MANAGING THE LEVEL OF CONSISTENCY FOR A FILE IN A DISTRIBUTED FILESYSTEM

• US 20160070741A1
• Filed: 09/10/2014
• Published: 03/10/2016
• Est. Priority Date: 09/10/2014
• Status: Active Grant

• Current Assignee: Panzura, Inc.
• Original Assignee: Panzura, Inc.
• Inventors: Lin, Yun, Jung, Steve Hyuntae, Taylor, John Richard, Anneboina, Vinay Kumar
• Granted Patent: US 9,990,372 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/172 Caching, prefetching or hoarding of files
• CPC Class Codes: G06F 16/1774 Locking methods, e.g. locking methods for file systems allowing shared and concurrent access to files
• CPC Class Codes: G06F 16/178 Techniques for file synchronisation in file systems
• CPC Class Codes: G06F 16/183 Provision of network file services by network file servers, e.g. by using NFS, CIFS network file access protocols H04L67/1097

Abstract

The disclosed embodiments disclose techniques for managing consistency for a file in a distributed filesystem. Two or more cloud controllers collectively manage distributed filesystem data that is stored in the cloud storage systems; the cloud controllers ensure data consistency for the stored data, and each cloud controller caches portions of the distributed filesystem. During operation, a cloud controller receives from a client a request to access the file. The cloud controller determines a level of consistency that is associated with the file, and then uses this level of consistency to determine whether to communicate with a peer cloud controller when handling the request.

Claims

1. A computer-implemented method for managing a level of consistency for a file in a distributed filesystem, the method comprising:

collectively managing the data of the distributed filesystem using two or more cloud controllers, wherein collectively managing the data comprises storing the data for the distributed filesystem in one or more cloud storage systems, wherein the cloud controllers cache and ensure data consistency for data stored in the cloud storage systems;

receiving at a cloud controller a request from a client to access the file;

determining a level of consistency that is associated with the file; and

determining from the level of consistency for the file whether to communicate with a peer cloud controller when handling the request.

2. The computer-implemented method of claim 1, wherein the cloud controller determines that the file is associated with local file consistency and does not need to be kept consistent across the other peer cloud controllers for the distributed filesystem.

3. The computer-implemented method of claim 2,

wherein the file already exists in the distributed filesystem and the cloud controller is configured to use only metadata and data that is already cached locally on the cloud controller when handling the request;

and

wherein not contacting any other peer cloud controllers or a cloud storage system to retrieve data or updates for the file facilitates reducing file access delays for the request.

4. The computer-implemented method of claim 2,

wherein upon determining that the file does not yet exist in the distributed filesystem and that additional data needs to be downloaded from a peer cloud controller and/or a cloud storage system to populate the file, the cloud controller is configured to use a local-only version of the file on the cloud controller;

and

wherein using the local-only version of the file facilitates handling the request without needing to retrieve data or updates associated with the file from peer cloud controllers or the cloud storage system, thereby reducing file access delays for the request.

5. The computer-implemented method of claim 1, wherein the cloud controller determines that the file needs write-after-write consistency.

6. The computer-implemented method of claim 5,

wherein the cloud controller determines that the request is a non-write request;

and

wherein the cloud controller uses the most up-to-date version of the file that is known to the cloud controller and does not check with any peer cloud controllers for more recent updates to the file.

7. The computer-implemented method of claim 5,

wherein the cloud controller determines that the request is a write request;

and

wherein the cloud controller is configured to ensure write-after-write consistency for the file by;

contacting the owning cloud controller for the file to determine whether the file has been updated; and

if a more recent version of the file exists on the owning cloud controller, synchronizing the local version of the file with the more recent version of the file and claiming ownership for the file to ensure that the file can be updated exclusively on the cloud controller for the duration of the file access.

8. The computer-implemented method of claim 5,

wherein new files in the distributed filesystem are associated with write-after-write consistency by default unless an alternative level of consistency is explicitly specified or determined by the cloud controller;

wherein each cloud controller is configured to periodically send incremental metadata snapshots to all of the other cloud controllers of the distributed filesystem to lazily propagate metadata changes for recently modified files throughout the distributed filesystem;

wherein read requests for files in the distributed filesystem are typically more frequent than write requests for files in the distributed filesystem; and

wherein a default of write-after-write consistency facilitates reducing network load and cloud controller overhead for the distributed filesystem by primarily incurring the overhead of communicating with peer cloud controllers during write accesses.

9. The computer-implemented method of claim 1, wherein the cloud controller determines that the file needs read-after-close consistency.

10. The computer-implemented method of claim 9, wherein the cloud controller is configured to ensure read-after-close consistency for the file by:

registering with all of the peer cloud controllers for the distributed filesystem to request change notifications for the file;

sending change notifications to any peer cloud controllers that have registered to receive notifications for the file whenever the cloud controller receives a write to the file from the client;

contacting the owning cloud controller for the file to determine whether the requested file has been updated; and

if a more recent version of the file exists on the owning cloud controller, synchronizing the local version of the file with the more recent version of the file.

11. The computer-implemented method of claim 1, wherein the cloud controller determines that the file is associated with real-time file consistency and that the request involves byte-range locking techniques that support multiple concurrent writes to the file.

12. The computer-implemented method of claim 1, wherein the cloud controllers for the distributed filesystem are configured to monitor and dynamically adjust the level of consistency that is associated with the file based on client accesses to the file and to balance performance and consistency for the distributed filesystem.

13. The computer-implemented method of claim 12, wherein an initial level of consistency for the file is determined based on at least one of the file'"'"'s name, the file'"'"'s type, the file'"'"'s geographic location, the file'"'"'s location in the distributed filesystem hierarchy, the application accessing the file, and/or file access patterns.

14. The computer-implemented method of claim 13,

wherein the file is initially created and accessed by a single client and assigned a low level of consistency by the cloud controller;

wherein the cloud controller subsequently determines that the file needs a higher level of consistency and that the cloud controllers for the distributed filesystem can support a higher level of consistency for the file, and dynamically increases the level of consistency for the file; and

wherein the cloud controller considers collaborative access patterns, available compute resources, and available network bandwidth to determine the initial and subsequent levels of consistency for the file.

15. The computer-implemented method of claim 14,

wherein the cloud controller subsequently decreases the level of consistency for the file upon determining at least one of:

a change in the number of distributed clients that are accessing the file;

a change in the types of client accesses for the file; and

a change in the set of resources that are available for the distributed filesystem.

16. A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for managing a level of consistency for a file in a distributed filesystem, the method comprising:

collectively managing the data of the distributed filesystem using two or more cloud controllers, wherein collectively managing the data comprises storing the data for the distributed filesystem in one or more cloud storage systems,

wherein the cloud controllers cache and ensure data consistency for data stored in the cloud storage systems;

receiving at a cloud controller a request from a client to access the file;

determining a level of consistency that is associated with the file; and

determining from the level of consistency for the file whether to communicate with a peer cloud controller when handling the request.

17. A cloud controller that manages a level of consistency for a file in a distributed filesystem, comprising:

a processor;

a storage mechanism that stores metadata for the distributed filesystem; and

a storage management mechanism;

wherein two or more cloud controllers collectively manage the data of the distributed filesystem;

wherein the cloud controller is configured to receive a request from a client to access the file;

wherein the storage management mechanism is configured to determine a level of consistency that is associated with the file; and

wherein the storage management mechanism is further configured to determine from the level of consistency for the file whether to communicate with a peer cloud controller when handling the request.