Striping directories across a striped volume set by the filenames contained in the directories

• US 8,224,864 B1
• Filed: 01/07/2008
• Issued: 07/17/2012
• Est. Priority Date: 01/07/2008
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: Network Appliance Incorporated (NetApp, Inc.)
• Inventors: Jernigan, Richard, Hyer, Robert
• Primary Examiner(s): Timblin, Robert
• Assistant Examiner(s): BOWEN, RICHARD L
• Application Number:
  US11/970,452
• Time in Patent Office: 1,653 Days
• Field of Search: None
• US Class Current: 707/802
• CPC Class Codes: G06F 16/137 Hash-based content-based indexing of textual data G06F16/31
• CPC Class Codes: G06F 16/1827 Management specifically adapted to NAS management of storage area networks [SAN] G06F3/067

Abstract

A method and apparatus for striping a directory across storage server nodes. Each storage server node services one of a plurality of persistent mass storage facilities. The method comprises mapping each filename in the directory to one of the persistent mass storage facilities. According to the mapping of the filenames, the directory is distributed across the storage server nodes for storage on the persistent mass storage facilities.

Claims

1. A method of operating a cluster of storage server nodes, the method comprising:

mapping a given one of a plurality of filenames in a directory to a given one of the storage server nodes according to a first mapping, each storage server node services one of a plurality of persistent mass storage facilities;

choosing an identifier of a meta-data container that stores meta-data attributes of the given filename, such that the identifier is mapped to the given storage server node according to a second mapping;

distributing the directory, including the filenames and pointer information that points filenames to corresponding identifiers of corresponding meta-data containers, across the storage server nodes for storage on the persistent mass storage facilities according to the first mapping of the filenames; and

in response to renaming the given filename to a second filename that is mapped to a different one of the storage server nodes according to the first mapping, moving the pointer information of the second filename to the different storage server node while retaining the meta-data attributes of the second filename on the given storage server node.

2. The method of claim 1, wherein mapping a given one of a plurality of filenames further comprises:

converting each of the filenames into a volume identifier to identify a volume serviced by one of the storage server nodes, the meta-data container associated with a file having the filename.

3. The method of claim 2, wherein converting each of the filenames into a volume identifier further comprises:

converting each of the filenames into a numerical string; and

hashing and mapping the numerical string into the volume identifier.

4. The method of claim 1, further comprising:

creating a new directory entry in the directory, the new directory entry including a new filename for a new file;

mapping the new filename to a volume serviced by one of the storage server nodes; and

assigning an identifier of a meta-data container to the new filename, the identifier identifying the mapped volume, wherein the volume stores a copy of meta-data attributes of the new file and a portion of the directory including the identifier.

5. The method of claim 1, further comprising:

receiving a request to retrieve meta-data attributes of a file in the directory; and

returning a file handle for the directory, the file handle indicating whether the directory is striped by the filenames.

6. The method of claim 1, further comprising:

receiving a request for a file at a first storage server node, the request specifying a requested filename associated with the file; and

forwarding the request from the first storage server node to a second storage server node in response to a determination that the requested filename is mapped to a volume serviced by the second storage server.

7. A storage system comprising:

a plurality of arrays of storage devices to store volumes that are organized as a striped volume set for a directory, wherein the directory contains filenames and pointer information that points the filenames to corresponding identifiers of corresponding meta-data containers, and wherein the directory is distributed across the striped volume set according to a first mapping of the filenames in the directory to the volumes; and

a plurality of storage server nodes coupled to the arrays of storage devices, each storage server node servicing at least one of the volumes in the striped volume set,

wherein a given one of the filename is mapped to a given one of the storage server nodes according to the first mapping, and an identifier of a meta-data container that stores meta-data attributes of the given filename is chosen such that the identifier is mapped to the given storage server node according to a second mapping, and

wherein in response to renaming the given filename to a second filename that is mapped to a different one of the storage server nodes according to the first mapping, the storage server nodes move the pointer information of the second filename to the different storage server node while retaining the meta-data attributes of the second filename on the given storage server node.

8. The storage system of claim 7, wherein the storage server nodes map the filenames by converting each of the filenames into a numerical string and hashing and mapping the numerical string into a volume identifier that identifies one of the volumes.

9. The storage system of claim 7, further comprising:

a switching fabric via which the storage server nodes communicate, the switching fabric forwards a request for a file from a first node of the storage server nodes to a second node of the storage server nodes in response to a determination by the first node that a filename of the file is mapped to one of the volumes serviced by the second node.

10. The storage system of claim 7, wherein each of the storage server nodes further comprises:

a network module to receive a request for a file from a client, and to hash a filename associated with the file to identify one of the storage server nodes as an owner of the filename; and

a data module to perform data operations according to the request.

11. The storage system of claim 7, wherein the meta-data container is associated with files having the filenames.

12. The storage system of claim 7, wherein the storage server nodes, in response to a request to create a new file having a new filename in the directory, store the mapping of the new filename and meta-data attributes of the new file in a same one of the volumes.

13. The storage system of claim 7, wherein a plurality of directories are distributed across the striped volume set, one or more of the directories distributed with different striping algorithms based on which the mapping of the filenames is performed.

14. A method of operating a cluster of storage server nodes, the method comprising:

receiving a request for a file having a filename, the request involving retrieving meta-data attributes of the file from a directory that holds the filename;

mapping the filename to a volume according to a first mapping;

directing the request to a given one of the storage server nodes servicing the volume according to the first mapping;

obtaining pointer information that points the filename to an identifier of a meta-data container that stores the meta-data attributes of the filename;

mapping the identifier to the given storage server node according to a second mapping and performing file operations according to the request,

wherein the identifier was chosen when creating the file such that the identifier is mapped to the given storage server node according to the second mapping, and

in response to renaming the filename to a second filename that is mapped to a different one of the storage server nodes according to the first mapping, moving the pointer information of the second filename to the different storage serve node while retaining the meta-data attributes of the second filename on the given storage server node.

15. The method of claim 14, wherein mapping the filename further comprises:

converting the filename into a numerical string; and

hashing and mapping the numerical string into a volume identifier that identifies the volume.

16. The method of claim 14, further comprising:

receiving the request at a first storage server node; and

forwarding the request from the first storage server node to a second storage server node through a switching fabric in response to a determination that the filename is mapped to the volume serviced by the second storage server.

17. The method of claim 14, further comprising:

creating a new file having a new filename in the directory; and

storing the mapping of the new filename and meta-data attributes of the new file in a same volume.

18. The method of claim 14, further comprising:

associating the directory with a file handle to provide access to the directory, the file handle indicating whether the directory is striped and which striping algorithm is used.

19. The method claim 14, further comprising:

distributing a plurality of directories across a set of volumes, one or more of the directories distributed with different striping algorithms based on which the mapping of the filename is performed.