Hierarchical mapping of free blocks of cylinder groups of file systems built on slices of storage and linking of the free blocks

• US 8,407,265 B1
• Filed: 11/19/2010
• Issued: 03/26/2013
• Est. Priority Date: 11/19/2010
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Scheer, Michael D., Bono, Jean-Pierre, Clark, Morgan, Tang, Hongliang, Veeraswamy, Sairam, Sethi, Pranit, Mathews, Alexander S.
• Primary Examiner(s): KIM, HONG CHONG
• Application Number:
  US12/950,996
• Time in Patent Office: 858 Days
• Field of Search: 707/823, 711/170, 711/111, 711/112, 711/114
• US Class Current: 707/823
• CPC Class Codes: G06F 16/1727 Details of free space management performed by the file system saving storage space on storage systems G06F3/0608; management of blocks in storage devices G06F3/064

Abstract

In a file server, a file system is built upon a volume of data storage. The file system includes multiple cylinder groups. Each cylinder group includes file system blocks. The file system blocks include allocated blocks and free blocks. The volume includes slices of storage, and each slice stores at least one of the cylinder groups. A hierarchical slice map has a top level that includes a count of free blocks in the file system, an upper level that includes a count of free blocks in groups of the slices, an intermediate level that includes a count of the free blocks in sub-groups of the slices, and a bottom level that includes a count of the free blocks in each slice of storage. To find a free block for allocation to a file, the slice map hierarchy is searched in a top-down fashion.

Claims

1. A computer-implemented method of operating a file server, the file server including data storage and a data processor, the data storage storing a logical volume and a file system built on the logical volume, the file system including multiple cylinder groups, each of the cylinder groups including file system blocks, the file system blocks including allocated blocks and free blocks of the data storage, the logical volume including slices of the data storage, and each slice of the data storage storing at least one of the cylinder groups, the method comprising the data processor executing computer instructions stored on a non-transitory computer-readable storage medium to perform the steps of:

(a) maintaining, in the file server, a hierarchical map of the free blocks of the data storage in the file system, the hierarchical map having an upper level including a summary of the free blocks of the data storage in groups of the slices, the hierarchical map having at least one intermediate level including a summary of the free blocks of the data storage in sub-groups of the slices, and the hierarchical map having a bottom level including a summary of the free blocks of the data storage in each of the slices of the data storage; and

(b) searching the hierarchical map in a top-down fashion to find a cylinder group of the file system including a free block of the data storage in the file system; and

(c) linking the free block of the data storage from the cylinder group found in step (b) to a file.

2. The computer-implemented method as claimed in claim 1, which further includes pre-allocating a number of the free blocks at the upper level of the hierarchical map, and then finding at a lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, and in response to finding at the lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, then releasing said some of the number of the pre-allocated free blocks from the pre-allocation of the number of the free blocks at the upper level.

3. The computer-implemented method as claimed in claim 1, wherein the hierarchical map includes a vector of slice group resource descriptors indicating a presence or absence of at least one free block in each of the groups of the slices, and for each of the groups of the slices, a respective vector of slice sub-group resource descriptors indicating a presence or absence of at least one free block in each of the sub-groups of the slices included in said each of the groups of the slices;

and

wherein step (b) of searching the hierarchical map in a top-down fashion includes scanning the vector of slice group resource descriptors until one of the slice group resource descriptors in the vector of slice group resource descriptors is found to indicate the presence of at least one free block in one of the groups of the slices, and upon finding said one of the slice group resource descriptors, then scanning the vector of slice sub-group resource descriptors indicating a presence or absence of at least one free block in each of the sub-groups of the slices included in said one of the groups of the slices; and

wherein the step (a) of maintaining the hierarchical map in the file server includes updating the levels of the hierarchical map in a top-down fashion through the hierarchical map to pre-allocate free blocks in the levels of the hierarchical map before the pre-allocated blocks are allocated in the file system, and after blocks allocated in the file system are freed in the file system, updating the levels of the hierarchical map in a bottom-up fashion through the file system to add the freed blocks to the levels of the hierarchical map, so that free blocks found at an upper level of the hierarchical map will also be found at a lower level of the hierarchical map during the searching of the hierarchical map in a top-down fashion in step (b).

