Protocol virtualization for a network file system

• US 20080126434A1
• Filed: 08/03/2006
• Published: 05/29/2008
• Est. Priority Date: 08/03/2006
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Swaminathan, Ram, Lai, Junwen, Uysal, Mustafa
• Granted Patent: US 8,990,270 B2
• Time in Patent Office: Days
• US Class Current: 707/201
• CPC Class Codes: G06F 16/188 Virtual file systems
• CPC Class Codes: H04L 67/1097 for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
• CPC Class Codes: H04L 69/08 Protocols for interworking; Protocol conversion

Abstract

A method of providing protocol virtualization for a network file system. In particular, at least one virtual file server is established in the data path between the clients and the physical file systems. The virtual file server presents a virtualized name space and a virtualized ID space to the clients, the virtualized name and ID space representing the physical files on the physical file servers. To the clients, the virtual file server appears to be file server. To the physical file servers, the virtual file server appears to be a client. Junction directories permit the virtual file servers to combine physically separate directories to appear as one, and permit migration of objects between the physical file servers without perception by the clients.

Claims

1. A method of protocol virtualization for a network file system, comprising:

establishing at least one virtual file server (VFS) in a data path between at least one client and at least one physical file servers (PFS);

presenting a virtualized name space to the at least one client, the virtualized name space mapped to individual name spaces of at least one PFS by the at least one VFS;

virtualizing an ID space;

providing the virtualized ID space to the plurality of clients by the VFSs, the clients perceiving the VFS as a file server;

re-formatting transaction requests from the at least one client by the at least one VFS for delivery to at least one PFS; and

re-formatting transaction responses from the at least one PFS by the at least one VFS for deliver to the at least one client.

2. The protocol virtualization method of claim 1, wherein the at least one VFS is transparent to the at least one client and the at least one PFS.

3. The protocol virtualization method of claim 1, wherein the re-formatting of the transaction requests and the re-formatting of the transaction responses are performed transparently.

4. The protocol virtualization method of claim 1, wherein within the virtualized ID space each file identified by a physical file handle (PFH) on a PFS, the PFH and PFS pair mapped to provide a virtual file handle (VFH), the map of VFHs stored by the at least one VFS.

5. The protocol virtualization method of claim 4, wherein with each file request from a first client, a first VFS receives a first VFH, the first VFS translating the first VFH to the associated PFH and PFS, the first VFS re-formatting the file request and directing the re-formatted file request and PFH to the PFS, the first VFS receiving at least one response from the PFS, the VFS re-formatting the response and transmitting the re-formatted response to the first client.

6. The protocol virtualization method of claim 1, wherein in response to a plurality of PFSs, establishing at least one junction directory upon at least one VFS, the junction directory identifying files mapped between at least two PFS.

7. The protocol virtualization method of claim 6, wherein for a directory request involving the junction directory, the VFS receives the request, identifies the at least two PFSs, re-formatting the request for each identified PFS, receives at least one response from each identified PFS, and returns a combined response.

8. The protocol virtualization of claim 1, wherein the method further includes file migration means for migrating files from a first PFS to a second PFS, wherein the migration is transparent to the at least one client.

9. The protocol virtualization method of claim 1, wherein the method further includes directory migration means for migrating directories from a first PFS to a second PFS, wherein the migration is transparent to the at least one client.

10. A method of protocol virtualization for a network file system, comprising:

establishing at least one virtual file server (VFS) in a data path between at least one client and at least one physical file server (PFS);

presenting a virtualized name space to the at least one client, the virtualized name space mapped to individual name spaces of at least one PFS by the at least one VFS;

virtualizing an ID space in response to a request from a client, within the ID space each file identified by a physical file handle (PFH) on each PFS, the PFH and PFS pair mapped to provide a virtual file handle (VFH), the map of VFHs stored by the at least one VFS; and

providing the VFHs to the plurality of clients by the VFSs, the clients perceiving the VFS as a file server;

wherein with each file request from a first client, a first VFS receives a first VFH, the first VFS translating the first VFH to the associated PFH and PFS, the first VFS re-formatting the file request and directing the re-formatted file request and PFH to the PFS, the first VFS receiving at least one response from the PFS, the VFS reformatting the response and transmitting the re-formatted response to the first client.

11. The protocol virtualization method of claim 10, wherein the at least one VFS is transparent to the at least one client and the at least one PFS.

12. The protocol virtualization method of claim 10, wherein:

in response to a plurality of PFSs, establishing at least one junction directory upon at least one VFS, the junction directory identifying files mapped between at least two PFS;

wherein for a directory request involving the junction directory, the VFS receives the request, identifies at least two PFSs, re-formatting the request for each identified PFS, receives at least one response from each identified PFS, and returns a combined response.

