CONSISTENT DISTRIBUTED STORAGE COMMUNICATION PROTOCOL SEMANTICS IN A CLUSTERED STORAGE SYSTEM

• US 20140281065A1
• Filed: 05/28/2014
• Published: 09/18/2014
• Est. Priority Date: 09/23/2011
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Patnaik, Pranab, Tan, Kai, Venkatesan, Vivek
• Granted Patent: US 9,256,377 B2
• Time in Patent Office: Days
• US Class Current: 710/74
• CPC Class Codes: G06F 12/0646 Configuration or reconfiguration
• CPC Class Codes: G06F 3/0605 by facilitating the interaction with a user or administrator
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0631 by allocating resources to storage systems
• CPC Class Codes: G06F 3/0665 at area level, e.g. provisioning of virtual or logical volumes
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 3/0683 Plurality of storage devices
• 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]

Abstract

Consistent distributed storage communication protocol semantics, such as SCSI target semantics, in a clustered storage system are disclosed. The system includes a mechanism for presenting a single distributed logical unit, comprising one or more logical sub-units, as a single logical unit of storage to a host system by associating each of the logical sub-units that make up the single distributed logical unit with a single host visible identifier that corresponds to the single distributed logical unit. The system further includes a mechanism to maintain consistent context information for each of the logical sub-units such that the logical sub-units are not visible to a host system as separate entities from the single distributed logical unit.

Claims

1. A method comprising:

in a clustered storage system that has multiple storage server nodes, implementing a first logical sub-unit associated with at least a data storage object in a first storage server node;

maintaining, for the first logical sub-unit, context information that is consistent with corresponding context information of a second logical sub-unit implemented in a second storage server node such that the first and second logical sub-units are part of a single distributed logical unit collectively made up by at least the first and second logical sub-units; and

associating the first and second logical sub-units with a single host-visible identifier of the single distributed logical unit so that a host can access either the first or the second logical sub-unit of the single distributed logical unit, such that the first and second logical sub-units are not visible to the host as entities separate from the single distributed logical unit.

2. The method of claim 1, wherein maintaining context information that is consistent with the corresponding context information includes:

implementing a first instance of a global task set for the first logical sub-unit of the single distributed logical unit to queue commands or task management functions that modify the context information;

coordinating entry of a first command into the first instance of the global task set such that the first instance of the global task set is consistent with a second instance of the global task set implemented for the second logical sub-unit;

processing a command from the first instance of the global task set; and

updating the context information for the first logical sub-unit to reflect the processing of the command.

3. The method of claim 1, wherein maintaining the context information includes coordinating a command from the host to both the first storage server node and the second storage server node to ensure consistency of the context information in the first logical sub-unit and the second logical sub-unit.

4. The method of claim 1, wherein the context information includes at least one of logical unit state information, logical unit configuration information, or logical unit metadata.

5. The method of claim 1, wherein the context information includes information to indicate whether the single distributed logical unit supports ALUA or not.

6. The method of claim 1 further comprising, detecting corruption of the context information associated with the first logical sub-unit and quarantining the first logical sub-unit by suspending communication to the first logical sub-unit until the first logical sub-unit recovers from the corruption.

7. The method of claim 1 further comprising:

implementing multiple virtual target ports, wherein one of the virtual target ports is associated with a first physical target port of the first storage server node and another one of the virtual target ports is associated with a second physical target port of the second storage server node; and

presenting the multiple virtual target ports to the host.

8. The method of claim 7, wherein a first target device among multiple target devices implemented in the clustered storage system is associated with one or more of the multiple virtual target ports.

9. A storage server node in a clustered storage system comprising:

an interface to detachably couple with an interconnect to enable communication amongst multiple storage server nodes of the clustered storage system;

a management module to enable management operations to be performed on the storage server node;

a data module to provide data storage services for the storage server node; and

a protocol module to allow a host to access the clustered storage system, the protocol module configured to implement a logical sub-unit of a single distributed logical unit and to associate the logical sub-unit with an identifier of the single distributed logical unit so that the host can perform a block-level access to the logical sub-unit of the single distributed logical unit, such that the logical sub-unit is not visible to the host as a separate entity from the single distributed logical unit.

10. The storage server node of claim 9, wherein the protocol module is further configured to maintain context information that is consistent with corresponding context information of another logical sub-unit implemented in another storage server node in the clustered storage system.

11. The storage server node of claim 10, wherein, to maintain the context information that is consistent with the another logical sub-unit, the protocol module is configured to:

implement a first instance of a global task set to queue commands that modify the context information;

coordinate entry of a first command into both the first instance and at least a second instance of the global task set implemented on the other storage server node to ensure the context information is consistent with the corresponding context information in the other storage server node after the first command is processed;

process the first command from the instance of the global task set; and

update the context information for the logical sub-unit to reflect the processing of the command.

12. The storage server node of claim 9, wherein the protocol module is further configured to generate a map that associates each logical sub-unit that collectively make up the single distributed logical unit with a set of storage objects, wherein the single distributed logical unit is divided into a number of fixed size logical blocks and wherein the block-level access enables the host to access at a granularity of a logical block.

13. The storage server node of claim 10, wherein the context information includes at least one of logical unit state information, logical unit configuration information, or logical unit metadata.

14. The storage server node of claim 10, wherein the protocol module is further configured to detect corruption of the context information associated with the logical sub-unit and quarantine the logical sub-unit by suspending communication with the logical sub-unit until the logical sub-unit recovers from the corruption.

15. The storage server node of claim 9, wherein the protocol module is further configured to:

implement multiple virtual SCSI target ports, wherein at least one of the multiple virtual SCSI target ports is associated with a physical target port of the storage server node; and

present the multiple SCSI target ports to the host.

16. The storage server node of claim 9 further comprising, multiple virtual storage servers configured to operate in a SAN mode.

17. The storage server node of claim 16, wherein the protocol module is further configured to:

implement multiple virtual SCSI target ports, wherein at least one of the multiple virtual SCSI target ports is associated with a virtual storage server of the multiple virtual storage servers.

18. A method comprising:

in a first storage server node in a clustered storage system that includes multiple storage server nodes implementing multiple logical sub-units, implementing a first logical sub-unit and a first task sub-manager for the first logical sub-unit;

receiving a command through a target port of the first storage server node by the first task sub-manager for the first logical sub-unit;

maintaining, for each of the logical sub-units by the first task sub-manager, context information that is consistent with corresponding context information of each of other logical sub-units such that the logical sub-units of the multiple storage server nodes collectively make up a single distributed logical unit;

wherein, the maintaining includes;

requesting the command to be queued in task sets of the logical sub-units in the multiple storage server nodes;

processing the command;

updating the context information associated with the processing of the command to the logical sub-units.

19. The method of claim 18, further comprising:

associating the single distributed logical unit with a target device;

creating multiple virtual target ports associated with the target device, wherein each of the multiple storage server nodes includes a virtual target port; and

presenting, on each of the multiple storage server nodes, the multiple virtual target ports to a host.

20. The method of claim 18 further comprising, generating a map associating a set of storage objects with each logical sub-unit of the single distributed logical unit.