A Novel Automated Tiered Storage Architecture for Achieving Both Cost Saving and QoE

Abstract

With the exponential growth of data from ICT equipment and the continued development of low-cost storage technology, the scale and amount of data are continually increasing in many areas and moving throughout the cloud. However, most of them are infrequently accessed. Data temperature describes the frequency of data access: hot storage is dedicated to storing frequently accessed data, while cold storage is designed for infrequently accessed data. In this paper, we propose and implement an architecture of an automated tiered storage system that optimizes data allocation in data centers. Our proposed approach brings mutual benefits to both service providers and end users. Users do not need to consider which storage media they want to save, and access and service providers do not need to analyze data access or manually classify data. By successfully predicting infrequently accessed data and moving them to the cold storage, we obtain significant cost saving. While having the benefit of storage cost savings, we also ensure a quality of experience through the correctness of the predicted hot data. The operational strategy varies among cloud storage service providers, and as a result, we characterize different scenarios and provide customized optimal solutions.