A Multi-tiered Storage Structure for Cloud Computing

During recent decades, cloud-based shared storage has become a popular storage solution. Nevertheless, current cloud-based I/O subsystem assumes that the storage devices are homogeneous. Recently, the multi-tiered storage system extends the storage hierarchy by using SSDs to cache data from the hard disks has received significant attention. Thus, in this paper, we design an I/O framework for multi-tiered storages in multitenant cloud-based shared storage environments. Specifically, we propose a two-tier per-user queuing structure to solve the user interference and device interference problem. Furthermore, on the basis of the credit-based request scheduling algorithm, we introduce the concept of device credits, instead user credits in previous schemes, to proper charge users in different types of storage devices. The experimental results show our proposed framework is effective in realizing the multi-tiered-based shared storage systems in cloud computing environments.