Adaptive Recovery with Reinforcement Learning in Cloud-of-Clouds Storage Systems

Cloud-of-clouds storage systems are widely used in online applications, where user data are encrypted, encoded, and stored in multiple clouds. When some cloud nodes fail, the storage systems need to reconstruct the lost data and store it in the substitute nodes. It is a challenge to reduce the time of data recovery process to ensure the data reliability. In this paper, we adopt a reinforcement learning-based data recovery (RLDR) approach to reduce the regeneration time. By employing Mento-Carlo method, our approach can construct the tree-topology based regeneration process, (a.k.a. regeneration tree), to effectively reduce the regeneration time. Through theoretical analysis, we apply the information flow graph to optimize the network traffic between clouds for given regeneration tree. We conduct extensive experiments with real-world traces to verify the merit of RLDR. The experimental results demonstrate that our scheme can significantly accelerate regeneration process. Specifically, RLDR can reduce the regeneration time by up to 92% and improve the throughput by up to 1210%, compared with the state-of-the-art alternatives. To the best of our knowledge, this is the first work which adopts the reinforcement learning paradigm to reduce the regeneration time in cloud-of-clouds storage systems.