Systems and method of implementing disk ownership in networked storage

• US 7,650,412 B2
• Filed: 12/21/2001
• Issued: 01/19/2010
• Est. Priority Date: 12/21/2001
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Sarma, Joydeep Sen, Aster, Radek, Coatney, Susan M., Rowe, Alan L.
• Primary Examiner(s): Vaughn, Jr.; William C
• Assistant Examiner(s): Bayard; Djenane M
• Application Number:
  US10/027,457
• Publication Number:
  US 20030120743A1
• Time in Patent Office: 2,951 Days
• Field of Search: 709/245, 709/203, 709/217, 709/226, 709/211, 709/208, 709/229, 711/112, 711/156, 714/4, 714/6
• US Class Current: 709/226
• CPC Class Codes: G06F 3/0607 by facilitating the process of upgrading existing storage systems, e.g. for improving compatibility between host and storage device
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0637 Permissions
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: H04L 63/10 for controlling access to devices or network resources

Abstract

A system and method for disk ownership in a network storage system. Each disk has two ownership attributes set to show that a particular file server owns the disk. In a preferred embodiment the first ownership attribute is the serial number of the file server being written to a specific location on each disk and the second ownership attribute is setting a SCSI-3 persistent reservation. In a system utilizing this disk ownership method, multiple file servers can read data from a given disk, but only the file server that owns a particular disk can write data to the disk.

Claims

1. A method of claiming ownership of a plurality of disks by a network device of a plurality of network devices in a network storage system, comprising:

writing ownership information to a predetermined area of each disk, wherein the predetermined area of the disk is sector 0 on the disk and the ownership information stored in sector 0 is definitive ownership data for determining ownership of the disk;

setting a small computer system interface (SCSI) reservation tag for each disk to a state of network device ownership to provide a two part indicia of ownership for each disk, where the two part indicia of ownership are both written to each disk;

creating a table on each network device in the network storage system;

identifying all disks owned by each network device using ownership information written to the predetermined area of each disk of the plurality disks and, for each identified disk, if a mismatch occurs between the ownership information on the predetermined area of the disk and the ownership defined by the SCSI reservation tag, then using the ownership information written to the predetermined area of the disk as definite ownership data without requiring the owned network device to send a second SCSI reservation tag;

in response to identifying, storing entries in the table, wherein each entry identifies an owned disk of the network device storing the table;

identifying, by a second network device, all disks owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each disk;

setting a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state; and

removing ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk into the unowned state.

2. The method of claim 1 wherein the ownership information further comprises a serial number of a network device.

3. The method of claim 1, wherein the network device comprises a file server.

4. The method of claim 1 wherein the small computer system interface reservation tag and the ownership information at the predetermined area of the disk indicate ownership by the same network device.

5. The method of claim 1 wherein the small computer system interface reservation tag is a small computer system interface level 3 persistent reservation tag.

6. A network storage system comprising:

a plurality of network devices;

one or more switches, each network device connected to at least one of the one or more switch;

a plurality of disks having a first ownership attribute written to a predetermined area of each disk and a second ownership attribute in the form of a small computer system interface (SCSI) reservation tag, wherein the first and second ownership attribute are written to each disk, each disk connected to at least one of the plurality of switches, wherein the predetermined area of the disk is sector 0 on the disk and the ownership information stored in sector 0 is definitive ownership data for determining ownership of the disk;

each network device of the plurality of network devices identifies all disks owned by that network device using ownership information written to the predetermined area of each disk of the plurality disks and, for each identified disk, if a mismatch occurs between the ownership information on the predetermined area of the disk and the ownership defined by the SCSI reservation tag, then using the ownership information written to the predetermined area of the disk as definite ownership data without requiring the owned network device to send a second SCSI reservation tag and each network device is configured with a table and to store entries in a table, wherein each entry identifies an owned disk of the network device storing the table; and

a second network device configured to identify all disks owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each disk, to set a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state, and to remove ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk to an unowned state.

7. The network storage system of claim 6, wherein the small computer system interface reservation tag is a small computer system interface level 3 persistent reservation tag.

8. The networked storage system of claim 6, wherein the small computer system interface reservation tag is set such that only the network device may write to the disk.

9. The network storage system of claim 6, wherein the first ownership attribute further comprises a serial number of the network device that owns that particular disk.

10. The network storage system of claim 6, wherein each of the plurality of file servers can read data from each of the plurality of disks.

11. The network storage system of claim 6, wherein only a network device that owns one of the plurality of disks can write data to the one disk.

