Time-dependent Pricing and Scheduling for Cloud Object Storage Service Providers

초록

There is increasing competition among cloud object storage service (COSS) providers as demand for COSSs grows. While various pricing models exist for commercial COSS providers, they do not effectively adapt to changing client demand and resource supply. As a result, many COSS providers are still facing fundamental problems in operational strategy to maximize their profit. In this paper, we propose TD-PnS based on the Lyapunov-drift-minus-profit technique which jointly and dynamically makes decisions on (i) service pricing, (ii) CPU clock scaling and encoding scheduling, (iii) network scheduling, and (iv) energy storage management to maximize COSS provider's profits. We also propose an additional version of TD-PnS, namely TD-PnS-Adv, that adds realistic aspects such as system stabilization. Finally, through trace-driven simulation using real dataset, we demonstrate that the proposed algorithms outperform existing algorithms and pricing models in terms of profit.

키워드