Applying theory of constraints-based approach to solve memory allocation of cloud storage

Cloud computing business models include three service models: software as a service, platform as a service, and infrastructure as a service (IaaS). Cloud storage is a new storage model in which data are stored on host companies and people store data into it through the Internet, a derivative of IaaS. In the cloud storage, the maintenance and purchasing cost of memory space need to be considered to avoid the possibility of running out. There is a conflict between hosts and clients: clients expected to increase the amount of backup memory space to avoid shortage. On the other hand, hosts expected to decrease the cost of backup memory space to maximise profit. In order to resolve this issue, this paper proposed a theory of constraints (TOC)-based approach to solve the memory allocation of cloud storage, which uses market information to build a rolling forecast. Moreover, this paper uses practical and simulative data of cloud storage, provided by a cloud storage company, to verify the effectiveness and feasibility of the proposed approach. After comparing the results that were obtained from the proposed TOC-based approach with the weighted moving average combined with demand-pull and buffer management (DPBM), exponential smoothing combined with DPBM, and traditional TOC methods, it was found that the proposed approach performs better than the listing approaches. Moreover, the proposed TOC-based memory replenishment strategy can decrease a large amount of on-hand memory without impacting service level in the environment of cloud storage.