4. The computer-implemented method as claimed in claim 1, wherein the step (a) of maintaining the hierarchical map in the file server includes updating the levels of the hierarchical map in a bottom-up fashion upon performing step (c) of linking the free block of the data storage from the cylinder group found in step (b) to the file.

5. The computer-implemented method as claimed in claim 1, wherein the hierarchical map includes a respective count of the free file system blocks for each of the groups of the slices, a respective count of the free file system blocks for each of the sub-groups of the slices, and a respective count of the free file system blocks for each of the slices, and

the step (a) of maintaining the hierarchical map in the file server includes, upon performing step (c) of linking the free block of the data storage from the cylinder group found in step (b) to the file:

(1) updating the hierarchical map by decrementing a respective one of the counts for a respective one of the groups of the slices of the data storage including the free block of the data storage that was allocated to the file, and (2) updating the hierarchical map by decrementing a respective one of the counts for a respective one of the sub-groups of the slices of the data storage including the free block of the data storage that was allocated to the file, and (3) updating the hierarchical map by decrementing a respective one of the counts for a respective one of the groups of the slices of the data storage including the free block of the data storage that was allocated to the file.

6. The computer-implemented method as claimed in claim 1, wherein the logical volume includes some slices that are not provisioned with the data storage so that the logical volume is sparse, and wherein the free block linked to the file is a block in a slice that is provisioned with the data storage, and

the step (a) of maintaining the hierarchical map of the free blocks in the file system includes maintaining in the hierarchical map:

(1) respective counts of slices that are provisioned with the data storage in the groups of the slices, and (2) respective counts of slices that are provisioned with the data storage in the sub-groups of the slices.

7. The computer-implemented method as claimed in claim 6, wherein the step (a) of maintaining the hierarchical map of the free blocks in the file system further includes, upon provisioning a slice with data storage, updating the levels of the hierarchical map in a bottom-up fashion to increment the respective counts of slices that are provisioned with the data storage and to increment counts of free blocks of the file system.

8. The computer-implemented method as claimed in claim 6, wherein the step (a) of maintaining the hierarchical map of the free blocks in the file system further includes, upon evacuating allocated blocks from a slice provisioned with data storage, updating the levels of the hierarchical map in a bottom-up fashion to increment counts of free blocks of the file system, and once the slice provisioned with data storage has been fully evacuated, updating the levels of the hierarchical map in a top-down fashion to decrement the respective counts of slices that are provisioned with the data storage and to decrement the counts of free blocks of the file system, and removing backing storage from the slice that has been fully evacuated.

9. The computer-implemented method as claimed in claim 1, wherein:

the data storage includes a plurality of types of physical storage, and the summary information in the hierarchical map includes, for each of the groups of the slices, a count of free blocks of each type of physical storage; and

the summary information in the hierarchical map includes, for each of the sub-groups of the slices, a count of free blocks of each type of physical storage; and

the step (b) of searching the hierarchical map in a top-down fashion is performed in response to a request to allocate a block of a specified one of the types of physical storage, and the step (b) of searching the hierarchical map in a top-down fashion finds an indication in the hierarchical map the that one of the groups of the slices includes at least one block of the specified one of the types of physical storage, and finds an indication in the hierarchical map that one of the sub-groups of the slices includes at least one block of the specified one of the types of physical storage, in order to find a cylinder group of the file system containing a free block of the specified one of the types of physical storage so that the free block linked in step (c) to the file is stored in the specified one of the types of physical storage.

