Enhancing Cloud Availability Through Failure-Oblivious Computing

In recent years, cloud computing has drawn much attention from both industry and academia. A cloud is typically hosted in a large-scale data center that must be well operated despite various failures that may cause severe damage. In this paper, we present a technique to increase availability by introducing a failure-oblivious concept into a software cloud platform. As in the original concept, we made the cloud plat- form continue its operation even when the presence of failures. To confirm the effectiveness of this approach, we have implemented that concept into our cloud platform, Kumoi, and evaluated it in terms of continued execution and influence from side-effects.