Apparatus and method for handling partially inconsistent states among members of a cluster in an erratic storage network

• US 8,793,527 B1
• Filed: 02/28/2013
• Issued: 07/29/2014
• Est. Priority Date: 02/28/2013
• Status: Active Grant

• Current Assignee: Peaxy, Inc.
• Original Assignee: Peaxy, Inc.
• Inventors: Franks, John
• Primary Examiner(s): Baderman, Scott
• Assistant Examiner(s): BRYAN, JASON B
• Application Number:
  US13/780,952
• Time in Patent Office: 516 Days
• US Class Current: 714/4.1
• CPC Class Codes: G06F 16/2365 Ensuring data consistency and integrity
• CPC Class Codes: G06F 3/0611 in relation to response time
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• 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 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

A method, for handling partially inconsistent states among members of a cluster in an erratic storage network that responds to requests from a client, includes assigning a unique commit identifier corresponding to each successive modification of state of the cluster when an update request has been successfully completed and correlating an incoming request to a given member of the cluster with any commit identifiers necessary to satisfy the request. The method further includes detecting, on the incoming request to the given member of the cluster, whether the commit identifiers necessary to satisfy the request have been applied to the given member, and processing the incoming request based on whether the commit identifiers necessary to satisfy the request have been applied to the given member. Operation of the members can be asynchronous.

Claims

1. A method for handling partially inconsistent states among members of a cluster in an erratic storage network that responds to requests from a client, the method comprising:

assigning a unique commit identifier corresponding to each successive modification of state of the cluster when an update request has been successfully completed;

correlating an incoming request with a given member of the cluster with any commit identifiers necessary to satisfy the request;

detecting, on the incoming request to the given member of the cluster, whether the commit identifiers necessary to satisfy the request have been applied to the given member; and

if the commit identifiers necessary to satisfy the request have been applied to the given member, then satisfying the incoming request;

if the commit identifiers necessary to satisfy the request have not been applied to the given member, then using a set of selection criteria to determine which one or more of the following processes shall be performed;

(i) asynchronously causing one or more other members of the cluster, having commit identifiers necessary to satisfy the request, to satisfy the request;

(ii) asynchronously causing one or more other members to provide to the given member the missing commit identifiers to satisfy the request, so that the given member can satisfy the request; and

(iii) waiting until the given member has been caused to receive the missing commit identifiers to satisfy the request, so that the given member can satisfy the request; and

performing the selected one or more processes.

2. The method for handling partially inconsistent states according to claim 1, wherein

if the commit identifiers necessary to satisfy the request have not been applied to the given member, then performing both

(i) asynchronously causing one or more other members of the cluster, having commit identifiers necessary to satisfy the request, to satisfy the request;

and

(ii) asynchronously causing one or more other members to provide to the given member the missing commit identifiers to satisfy the request, so that the given member can satisfy the request.

3. The method according to claim 1, wherein the set of selection criteria includes one or more of the following criteria associated with the given member of the cluster or any other member of the cluster:

storage unit activity, processor load, network bandwidth available, memory available, thermal conditions, and electrical power consumption, and configuration settings.