12. The network storage system of claim 6, wherein the network devices comprise file servers.

13. A network storage system comprising:

a plurality of network devices;

one or more switches;

a plurality of disks; and

means for writing ownership information to a predetermined area of each disk of the plurality of disks, wherein the predetermined area of the disk is sector 0 on the disk and the ownership information stored in sector 0 is definitive ownership data for determining ownership of the disk;

means for setting a small computer system interface (SCSI) reservation tag of each disk to provide a two part indicia of ownership, where the two part indicia of ownership are written to each disk;

means for creating a table on each network device in the network storage system;

means for identifying all disks owned by each network device using ownership information written to the predetermined area of each disk of the plurality disks and, for each identified disk, if a mismatch occurs between the ownership information on the predetermined area of the disk and the ownership defined by the SCSI reservation tag, then using the ownership information written to the predetermined area of the disk as definite ownership data without requiring the owned network device to send a second SCSI reservation tag;

in response to identifying, means for storing entries in the table, wherein each entry identifies an owned disk of the network device storing the table;

means for identifying, by a second network device, all disks owned by a first network device in reponse to a failure of the first network device, wherein each network device can read ownership information of each disk

means for setting a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state; and

means for removing ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk into the unowned state.

14. The network storage system of claim 13, wherein the network devices comprise file servers.

15. A network storage system comprising:

one or more switches interconnected to form a switching fabric;

a plurality of disks, each of the disks connected to at least one of the switches, each disk storing a first ownership attribute to a predetermined area of a disk and each disk associated with a second ownership attribute in the form of a small computer system interface reservation, wherein the predetermined area of the disk stores definitive ownership data for determining ownership of the disk and the small computer system interface reservation allows other network devices to read the ownership attribute from the disks;

one or more network devices, interconnected with the switching fabric, each of the network devices being configured to own a predetermined set of disks of the plurality of disks though use of the first and second ownership attributes, wherein each network device identifies all disks owned by the network device using ownership information written to the predetermined area of each disk of the plurality disks and, for each identified disk, if a mismatch occurs between the ownership information on the predetermined area of the disk and the ownership defined by the SCSI reservation tag, then using the ownership information written to the predetermined area of the disk as definite ownership data without requiring the owned network device to send a second SCSI reservation tag and each network device is configured with a table and to store entries in a table, wherein each entry identifies an owned disk of the network device storing the table; and

a second network device configured to identify all disks owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each disk, to set a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state, and to remove ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk into the unowned state.

16. The network storage system of claim 15, wherein the first ownership attribute further comprises a serial number of one of the one or more network devices.

17. The network storage system of claim 15, wherein the small computer system interface reservation is a small computer system interface level 3 persistent reservation.

18. The network storage system of claim 15, wherein each of the network devices further comprises a disk ownership table, the disk ownership table containing ownership data for each of the disks.

19. The network storage system of claim 18, wherein the ownership table further comprises a world wide name for each of the disks, the world wide name being used for identification of each of the disks.

20. A computer-readable storage medium containing executable program instructions executed by a processor, comprising:

program instructions that write ownership information to a predetermined area of a disk, wherein the predetermined area of the disk stores definitive ownership data for determining ownership of the disk;

program instructions that set a small computer system interface reservation tag for the disk to a state of network device ownership to provide a two part indicia of ownership for the disk, where the two part indicia of ownership are both written to the disk and the small computer system interface reservation tag allows other network devices to read the ownership information from the disks;

program instructions that create a table on each network device in the network storage system;

program instructions that identify all disks owned by the network device using ownership information written to the predetermined area of each disk of the plurality disks and, for each identified disk, if a mismatch occurs between the ownership information on the predetermined area of the disk and the ownership defined by the SCSI reservation tag, then using the ownership information written to the predetermined area of the disk as definite ownership data without requiring the owned network device to send a second SCSI reservation tag;

in response to identifying, program instructions that store entries in the table, wherein each entry identifies an owned disk of the network device storing the table;

program instructions that identify, by a second network device, all disks owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each disk;

program instructions that set a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state; and

program instructions that remove ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk into the unowned state.

21. A method for a network device to manage ownership of one or more storage devices in a network storage system, comprising:

reading ownership information from a predetermined area of each storage device, wherein the predetermined area of each storage device is sector 0 on the disk and the ownership information stored in sector 0 is definitive ownership data for determining ownership of the storage device;

in response to reading the ownership information, creating an ownership table that stores entries where each entry identifies a storage device owned by the network device, wherein the ownership is stored within the network device;

reading a small computer system interface (SCSI) reservation tag from each storage device, wherein the SCSI reservation tag allows other network devices to read the ownership information from each storage device;

comparing the SCSI reservation tag to the ownership information of the same storage device and, if there is not a match, changing the SCSI reservation tag to match the ownership information;

configuring the one or more storage devices identified in the ownership table into at least one volume for use by the network device;

identifying, by a second network device, all storage devices owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each storage device;

setting a SCSI release tag for each storage device owned by the first network device, in response to the failure of the first network device, to transfer the storage device to an unowned state; and

removing ownership information stored in the predetermined area of each storage device owned by the first network device to complete transferring each storage device into the unowned state.

