FABRIC ENCAPSULATED RESILIENT STORAGE

• US 20180007134A1
• Filed: 06/30/2016
• Published: 01/04/2018
• Est. Priority Date: 06/30/2016
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: GUIM BERNAT, Francesc, RIVAS BARRAGAN, Daniel, DOSHI, Kshitij A., SCHMISSEUR, Mark A., LARSEN, Steen
• Granted Patent: US 10,785,295 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 3/0605 by facilitating the interaction with a user or administrator
• CPC Class Codes: G06F 3/0617 in relation to availability
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0644 Management of space entities, e.g. partitions, extents, pools
• CPC Class Codes: G06F 3/0649 Lifecycle management
• 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 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays
• CPC Class Codes: H04L 49/356 for storage area networks
• CPC Class Codes: H04L 67/1095 Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
• 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 67/303 Terminal profiles

Abstract

Fabric encapsulated resilient storage is hardware-assisted resilient storage in which the reliability capabilities of a storage server are abstracted and managed transparently by a host fabric interface (HFI) to a switch. The switch abstracts the reliability capabilities of a storage server into a level of resilience in a hierarchy of levels of resilience. The resilience levels are accessible by clients as a quantifiable characteristic of the storage server. The resilience levels are used by the switch fabric to filter which storage servers store objects responsive to client requests to store objects at a specified level of resilience.

Claims

1. A computer-implemented method comprising:

in a switch having access to storage nodes and client nodes;

abstracting characteristics of each storage node into a level of resilience in a hierarchy of levels of resilience, the abstracted level of resilience representing how reliably each storage node stores objects;

causing an object to be stored on one or more of the storage nodes based on their respective abstracted levels of resilience.

2. The computer-implemented method of claim 1, wherein causing the object to be stored on one or more of the storage nodes based on their respective abstracted levels of resilience includes:

receiving from the client node a request to store the object at a specified level of resilience;

determining which storage nodes store objects at the specified level of resilience; and

initiating a replication flow of messages to cause the object to be stored on any one or more of the storage nodes that store objects at the specified level of resilience.

3. The computer-implemented method of claim 2, further comprising, in the switch, configuring a host fabric interface (HFI) with the hierarchy of levels of resilience, wherein the level of resilience specified in the request is obtained through the HFI.

4. The computer-implemented method of claim 2, further comprising, in the switch:

determining none of the storage nodes store objects at the specified level of resilience; and

forwarding the request to another switch configured to access other storage nodes that store objects at the specified level of resilience.

5. The computer-implemented method of claim 2, further comprising, in the switch:

receiving the request from a storage node unable to store the object at the specified level of resilience.

6. The computer-implemented method of claim 1, further comprising, in the switch:

configuring a granularity of the hierarchy of levels of resilience into which characteristics of each storage node are abstracted, wherein hierarchies configured with greater granularity cause the storage characteristics to be abstracted into a greater number of levels of resilience than hierarchies configured with lesser granularity.

7. The computer-implemented method of claim 1, wherein how reliably the storage node stores objects represents one or more of an assurance of availability of stored objects and a protection against loss of stored objects, the assurance of availability and the protection against loss inferred from the storage characteristics of the storage node.

8. The computer-implemented method of claim 1, wherein:

the characteristics include any one or more of a static characteristic and a dynamic characteristic; and

further wherein the static characteristic is any one or more of a predefined reliability factor quantifying a reliability profile of the storage node, a number of disks comprising the storage node, and a type of disk technology used by the storage node, and the dynamic characteristic is any one or more of a load and an availability of the storage node over time.

9. The computer-implemented method of claim 1, further comprising, in the switch:

registering the characteristics of the storage nodes in a memory of the switch; and

updating the registered characteristics in the memory to reflect changes in the characteristics of the storage nodes.

10. A system comprising:

a switch having a memory configured to store characteristics of storage servers accessible through the switch;

a processor on the switch configured to abstract the stored characteristics of each storage server into a level of resilience in a hierarchy of levels of resilience, the abstracted level of resilience representing how reliably each storage node stores objects; and

a fabric on the switch configured to cause objects to be stored on one or more of the storage servers based on their respective abstracted levels of resilience.

11. The system of claim 10, further comprising:

a host fabric interface (HFI) to the switch configured to expose the hierarchy of levels of resilience to clients in communication with the switch; and

an ingress queue in the switch to receive requests from the clients to store objects;

the processor configured to determine which storage servers store objects at a requested level of resilience responsive to receiving a request to store an object at the requested level of resilience; and

the fabric initiating a replication flow of messages to cause the object to be stored on any one or more of the storage servers determined to store objects at the requested level of resilience.

12. The system of claim 11, further comprising:

the processor determining that none of the storage servers accessible through the switch store objects at the requested level of resilience; and

the fabric forwarding the request to another switch configured to access other storage servers that store objects at the specified level of resilience.

13. The system of claim 11, wherein the ingress queue is to receive the requests from a remote storage server unable to store objects at the requested level of resilience.

14. The system of claim 10, wherein the hierarchy of levels of resilience into which the processor abstracts the characteristics of each storage node is configurable to provide clients with varied levels of resilient storage.

15. The system of claim 10, wherein how reliably each storage server stores objects represents one or more of an assurance of availability of stored objects and a protection against loss of stored objects, the assurance of availability and the protection against loss inferred from the storage characteristics of the storage server.

16. The system of claim 10, wherein:

the characteristics include any one or more of a static characteristic and a dynamic characteristic; and

further wherein the static characteristic is any one or more of a predefined reliability factor quantifying a reliability profile of the storage node, a number of disks comprising the storage node, and a type of disk technology used by the storage node, and the dynamic characteristic is any one or more of a load and an availability of the storage node over time.

17. The system of claim 10, further comprising:

the processor configured to update the characteristics stored in the switch'"'"'s memory to reflect changes in the characteristics of the storage servers.

18. At least one computer readable storage medium including instructions that, when executed on a machine, cause the machine to:

abstract characteristics of a storage server in communication with the machine into a level of resilience in a hierarchy of levels of resilience, the abstracted level of resilience representing how reliably the storage server stores objects; and

store an object the storage server based on the abstracted level of resilience.

19. The at least one computer readable storage medium of claim 18, wherein to store the object on the storage server based on the abstracted level of resilience includes:

to receive from a client in communication with the machine a request to store the object at a requested level of resilience;

to determine whether the storage server stores objects at the requested level of resilience;

to initiate a replication flow of messages to the storage server that cause the storage server to store the object at the requested level of resilience.

20. The at least one computer readable storage medium of claim 18, wherein the characteristics include any one or more of a static characteristic and a dynamic characteristic;

and

further wherein the static characteristic is any one or more of a predefined reliability factor quantifying a reliability profile of the storage server, a number of disks comprising the storage server, and a type of disk technology used by the storage server, and the dynamic characteristic is any one or more of a load and an availability of the storage server over time.