Towards a zero-knowledge model for disk drives

In this paper, we present a model for disk drives with zero knowledge about the modeled drive. This model is part of our proposal to design a storage system capable of extracting all potential performance and capacity available in a heterogeneous environment with as little human interaction as possible. To make the model, our system automatically learns the behavior of the drive without expecting any prior knowledge about it from the user. In order to achieve this zero-knowledge model, we have studied three approaches: linear approximation, quadratic approximation and neural networks. We have implemented and evaluated these three approaches and found that neural networks are a great mechanism to model drive behavior. This approach has errors below 10% in read operations.