Creating point-in-time copies of file maps for multiple versions of a production file to preserve file map allocations for the production file

• US 8,620,973 B1
• Filed: 09/25/2012
• Issued: 12/31/2013
• Est. Priority Date: 09/25/2012
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Veeraswamy, Sairam, Bixby, Peter C., Madany, Peter W.
• Primary Examiner(s): AL HASHEMI, SANA A
• Application Number:
  US13/626,741
• Time in Patent Office: 462 Days
• Field of Search: 707/639, 707/673, 707/705, 707/711, 707/822, 707/825, 707/829
• US Class Current: 707/822
• CPC Class Codes: G06F 16/128 Details of file system snapshots on the file-level, e.g. snapshot creation, administration, deletion use of snapshots for error detection or correction G06F11/14, G06F11/16

Abstract

To preserve the contiguity of file map extents of a production file when making a series of snapshot copies, the snapshots share indirect blocks of the production file through virtual pointers. When a write to a data block of the production file is first done since the time of the most recent snapshot so that the most recent snapshot can no longer share one of the contiguous indirect blocks, a new indirect block is allocated to store the file mapping metadata for the most recent snapshot, and a virtual pointer for the snapshots is changed to point to this new indirect block. Therefore the change in the virtual pointer changes the file mapping metadata for any number of snapshots sharing the new indirect block so that the method is scalable.

Claims

1. A method of operating a file server, the file server having data storage storing a production file, a data processor coupled to the data storage for access to the production file, and non-transitory computer readable storage medium coupled to the data processor and storing computer instructions, the production file having data blocks and indirect blocks that are allocated to the production file and contain pointers to the data blocks for mapping block offsets in a logical extent of the production file to corresponding ones of the data blocks, and the computer instructions, when executed by the data processor, create a series of snapshot copies of the production file while preserving the allocations of the indirect blocks to the production file by performing the steps of:

(a) creating a first snapshot copy of the production file by allocating a first inode to the first snapshot copy and building a first tree of pointers extending from the first inode to a first group of the indirect blocks of the production file; and

then

(b) creating a second snapshot copy of the production file by allocating a second inode to the second snapshot copy and building a second tree of pointers extending from the second inode to a second group of the indirect blocks of the production file, wherein the first tree of pointers and the second tree of pointers share pointers to indirect blocks that are included in a third group of indirect blocks that are included in both the first group of the indirect blocks of the production file and the second group of the indirect blocks of the production file, and then

(c) in response to a request from a client of the file server to write a block of new data to a specified block offset in the logical extent of the production file, writing the block of new data to the data storage, searching one of the first tree of pointers and the second tree of pointers to find one of the pointers to one of the indirect blocks that maps the specified block offset in the logical extent of the production file to a corresponding one of the data blocks including a block of old data, said one of the indirect blocks being included in the third group of indirect blocks, and allocating a new indirect block to the snapshots, copying contents of said one of the indirect blocks to the new indirect block, changing said one of the pointers to said one of the indirect blocks to point to the new indirect block, and changing the contents of one of the new indirect block and said one of the indirect blocks so that the new indirect block maps the block of old data to the specified block offset in the logical extent of the first snapshot copy and the second snapshot copy, and said one of the indirect blocks maps the block of new data to the specified block offset in the logical extent of the production file.

2. The method as claimed in claim 1, which further includes an initial step of creating the production file by allocating an extent of contiguous indirect blocks including the first group of the indirect blocks of the production file and the second group of the indirect blocks of the production file, and setting a direct mapped file type attribute of the production file.

3. The method as claimed in claim 1, wherein step (a) includes executing a background task to build the first tree of pointers, and interrupting the background task in response to a client request to write data to a block offset in the production file mapped by a certain indirect block of the production file by extending a branch of the first tree of pointers from the first inode to the certain indirect block of the production file and then writing data to the block offset in the production file mapped by the certain indirect block of the production file.

4. The method as claimed in claim 1, wherein the building of the first tree of pointers includes allocating and linking indirect blocks between the first inode and said first group of the indirect blocks of the production file, and allocating and linking pointer objects between the indirect blocks of the first tree of pointers and the first group of the indirect blocks of the production file so that each of the indirect blocks of the first tree of pointers is linked to each of the indirect blocks in the first group of the indirect blocks of the production file through a respective one of the pointer objects and the respective one of the pointer objects includes a respective pointer to said each of the indirect blocks in the first group of the indirect blocks of the production file.

5. The method as claimed in claim 4, wherein the building of the second tree of pointers includes allocating and linking indirect blocks between the second inode and the pointer objects that include pointers to the indirect blocks in the third group of the indirect blocks of the production file.

