes to one or more data on a first database server, wherein the filesystem operations comprise file system commands intercepted from a file system stack;

a filestream manager configured to intercept one or more database operations that make one or more changes to the data, wherein the database operations comprise database commands intercepted from a database stack;

a chunker of the first database server configured to;

merge the filesystem operations and the database operations into a single data change comprising a chunk;

add one or more log records having the chunk into a database log; and

write one or more database log blocks of the database log, wherein the chunker is configured to;

analyze the information to determine an offset location and a length for each change made by each filesystem operation and each database operation; and

merge two or more offset locations and two or more lengths into a single data change having a merged offset location and merged length, wherein the two or more offset locations and the two or more lengths are adjacent to or overlapping with each other;

a log reader configured to;

read the database log blocks;

determine whether one of the database log blocks in the database log includes one of the log records having the chunk; and

if the one of the database log blocks includes the one of the log records, then send data that corresponds to the one of the log records to one or more second database management systems.

16. The database management system of claim 15, further comprising:

one or more database components configured to;

receive one or more notifications pertaining to the intercepted filesystem operations; and

send the notifications to the chunker, wherein the notifications comprise information related to the filesystem operations and the database operations.