10. The computer-implemented method as claimed in claim 1, wherein:

the hierarchical map includes respective nodes for the groups of the slices, and respective nodes for the sub-groups of the slices, and there is an algebraic relationship between addresses of the nodes for the groups of the slices and addresses of the nodes of the sub-groups of the slices, and

the step (b) of searching the hierarchical map in a top-down fashion to find a cylinder group of the file system including searching an upper level of the hierarchical map to find an index (i) of one of the groups of the slices having the cylinder group found in step (b), and then performing an algebraic computation upon the index (i) to compute an address of the node of said one of the sub-groups of the slices having the cylinder group found in step (b), and using the computed address of the node of said one of the sub-groups of the slices having the cylinder group found in step (b) to search the node of said one of the sub-groups of the slices having the cylinder group found in step (b).

11. A file server comprising:

data storage storing a logical volume and a file system built on the logical volume, the file system including multiple cylinder groups, each of the cylinder groups including file system blocks, the file system blocks including allocated blocks and free blocks of the data storage, the logical volume including slices of the data storage, and each slice of the data storage storing at least one of the cylinder groups,

a data processor coupled to the data storage for storing data in the data storage and for reading stored data from the data storage; and

a non-transitory computer readable storage medium storing computer instructions that, when executed by the data processor, perform the steps of;

(a) maintaining, in the file server, a hierarchical map of the free blocks of the data storage in the file system, the hierarchical map having an upper level including a summary of the free blocks of the data storage in groups of the slices, the hierarchical map having at least one intermediate level including a summary of the free blocks of the data storage in sub-groups of the slices, and the hierarchical map having a bottom level including a summary of the free blocks of the data storage in each of the slices of the data storage;

(b) searching the hierarchical map in a top-down fashion to find a cylinder group of the file system including a free block of the data storage in the file system; and

(c) linking the free block of the data storage from the cylinder group found in step (b) to the file.

12. The file server as claimed in claim 11, wherein the computer instructions, when executed by the data processor, further perform the steps of:

pre-allocating a number of the free blocks at the upper level of the hierarchical map, and then finding at a lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, and in response to finding at the lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, then releasing said some of the number of the pre-allocated free blocks from the pre-allocation of the number of the free blocks at the upper level.

13. The file server as claimed in claim 12, wherein the hierarchical map includes a vector of slice group resource descriptors indicating a presence or absence of at least one free block in each of the groups of the slices, and for each of the groups of the slices, a respective vector of slice sub-group resource descriptors indicating a presence or absence of at least one free block in each of the sub-groups of the slices included in said each of the groups of the slices;

and

wherein step (b) of searching the hierarchical map in a top-down fashion includes scanning the vector of slice group resource descriptors until one of the slice group resource descriptors in the vector of slice group resource descriptors is found to indicate the presence of at least one free block in one of the groups of the slices, and upon finding said one of the slice group resource descriptors, then scanning the vector of slice sub-group resource descriptors indicating a presence or absence of at least one free block in each of the sub-groups of the slices included in said one of the groups of the slices; and

wherein the step (a) of maintaining the hierarchical map in the file server includes updating the levels of the hierarchical map in a top-down fashion through the hierarchical map to pre-allocate free blocks in the levels of the hierarchical map before the pre-allocated blocks are allocated in the file system, and after blocks allocated in the file system are freed in the file system, updating the levels of the hierarchical map in a bottom-up fashion through the file system to add the freed blocks to the levels of the hierarchical map, so that free blocks found at an upper level of the hierarchical map will also be found at a lower level of the hierarchical map during the searching of the hierarchical map in a top-down fashion in step (b).

14. The file server as claimed in claim 11, wherein the step (a) of maintaining the hierarchical map in the file server includes updating the levels of the hierarchical map in a bottom-up fashion upon performing step (c) of linking the free block of the data storage from the cylinder group found in step (b) to the file.

15. The file server as claimed in claim 11, wherein the hierarchical map includes a respective count of the allocated or free file system blocks for each of the groups of the slices, a respective count of the allocated or free file system blocks for each of the sub-groups of the slices, and a respective count of the allocated or free file system blocks for each of the slices, and

the step (a) of maintaining the hierarchical map in the file server includes, upon performing step (c) of linking the free block of the data storage from the cylinder group found in step (b) to the file:

(1) updating the hierarchical map by incrementing or decrementing a respective one of the counts for a respective one of the groups of the slices of the data storage including the free block of the data storage that was allocated to the file, and (2) updating the hierarchical map by incrementing or decrementing a respective one of the counts for a respective one of the sub-groups of the slices of the data storage including the free block of the data storage that was allocated to the file, and (3) updating the hierarchical map by incrementing or decrementing a respective one of the counts for a respective one of the of the groups of the slices of the data storage including the free block of the data storage that was allocated to the file.