6. The method as claimed in claim 1, wherein the building of the first tree of pointers includes allocating and linking indirect blocks between the first inode and the first group of the indirect blocks of the production file, and allocating and linking pointer objects between the indirect blocks of the first tree of pointers and the first group of the indirect blocks of the production file so that each of the indirect blocks of the first tree of pointers is linked to each of the indirect blocks of the first group of the indirect blocks of the production file through a respective one of the pointer objects;

the building of the second tree of pointers includes allocating and linking indirect blocks between the second inode and the pointer objects that include pointers to the indirect blocks in the third group of the indirect blocks of the production file; and

the building of the second tree of pointers includes de-duplicating indirect blocks in at least one of the first tree of pointers and the second tree of pointers so that indirect blocks are shared between the first tree of pointers and the second tree of pointers.

7. The method as claimed in claim 1, wherein the step (c) includes searching the first tree of pointers and not the second tree of pointers to find said one of the pointers to said one of the indirect blocks that maps the specified block offset in the logical extent of the production file to the corresponding one of the data blocks including the block of old data.

8. The method as claimed in claim 1, wherein the writing of the block of new data to the data storage in the step (c) includes allocating a new data block of the data storage and writing the block of new data to the new data block of the data storage, and the step (c) includes changing the contents of said one of said indirect blocks by changing a block pointer in said one of said indirect blocks to point to the new data block of the data storage.

9. A method of operating a file server, the file server having data storage storing a production file, a data processor coupled to the data storage for access to the production file, and non-transitory computer readable storage medium coupled to the data processor and storing computer instructions, the production file having data blocks and indirect blocks that are allocated to the production file and contain pointers to the data blocks for mapping block offsets in a logical extent of the production file to corresponding ones of the data blocks, and the computer instructions, when executed by the data processor, create a series of snapshot copies of the production file while preserving the allocations of the indirect blocks to the production file by performing the steps of:

(a) creating a first snapshot copy of the production file by allocating a first inode to the first snapshot copy and building a first tree of pointers extending from the first inode to a first group of the indirect blocks of the production file; and

then

(b) creating a second snapshot copy of the production file by allocating a second inode to the second snapshot copy and building a second tree of pointers extending from the second inode to a second group of the indirect blocks of the production file, wherein the first tree of pointers and the second tree of pointers share pointers to a third group of indirect blocks that are allocated to the production file and that are included in both the first group of the indirect blocks of the production file and in the second group of the indirect blocks of the production file; and

then

(c) in response to a request from a client of the file server to write a block of new data to a specified block offset in the logical extent of the production file, allocating a new block of the data storage, writing the block of new data to the new block of data storage, searching the first tree of pointers to find one of the pointers to one of the indirect blocks that maps the specified block offset in the logical extent of the production file to a corresponding one of the data blocks including a block of old data, said one of the indirect blocks being included in the third group of indirect blocks, and allocating a new indirect block to the snapshots, copying contents of said one of the indirect blocks to the new indirect block, changing the pointers to said one of the indirect blocks to point to the new indirect block, and changing the contents of said one of the indirect blocks so that the new indirect block maps the block of old data to the specified offset in the logical extent of the first snapshot copy and the second snapshot copy, and said one of the indirect blocks maps the block of new data to the specified block offset in the logical extent of the production file.

10. The method as claimed in claim 9, which further includes an initial step of creating the production file by allocating an extent of contiguous indirect blocks including the first group of the indirect blocks of the production file and the second group of the indirect blocks of the production file, and setting a direct mapped file type attribute of the production file.

11. The method as claimed in claim 9, wherein step (a) includes executing a background task to build the first tree of pointers, and interrupting the background task in response to a client request to write data to a block offset in the production file mapped by a certain indirect block of the production file by extending a branch of the first tree of pointers from the first inode to the certain indirect block of the production file and then writing data to the block offset in the production file mapped by the certain indirect block of the production file.

12. The method as claimed in claim 9, wherein the building of the first tree of pointers includes allocating and linking indirect blocks between the first inode and the first group of the indirect blocks of the production file, and allocating and linking pointer objects between the indirect blocks of the first tree of pointers and the first group of the indirect blocks of the production file so that each of the indirect blocks of the first tree of pointers is linked to each of the indirect blocks in the first group of the indirect blocks of the production file through a respective one of the pointer objects and the respective one of the pointer objects includes a respective pointer to said each of the indirect blocks in the first group of the indirect blocks of the production file.

