Fault Tolerance in Cloud Using Reactive and Proactive Techniques

Fault tolerance plays a vital role in ensuring high serviceability and reliability in cloud. A lot of research is currently under way to analyze how cloud can provide fault tolerance for an application. The work proposes a reactive fault tolerant technique that uses check pointing to tolerate the fault. The work proposes a VMμ Checkpoint framework to protect both VMs and applications in the VMs against transient errors. The VM-μCheckpoint mechanism is implemented using CoW-PC (Copy on Write – Presave in cache) algorithm. The CoW-PC algorithm presaves all the tasks running on the VM’s in a cache memory. When there is any transient failure happening in VMs, it is noted and it is recovered using last presaved checkpoint from the cache memory. Once the tasks are executed successfully, the presaved checkpoints are deleted automatically from the cache memory. Thus the algorithm uses in place and in memory recovery of checkpoints that reduces the checkpoint overhead and improves the performance.