Fault Tolerant Distributed Computing

In a distributed computing model each peer is given authority for a portion of the data. In essence, the computing workload is distributed to all peers in the network and thus the workload on each computing device is kept to a minimum. Adding fault tolerance to distributed computing allows it to detect hackers. One such fault tolerant method is a triple authority system, in which three peers perform the same calculations. This allows it to identify the peer being hacked in real-time, and correcting the hacked data in real-time.

Authority assignment is performed dynamically at start-up or can be pre-assigned. When data is received from a computing device, the data is compared with all other authorities of the data. If the authorities do not agree then a conflict is raised for further processing. Depending on the policy settings several conflicts may need to be received successfully for a set duration before the sending device is considered to be malfunctioning. Once a device is found to be malfunctioning, policy rules are applied to either take the device offline or reassign authority within the network. This essentially heals the network as the malfunction device is no longer in a position of authority and cannot cause any harm.

In essence, fault tolerant distributed computing addresses security by dynamically assigning authority, detecting conflict, self-correcting and by removing conflicting computing device from a position of authority.

The network engine is modified so that the fault tolerant distributed computing model can coexist with the client-server model. This allows the developer to dynamically select at run-time the client-server model or fault tolerant distributed computing model.

For more information contact us at info@akimbotechnologies.com.


<< Technology Overview   NAT Traversal >>