Reactive Hybrid Model for Fault Mitigation in Real-Time Cloud Computing

Materials And Methods

With high demand for cloud services and resources by users who migrated to the cloud for high level of productivity, cloud computing systems have experienced an increase in outages or failures particularly in real-time cloud computing. However, researchers have applied different fault tolerant techniques to mitigate the effect of failure. Some of these researches are not without their limitations.

Paul and Visuwasum (2012), worked on the research titled “Checkpoint-based Intelligent Fault Tolerance for Cloud Service Providers.” The research was motivated by the need to achieve reliability for real time computing on Cloud Infrastructure. The research objective was to propose a smart checkpoint infrastructure for virtualized service provider and fault tolerance model for real time computing. In the methodology, checkpoints are stored in Hadoop distributed files system as it allows fast resumption of task execution after a node crash. The method also improves fault tolerance since the checkpoints were distributed and replicated in all the nodes of the service provider. However, the research failed to consider the any form of check point overheads. That is the cost, time, and frequency of replicating checkpoints in all the nodes of the service provider.

In the work of Kalanirnika and Sivagami (2015), ‘Fault tolerance in cloud using proactive and reactive techniques’, the research was motivated by the need to provide a more reliable cloud computing system. The researchers proposed a reactive fault tolerance technique that uses check pointing to tolerate faults. The methodology involved using VM-µ a virtual machine framework to tolerate transient errors.