13. The protocol virtualization method of claim 10, wherein each PFS is an NFS server, each VFS perceived as an NFS server by each client, each VFS perceived as a client by each PFS.

14. The protocol virtualization method of claim 10, wherein the mapping of each PFH and PFS pair to a VFH is unique and arbitrary.

15. The protocol virtualization of claim 10, wherein there are a plurality of VFSs, the map of VFHs allocated between the VFSs.

16. The protocol virtualization method of claim 15, wherein for the translation of the first VFH, the first VFS consults its allocation of the map;

in response to the translation occurring in the allocation, the first VFH performing the translation directly, and

in response to the translation being unavailable in the allocation performing a hash upon the VFH to identify a second VFS having the appropriate allocation and requesting the translation from the second VFS.

17. The protocol virtualization method of claim 16, wherein requested translations are locally cached by the first VFH for a tunable period of time.

18. The protocol virtualization method of claim 10, wherein the method is stored on a computer readable medium as a computer program which, when executed by a computer will perform the steps of protocol virtualization.

19. The protocol virtualization of claim 10, further including a method of file migration, comprising:

selecting a VFS to act as a choke point for all access transactions involving the files to be moved;

informing all other VFSs of the selected VFS choke point, all other VFSs directing requests to VFHs translating to involved files to the choke point;

copying each involved file from a source directory PFS to a destination directory PFS;

updating the map of PFH, PFS and associated VFH entries to reflect the destination directory; and

releasing the VFS choke point and informing all other VFSs of the release; and

propagating the updated VFHs to the other VFSs;

wherein for each read request transaction received by the VFS choke point, accessing and returning data from the source directory;

wherein for each write request transaction received by the VFS choke point, the VFS choke point delivering the write transaction to both the source directory and the destination directory.

20. The protocol virtualization of claim 10, further including a method of directory migration, comprising:

selecting a VFS to act as a choke point for all access transactions involving a source directory on a source PFS to be moved;

informing all other VFSs of the selected VFS choke point, all other VFSs directing requests to VFHs translating to involved files under the source directory to the choke point;

performing destination setup, including;

establishing a destination directory on a destination PFS corresponding to the source directory;

transforming the source directory into a junction directory consisting of all existing objects within the source directory and the destination directory;

retrieving a list of all objects within the source directory;

copying each file object from the source directory to the destination directory;

in response to a sub-directory object, recursively performing the destination setup for each sub-directory object;

updating the map of PFH, PFS and associated VFH entries to reflect the at least one destination directory; and

releasing the VFS choke point and informing all other VFSs of the release; and

propagating the updated VFHs to the other VFSs;

wherein for each read request transaction received by the VFS choke point, accessing and returning data from the source directory;

wherein for each write request transaction received by the VFS choke point, the VFS choke point delivering the write transaction to both the source directory and the destination directory.

21. A network file system, comprising:

at least one unmodified client;

at least one unmodified physical file server (PFS), each PFS having at least one file identified by an associated physical file handle (PFH);

at least one virtual file server (VFS) in a data path between the at least one client and at least one PFS;

a virtualized name space mapped to the at least one PFS, the name space provided to the at least one client by the at least one VFS;

a virtualized ID space established in response to a request from a client, within the ID space each file identified by a physical file handle (PFH) on a PFS, the PFH and PFS pair mapped to provide a virtual file handle (VFH), the map of VFHs stored by the at least one VFS.

22. The network file system of claim 21, wherein with each file request from a first client, a first VFS is operable to receive a first VFH, the first VFS translating the first VFH to the associated PFH and PFS, the first VFS re-formatting the file request and directing the re-formatted file request and the PFH to the PFS, the first VFS receiving at least one response from the PFS, the VFS re-formatting the response and transmitting the re-formatted response to the first client.

23. The network file system of claim 21, further including a junction directory upon at least one VFS, the junction directory identifying files mapped between at least two PFSs.

24. The network file system of claim 23, wherein for a directory request involving the junction directory, a first VFS is operable to receive the request, identifies the at least two PFSs, re-formatting the request for each identified PFS, receives at least one response from each identified PFS, and returns a combined response.

25. The network file system claim 21, wherein the mapping of each PFH and PFS pair to a VFH is unique and arbitrary.

26. The network file system claim 21, wherein there are a plurality of VFSs, the map of VFHs allocated between the VFSs.

27. The network file system claim 21, wherein for the translation of the first VFH the first VFS is operable to consult its allocation of the map;

in response to the translation occurring in the allocation, the first VFH performing the translation directly;

in response to the translation being unavailable in the allocation performing a hash upon the VFH to identify a second VFS having the appropriate allocation and requesting the translation from the second VFS.

28. The network file system of claim 27, wherein each VFH is operable to cache requested translations for a tunable period of time.