Method and system for coordinated multiple cluster failover

• US 8,429,450 B2
• Filed: 05/28/2010
• Issued: 04/23/2013
• Est. Priority Date: 04/04/2007
• Status: Expired due to Fees

• Current Assignee: Vision Solutions Incorporated (Precisely Software Incorporated)
• Original Assignee: Vision Solutions Incorporated (Precisely Software Incorporated)
• Inventors: Brown, David E., Simpson, Scott
• Primary Examiner(s): CHU, GABRIEL L
• Application Number:
  US12/790,628
• Publication Number:
  US 20100241896A1
• Time in Patent Office: 1,061 Days
• Field of Search: None
• US Class Current: 714/13
• CPC Class Codes: G06F 11/1482 by means of middleware or OS functionality
• CPC Class Codes: G06F 11/2033 switching over of hardware resources
• CPC Class Codes: G06F 11/2035 without idle spare hardware
• CPC Class Codes: G06F 11/2038 with a single idle spare processing component
• CPC Class Codes: G06F 11/2041 with more than one idle spare processing component
• CPC Class Codes: G06F 11/2046 where the redundant components share persistent storage G06F11/2043 takes precedence

Abstract

Hyperclusters are a cluster of clusters. Each cluster has associated with it one or more resource groups, and independent node failures within the clusters are handled by platform specific clustering software. The management of coordinated failovers across dependent or independent resources running on heterogeneous platforms is contemplated. A hypercluster manager running on all of the nodes in a cluster communicates with platform specific clustering software regarding any failure conditions, and utilizing a rule-based decision making system, determines actions to take on the node. A plug-in extends exit points definable in non-hypercluster clustering technologies. The failure notification is passed to other affected resource groups in the hypercluster.

Claims

1. A method for coordinating availability of data processing resources between a first cluster of nodes each controlled by a respective first cluster manager and multiple other clusters of nodes each controlled by a respective second cluster manager, the method comprising:

receiving a disruption signal from an exit program of one of the first cluster managers, the disruption signal being representative of a disruption event associated with a specific one of the nodes of the first cluster, the disruption signal being received by a first hypercluster manager of the specific one of the nodes of the first cluster;

deriving a local action code and a hypercluster event code from a hypercluster rules list, the local action code and the hypercluster event code corresponding to the disruption event, the local action code having an associated local cluster activation sequence for regulating the operation of the first cluster of nodes, and the hypercluster event code having an associated cluster activation sequence for regulating the operation of the respective multiple other clusters of nodes;

transmitting the hypercluster event code to the multiple other clusters of nodes, each of the nodes of the multiple other clusters of nodes including a second hypercluster manager for execution of a remote cluster activation sequence thereon;

receiving a token on the first hypercluster manager transmitted thereto in response to a completed execution of the remote cluster activation sequence on a one of the node of the multiple other clusters of nodes; and

executing the local cluster activation sequence on the first cluster of nodes upon receipt of the token;

wherein the first cluster of nodes and the multiple other clusters of nodes each function independent of each other, the nodes of the first cluster of nodes communicate with each other by a first set of messages, and the nodes of a given one of the multiple other clusters of nodes communicate with each other by a second set of messages.

2. The method of claim 1, further comprising:

synchronizing views of the first and the multiple other clusters of nodes amongst each of the first and the multiple other clusters, view synchrony being determined by another token.

3. The method of claim 2, wherein deriving the hypercluster event code includes:

translating the disruption event to a universal event code with a translation table, the translation table including a first sequence of disruption events and a second sequence of universal event codes correlated thereto.

4. The method of claim 3, wherein the universal event code is referenced to derive the hypercluster event code from the hypercluster rules list.

5. The method of claim 2, wherein the local and remote cluster activation sequences include dependencies therebetween, the dependencies establishing the timing and order of the local and remote cluster activation sequences.

6. The method of claim 5, wherein transmitting the hypercluster event code to the other clusters of nodes is in response to receiving a confirmation code representative of completion of one step in the cluster activation sequence as defined by the dependencies.

7. The method of claim 2, wherein the remote cluster activation sequence is being executed on one of the nodes of the multiple other clusters.