Self healing grid architecture for decentralized component-based systems
View Patent ↗A self-healing and self-optimizing grid architecture can be provided in accordance with the present invention. Specifically, the architecture can include a mechanism for detecting component failures, and even degraded component performance, within peer components in a hosting service. Once a failure has been detected, the detecting peer to undertake remedial action to recreate and redeploy the component in the hosting system. In particular, the detecting component can acquire the behavior of the failed component and the detecting component can instantiate an instance of the behavior in another server in the grid. Thus, the mechanism described herein can be analogized to biotechnical DNA as every component in the hosting service can maintain an awareness of the state of the entire system and can recreate the entire system through knowledge provided by grid services DNA.
1. A method of self-healing in a Web services grid comprising a plurality of hosting service components, said method comprising the steps of:
detecting in at least one of the hosting service components a failure of a peer hosting service component;
loading a grid services distributed network agreement (DNA), said grid services DNA specifying sufficient resource data necessary to redeploy any one failed hosting service component in the Web services grid; and,
redeploying said failed peer hosting service component based upon an associated behavior specified in said grid services DNA, said redeploying step comprising
alerting other peer hosting service components in the Web services grid of said detected failure;
including with said alert a timestamp;
receiving acknowledgments of said alert from said other peer hosting service components;
computing a lowest timestamp among any timestamps included in said acknowledgments; and,
if said time stamp included with said alert is computed to be said lowest timestamp, performing said redeploying step, but if said time stamp included with said alert is computed not to be said lowest timestamp, suppressing said redeploying step.
2. The method of claim 1 , wherein said redeploying step comprises the steps of:
deploying said failed peer component in a new physical location other than the physical location in which said failed peer component failed, said deployed failed peer component retrieving its prior state from said grid services DNA;
updating said grid services DNA to specify said new physical location; and, forwarding said updated grid services DNA to others of said hosting service components.
3. A method of self-healing in a Web services grid comprising a plurality of hosting service components, comprising the steps of:
detecting in at least one of the hosting service components a failure of a peer hosting service component;
loading a grid services distributed network agreement (DNA), said grid services DNA specifying sufficient resource data necessary to redeploy any one failed hosting service component in the Web services grid;
redeploying said failed peer hosting service component based upon an associated behavior specified in said grid services DNA; and,
serializing state information for each of the hosting service components to fixed storage at a location specified by said grid services DNA.
4. A machine readable storage having stored thereon a computer program for self-healing in a Web services grid comprising a plurality of hosting service components, said computer program comprising a routine set of instructions which when executed cause the machine to perform, the steps of:
detecting in at least one of the hosting service components a failure of a peer hosting service component;
loading a grid services distributed network agreement (DNA), said grid services DNA specifying sufficient resource data necessary to redeploy any one failed hosting service component in the Web services grid; and,
redeploying said failed peer hosting service component based upon an associated behavior specified in said grid services DNA, said redeploying step comprising
alerting other peer hosting service components in the Web services grid of said detected failure;
including with said alert a timestamp;
receiving acknowledgments of said alert from said other peer hosting service components;
computing a lowest timestamp among any timestamps included in said acknowledgments; and,
if said time stamp included with said alert is computed to be said lowest timestamp, performing said redeploying step, but if said time stamp included with said alert is computed not to be said lowest timestamp, suppressing said redeploying step.
5. The machine readable storage of claim 4 , wherein said redeploying step comprises the steps of:
deploying said failed peer component in a new physical location other than the physical location in which said failed peer component failed, said deployed failed peer component retrieving its prior state from said grid services DNA;
updating said grid services DNA to specify said new physical location; and,
forwarding said updated grid services DNA to others of said hosting service components.