13. A file server comprising:

Data storage storing a production file;

a data processor coupled to the data storage for access to the production file; and

non-transitory computer readable storage medium coupled to the data processor and storing computer instructions;

the production file having data blocks and indirect blocks that are allocated to the production file and contain pointers to the data blocks for mapping block offsets in a logical extent of the production file to corresponding ones of the data blocks, and the computer instructions, when executed by the data processor, create a series of snapshot copies of the production file while preserving the allocations of the indirect blocks to the production file by performing the steps of;

(a) creating a first snapshot copy of the production file by allocating a first inode to the first snapshot copy and building a first tree of pointers extending from the first inode to a first group of the indirect blocks of the production file; and

then

(b) creating a second snapshot copy of the production file by allocating a second inode to the second snapshot copy and building a second tree of pointers extending from the second inode to a second group of the indirect blocks of the production file, wherein the first tree of pointers and the second tree of pointers share pointers to a third group of indirect blocks that are included in both the first group of the indirect blocks of the production file and the second group of the indirect blocks of the production file, and then

(c) in response to a request from a client of the file server to write a block of new data to a specified block offset in the logical extent of the production file, writing the block of new data to the data storage, searching one of the first tree of pointers and the second tree of pointers to find one of the pointers to one of the indirect blocks that maps the specified block offset in the logical extent of the production file to a corresponding one of the data blocks including a block of old data, said one of the indirect blocks being included in the third group of indirect blocks, and allocating a new indirect block to the snapshots, copying contents of said one of the indirect blocks to the new indirect block, changing said one of the pointers to said one of the indirect blocks to point to the new indirect block, and changing the contents of one of the new indirect block and said one of the indirect blocks so that the new indirect block maps the block of old data to the specified block offset in the logical extent of the first snapshot copy and the second snapshot copy, and said one of the indirect blocks maps the block of new data to the specified block offset in the logical extent of the production file.

14. The file server as claimed in claim 13, wherein the first group of the indirect blocks of the production file and the second group of the indirect blocks of the production file are included in a contiguous extent of indirect blocks allocated to the production file, and the production file has an attribute indicting a direct mapped file type.

15. The file server as claimed in claim 13, wherein step (a) includes executing a background task to build the first tree of pointers, and interrupting the background task in response to a client request to write data to a block offset in the production file mapped by a certain indirect block of the production file by extending a branch of the first tree of pointers from the first inode to the certain indirect block of the production file and then writing data to the block offset in the production file mapped by the certain indirect block of the production file.

16. The file server as claimed in claim 13, wherein the building of the first tree of pointers includes allocating and linking indirect blocks between the first inode and the first group of the indirect blocks of the production file, and allocating and linking pointer objects between the indirect blocks of the first tree of pointers and the first group of the indirect blocks of the production file so that each of the indirect blocks of the first tree of pointers is linked to each of the indirect blocks in the first group of the indirect blocks of the production file through a respective one of the pointer objects and the respective one of the pointer objects includes a respective pointer to said each of the indirect blocks in the first group of the indirect blocks of the production file.

17. The file server as claimed in claim 16, wherein the building of the second tree of pointers includes allocating and linking indirect blocks between the second inode and the pointer objects that include pointers to the indirect blocks in the third group of the indirect blocks of the production file.

18. The file server as claimed in claim 13, wherein the building of the first tree of pointers includes allocating and linking indirect blocks between the first inode and the first group of the indirect blocks of the production file, and allocating and linking pointer objects between the indirect blocks of the first tree of pointers and the first group of the indirect blocks of the production file so that each of the indirect blocks of the first tree of pointers is linked to each of the indirect blocks in the first group of the indirect blocks of the production file through a respective one of the pointer objects;

and

the building of the second tree of pointers includes allocating and linking indirect blocks between the second inode and the pointer objects that are between the indirect blocks of the first tree of pointers and the third group of the indirect blocks of the production file; and

the building of the second tree of pointers includes de-duplicating indirect blocks in at least one of the first tree of pointers and the second tree of pointers so that indirect blocks are shared between the first tree of pointers and the second tree of pointers.

19. The file server as claimed in claim 13, wherein the step (c) includes searching the first tree of pointers and not the second tree of pointers to find said one of the pointers to said one of the indirect blocks that maps the specified block offset in the logical extent of the production file to the corresponding one of the data blocks including the block of old data.

20. The files server as claimed in claim 13, wherein the writing of the block of new data to the data storage in the step (c) includes allocating a new data block of the data storage and writing the block of new data to the new data block of the data storage, and the step (c) includes changing the contents of said one of said indirect blocks by changing a block pointer in said one of said indirect blocks to point to the new data block of the data storage.