22. The method of claim 21 further comprising:

setting ownership information at the predetermined area of each storage device.

23. The method of claim 21 wherein the step of configuring further comprises:

organizing the one or more storage devices into at least one Redundant Array of Independent Disks (RAID) group.

24. The method of claim 21 further comprising:

wherein the predetermined area of the one or more storage devices is sector zero of the one or more storage devices.

25. The method of claim 21 further comprising:

wherein the ownership information is a serial number of the network device that owns that particular storage device.

26. The method of claim 21 further comprising:

wherein the ownership table includes a world wide name for each of the storage devices, the world wide name being used to identify each of the storage devices.

27. A network device for managing ownership of one or more storage devices in a network storage system, comprising:

means for reading ownership information from a predetermined area of each storage device, wherein the predetermined area of each storage device stores definitive ownership information for determining ownership of the storage device;

in response to reading the ownership information, means for creating an ownership table that stores entries where each entry identifies a storage device owned by the network device, wherein the ownership is stored within the network device;

means for reading a small computer system interface (SCSI) reservation tag from each storage device, wherein the SCSI reservation tag allows other network devices to read the ownership information from each storage device;

means for comparing the SCSI reservation tag to the ownership information of the same storage device and, if there is not a match, changing the SCSI reservation tag to match the ownership information;

means for configuring the one or more storage devices identified in the ownership table into at least one volume for use by the network device;

means for identifying, by a second network device, all storage devices owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each storage device;

means for setting a SCSI release tag for each storage device owned by the first network device, in response to the failure of the first network device, to transfer the storage device to an unowned state; and

means for removing ownership information stored in the predetermined area of each storage device owned by the first network device to complete transferring each storage device into the unowned state.

28. A computer readable storage medium containing executable program instructions executed by a processor, comprising:

program instructions that read ownership information from a predetermined area of each storage device of one or more storage devices in a network storage system, wherein the predetermined area of each storage device stores definitive ownership information for determining ownership of the storage device;

in response to reading the ownership information, program instructions that create an ownership table stores entries where each entry identifies a storage devices owned by the network device, wherein the ownership is stored within the network device;

program instructions that read a small computer system interface (SCSI) reservation tag from each storage device, wherein the SCSI reservation tag allows other network devices to read the ownership information from each storage device;

program instructions that compare the SCSI reservation tag to the ownership information of the same storage device and, if there is not a match, changing the SCSI reservation tag to match the ownership information;

program instructions that configure the one or more storage devices identified in the ownership table into at least one volume for use by the network device;

program instructions that identify, by a second network device, all storage devices owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each storage device;

program instructions that set a SCSI release tag for each storage device owned by the first network device, in response to the failure of the first network device, to transfer the storage device to an unowned state; and

program instructions that remove ownership information stored in the predetermined area of each storage device owned by the first network device to complete transferring each storage device into the unowned state.

29. A network storage system, comprising:

one or more storage devices, each storage device having a predetermined area for storing ownership information and each storage device having a small computer system interface (SCSI) reservation tag, wherein the predetermined area of each storage device stores definitive ownership information for determining ownership of the storage device, and the SCSI reservation tag allows other network devices to read the ownership information from each storage device;

at least one network device having an ownership table constructed based upon the ownership information from each storage device, wherein the ownership is stored within the network device;

the at least one network device having an ownership layer for comparing the SCSI reservation tag to the ownership information of the same storage device and, if there is not a match, changing the SCSI reservation tag to match the ownership information; and

the at least one network device having a disk storage layer for configuring the one or more storage devices identified in the ownership table into at least one volume for use by the network device; and

a second network device configured to identify all disks owned by a first network device in response to a failure of the first network device, wherein each network device can read ownership information of each disk, to set a SCSI release tag for each disk owned by the first network device, in response to the failure of the first network device, to transfer the disk to an unowned state, and to remove ownership information stored in the predetermined area of each disk owned by the first network device to complete transferring each disk into the unowned state.

30. The network storage system of claim 29 further comprising:

the ownership layer adapted to set ownership information at the predetermined area of each storage device.

31. The network storage system of claim 29 further comprising:

the disk storage layer organizing the one or more storage devices into at least one Redundant Array of Independent Disks (RAID) group.

32. The network storage system of claim 29 further comprising:

wherein the predetermined area of the one or more storage devices is sector zero of the one or more storage devices.

33. The network storage system of claim 29 further comprising:

wherein the ownership information is a serial number of the network device that owns that particular storage device.

34. The network storage system of claim 29 further comprising:

wherein the ownership table includes a world wide name for each of the storage devices, the world wide name being used to identify each of the storage devices.