×

Fault tolerant multi-node computing system using periodically fetched configuration status data to detect an abnormal node

  • US 7,870,439 B2
  • Filed: 05/26/2004
  • Issued: 01/11/2011
  • Est. Priority Date: 05/28/2003
  • Status: Active Grant
First Claim
Patent Images

1. A fault tolerant computing system comprising:

  • a plurality of processing nodes interconnected by a communication medium, wherein said processing nodes are configured either in a substantially equal configurations for parallel-running a plurality of uniquely different versions of identical application programs or in a plurality of uniquely different configurations for parallel-running a plurality of substantially equal versions of identical application programs; and

    a fault detector connected to said processing nodes via said communication medium for periodically collecting configuration status data from said processing nodes and mutually verifying the configuration status data of said processing nodes with each other for detecting an abnormal node whose operating state is beyond a range of normal deviations of the uniquely different configuration of the node, wherein said fault detector comprises means for detecting configuration status data whose value differs significantly from a data set, based on a statistical test, and formed by the configuration status data of all of said processing nodes and identifying one of said processing nodes which provides the detected configuration status data, andwherein the configuration status data includes, at least, information regarding the processing node'"'"'s operating state and memory usage.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×