16. The file server as claimed in claim 11, wherein the logical volume includes some slices that are not provisioned with the data storage so that the logical volume is sparse, and wherein the free block linked to the file is a block in a slice that is provisioned with the data storage, and

the step (a) of maintaining the hierarchical map of the free blocks in the file system includes maintaining in the hierarchical map:

(1) respective counts of slices that are provisioned with the data storage in the groups of the slices, and (2) respective counts of slices that are provisioned with the data storage in the sub-groups of the slices.

17. The file server as claimed in claim 11, wherein:

the data storage includes a plurality of types of physical storage, and the summary information in the hierarchical map includes, for each of the groups of the slices, an indication of whether or not said each of the groups of the slices includes any blocks of each type of physical storage; and

the summary information in the hierarchical map includes, for each of the sub-groups of the slices, an indication of whether or not said each of the sub-groups of the slices includes any blocks of each type of physical storage; and

the step (b) of searching the hierarchical map in a top-down fashion is performed in response to a request to allocate a block of a specified one of the types of physical storage, and the step (b) of searching the hierarchical map in a top-down fashion finds an indication in the hierarchical map the that one of the groups of the slices includes at least one block of the specified one of the types of physical storage, and finds an indication in the hierarchical map that one of the sub-groups of the slices includes at least one block of the specified one of the types of physical storage, in order to find a cylinder group of the file system containing a free block of the specified one of the types of physical storage so that the free block linked in step (c) to the file is stored in the specified one of the types of physical storage.

18. The file server as claimed in claim 11, wherein:

the hierarchical map includes respective nodes for the groups of the slices, and respective nodes for the sub-groups of the slices, and there is an algebraic relationship between addresses of the nodes for the groups of the slices and addresses of the nodes of the sub-groups of the slices, and

the step (b) of searching the hierarchical map in a top-down fashion to find a cylinder group of the file system including searching an upper level of the hierarchical map to find an index (i) of one of the groups of the slices having the cylinder group found in step (b), and then performing an algebraic computation upon the index (i) to compute an address of the node of said one of the sub-groups of the slices having the cylinder group found in step (b), and using the computed address of the node of said one of the sub-groups of the slices having the cylinder group found in step (b) to search the node of said one of the sub-groups of the slices having the cylinder group found in step (b).

19. A non-transitory computer-readable storage medium containing instructions that, when executed by a data processor of a file server, perform the steps of:

(a) maintaining, in the file server, a hierarchical map of free blocks of data storage in a file system built upon a volume of slices of the data storage, each of the slices of the data storage containing at least one cylinder group of the file system, the hierarchical map having an upper level including a summary of the free blocks of the data storage in groups of the slices, the hierarchical map having at least one intermediate level including a summary of the free blocks of the data storage in sub-groups of the slices, and the hierarchical map having a bottom level including a summary of the free blocks of the data storage in each of the slices of the data storage; and

(b) searching the hierarchical map in a top-down fashion to find a cylinder group of the file system including a free block of the data storage in the file system; and

(c) linking the free block of the data storage from the cylinder group found in step (b) to a file.

20. The non-transitory computer-readable storage medium as claimed in claim 19, wherein the instructions, when executed by a data processor of a file server, further perform the steps of:

pre-allocating a number of the free blocks at the upper level of the hierarchical map, and then finding at a lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, and in response to finding at the lower level of the hierarchical map that some of the number of the pre-allocated free blocks do not have desired characteristics, then releasing said some of the number of the pre-allocated free blocks from the pre-allocation of the number of the free blocks at the upper level.