Methods and systems for self-healing in connected computing environments
AssigneeNetApp, Inc.
InventorKausik Ghatak
Life of this patent
Estimating the term…
Abstract
Methods and systems for networked systems are provided. A reinforcement learning (RL) agent is deployed during runtime of a networked system having at least a first component and a second component. The RL agent detects a first degradation signal in response to an error associated with the first component and a second degradation signal from the second component, the second degradation signal generated in response to the error. The RL agent identifies from a learned data structure an action for fixing degradation, at both the first component and the second component; and continues to update the learned data structure, upon successful and unsuccessful attempts to fix degradation associated with the first component and the second component.
Claims20
A method, comprising:
- representing, by a processor, a plurality of resources of a networked system by a plurality of interconnected nodes of a graphical structure, each node representing a resource, and a graphical edge between connected nodes representing an effect of a source node on another node;
- associating, by the processor, a policy to each node, the policy including an influence factor indicating an impact of a degradation of a first node on a second node and an indicator for triggering an action based on a severity level of a degradation signal impacting each node;
- generating, by the processor, a data structure during a simulation phase for storing one or more potential solution corresponding to an injected error associated with the first node that results in degradation of the second node, the degradation based on the influence factor corresponding to the first node and the potential solution triggered based on the policy associated with the first and second node;
- detecting, by the processor, an error similar to the injected error, during runtime of the networked system;
- and
- selecting, by the processor, the one or more potential solution from the data structure based on a maximum expected return to self-heal the first node and the second node.
A non-transitory machine-readable storage medium having stored thereon instructions for performing a method, comprising machine executable code which when executed by at least one machine, causes the machine to:
- represent a plurality of resources of a networked system by a plurality of interconnected nodes of a graphical structure, each node representing a resource, and a graphical edge between connected nodes representing an effect of a source node on another node;
- associate a policy to each node, the policy including an influence factor indicating an impact of a degradation of a first node on a second node and an indicator for triggering an action based on a severity level of a degradation signal impacting each node;
- generate a data structure during a simulation phase for storing one or more potential solution corresponding to an injected error associated with the first node that results in degradation of the second node, the degradation based on the influence factor corresponding to the first node and the potential solution triggered based on the policy associated with the first and second node;
- detect an error similar to the injected error, during runtime of the networked system;
- and
- select the one or more potential solution from the data structure based on a maximum expected return to self-heal the first node and the second node.
A system, comprising:
- a memory containing machine readable medium comprising machine executable code having stored thereon instructions;
- and a processor coupled to the memory to execute the machine executable code to:
- represent a plurality of resources of a networked system by a plurality of interconnected nodes of a graphical structure, each node representing a resource, and a graphical edge between connected nodes representing an effect of a source node on another node;
- associate a policy to each node, the policy including an influence factor indicating an impact of a degradation of a first node on a second node and an indicator for triggering an action based on a severity level of a degradation signal impacting each node;
- generate a data structure during a simulation phase for storing one or more potential solution corresponding to an injected error associated with the first node that results in degradation of the second node, the degradation based on the influence factor corresponding to the first node and the potential solution triggered based on the policy associated with the first and second node;
- detect an error similar to the injected error, during runtime of the networked system;
- and
- select the one or more potential solution from the data structure based on a maximum expected return to self-heal the first node and the second node.
Dependencies are inferred from the claim text.